Italy

COMPILATION OF ITALIAN HICP BY DIFFERENT GROUPS OF HOUSEHOLDS

Geneva, 2023, June 7-9

UNECE CPI Expert Group meeting

Ilaria Arigoni, Istat (Italy) ([email protected])

Alessandro Brunetti, Istat (Italy) ([email protected])

Valeria de Martino, Istat (Italy)([email protected])

Federico Polidoro, Istat (Italy) ([email protected])

Outline

2

• Inflation in Italy and in the Euro area in 2022-2023: an overview

• Current Istat methodology to compile HICP by five groups of households

• Changing from expenditure to income the variable to identify the groups of

households: main outcomes

• The impact of inflation on the income-based groups of households

• Characteristics of the households in the extreme groups and comparison

between their distributions in the five groups by expenditure and by income

• Is it enough focusing on the weights to measure the actual impact of inflation

on the poorest people?

• Some concluding remarks and perspectives

UNECE CPI Expert Group meeting 2023, June 7-9

Inflation in Italy and in the Euro area in 2022-23: an overview

3

• 2022, as well the final part of 2021, have been characterized, in Italy, in the

European Union (EU) and in the world, by a sharp increase of the rates of change

of consumer price indices that are slowly decreasing in the first part of 2023

• Italian inflation measured by HICP has raised from +1.0% in July 2021 (+2.2% in

the Euro Area) to +12.5% in November 2022 (+10.1% in the Euro area), slowing

down respectively to +7.0% and +8.7% in April 2023.

• Given the impact of energy prices on the sharp raise and on the recent slowdown

of inflation, the overall HICP excluding energy has gone on speeding up (arriving in

March 2023 at +7.9% in the EA and at +6.9% in Italy) and starting declining only in

April 2023 (+7.4% the EA; +6.7% in Italy)

• Yearly rates of change of food prices are still very high (in April 2023 +13.5% in the

EA, +11.0% in Italy)

UNECE CPI Expert Group meeting 2023, June 7-9

Inflation in Italy and in the Euro area in 2022-23: an overview

4

Figure 1. HICP Indices and annual rates of change. Italy and Euro area. 2016 – 2023. Percentage values

7.0

8.7

123.13

121.4

80.0

85.0

90.0

95.0

100.0

105.0

110.0

115.0

120.0

125.0

130.0

-2.0

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

Ja n-1

6

Ap r-1

6

Ju l-1

6

Oc t-1

6

Ja n-1

7

Ap r-1

7

Ju l-1

7

Oc t-1

7

Ja n-1

8

Ap r-1

8

Ju l-1

8

Oc t-1

8

Ja n-1

9

Ap r-1

9

Ju l-1

9

Oc t-1

9

Ja n-2

0

Ap r-2

0

Ju l-2

0

Oc t-2

0

Ja n-2

1

Ap r-2

1

Ju l-2

1

Oc t-2

1

Ja n-2

2

Ap r-2

2

Ju l-2

2

Oc t-2

2

Ja n-2

3

Ap r-2

3

Euro area m/m-12 (left axis) Italy m/m-12 (left axis) Euro area index Italy Index

UNECE CPI Expert Group meeting 2023, June 7-9

Current Istat methodology to compile HICP by five groups of households

5

• Since 2005, Istat has been compiling and disseminating a measure of the impact of

the inflation on five different groups of households of equal dimension ordered by

their spending power (from the lowest of the first group to the highest of the fifth) used

as a proxy of their income conditions.

• Indices of consumer prices are compiled considering the different structure of

consumption expenditure of each group of households (summarized in the system of

weights).

• HICPs by population subgroups are “satellite” indices of HICP: they share the set of

basic information (basket of products and price elementary data) and the

methodology of Italian HICP, but they are different each other for the system of

weights used for their calculation.

UNECE CPI Expert Group meeting 2023, June 7-9

6

• Weights for the five subgroups of households based on HBS data

• To estimate the weights, consumption expenditures are equivalized by using an

appropriate equivalence scale (Carbonaro scale), that considers the effects of

economies of scale and makes them comparable to that of a two-member household,

and, as such, among different-size households

• Households ordered by equivalent consumption expenditure, are organized by specific

cut-point values and divided into five groups of equal size (equivalent-expenditure

fifths)

• In a situation of perfect equality, a share of 20% of the total expenditure sustained by

all the households would be placed in each fifth: actually, in 2021, in terms of

equivalent expenditure, that of the last fifth was about 5 times that of the first fifth

(inequality measure on expenditure side).

Current Istat methodology to compile HICP by five groups of households

UNECE CPI Expert Group meeting 2023, June 7-9

7

Figure 2. Expenditure weights by 5 households groups and main special aggregates in 2023 (HBS year 2022)

UNECE CPI Expert Group meeting 2023, June 7-9

Current Istat methodology to compile HICP by five groups of households

8

• The methodology adopted using households’ expenditure data has been transferred to

households’ income data derived from HBS

• In this case to detect the households’ groups, households’ income data are equivalized

by using OECD-modified equivalence scale

• Households ordered by equivalent income, are organized by specific cut-point values

and divided into five groups of equal size (equivalent-income fifths) from the poorest

one (the first) to the wealthiest one (the fifth)

• Thus, the weights are estimated using the expenditure data of each income group of

households

• The inter quintile ratio between the fifth and the first income group is equal to 3.9,

whereas the fifth group spends 2.07 times what the first group spends (in 2021)

From expenditure to income the variable to identify the groups of households: main outcomes

UNECE CPI Expert Group meeting 2023, June 7-9

9

Figure 3. Expenditure and income weights. First group of households. Main special aggregates (2022, HBS 2021)

UNECE CPI Expert Group meeting 2023, June 7-9

From expenditure to income the variable to identify the groups of households: main outcomes

219,419

112,662

145,527

207,524

100,205

115,613

0

50,000

100,000

150,000

200,000

250,000 P

ro c e ss

e d

fo o

d (

in c l.

a lc

o h

o l a n

d

to b

a c c o

)

U n

p ro

c e ss

e d

fo o

d

E n

e rg

y

N o

n -e

n e rg

y

in d

u st

ri a l

g o

o d

s

S e rv

ic e s

re la

te d

t o

h o

u si

n g

S e rv

ic e s

re la

te d

t o

c o

m m

u n

ic a ti

o

n

S e rv

ic e s

re la

te d

t o

re c re

a ti

o n

S e rv

ic e s

re la

te d

t o

tr a n

sp o

rt s

S e rv

ic e s

m is

c e ll a n

e o

u s

Exp weights first group

Income weights first group

10 UNECE CPI Expert Group meeting 2023, June 7-9

From expenditure to income the variable to identify the groups of households: main outcomes

Figure 4. Expenditure and income weights. Fifth group of households. Main special aggregates (2022, HBS 2021)

115,474

49,340 67,376

124,251

53,387 79,555

0

50,000

100,000

150,000

200,000

250,000

300,000

350,000

400,000

P ro

c e ss

e d

f o

o d

( in

c l.

a lc

o h

o l a n

d t

o b

a c c o

)

U n

p ro

c e ss

e d

f o

o d

E n

e rg

y

N o

n -e

n e rg

y i n

d u

st ri

a l

g o

o d

s

S e rv

ic e s

re la

te d

t o

h o

u si

n g

S e rv

ic e s

re la

te d

t o

c o

m m

u n

ic a ti

o n

S e rv

ic e s

re la

te d

t o

re c re

a ti

o n

S e rv

ic e s

re la

te d

t o

tr a n

sp o

rt s

S e rv

ic e s

m is

c e ll a n

e o

u s

Exp weights fifth group

Income weights fifth group

11

• The structure of weights is similar but not the same between those referred to the

groups detected by expenditure data and those referred to the groups detected by

income data

• Specifically:

✓ For the first group (low income/low expenditure) the weights of the aggregates affected by

higher increase of consumer prices in 2022, decrease in relative terms (by 1.19

percentage points for unprocessed food, by almost 3 p.p. for Energy)

✓ For the fifth group (high income/high expenditure), vice versa the weights of the aggregates

affected by higher increase of consumer prices in 2022, increase in relative terms (by

0.88 percentage points for unprocessed food, by almost 1.22 p.p. for Energy)

From expenditure to income the variable to identify the groups of households: main outcomes

UNECE CPI Expert Group meeting 2023, June 7-9

The impact of inflation on the income-based groups of households

12 UNECE CPI Expert Group meeting 2023, June 7-9

14.5

10.7

17.9

9.9

-5.0

0.0

5.0

10.0

15.0

20.0

F eb

-1 8

A p

r- 18

Ju n

-1 8

A u

g -1

8

O ct

-1 8

D ec

-1 8

F eb

-1 9

A p

r- 19

Ju n

-1 9

A u

g -1

9

O ct

-1 9

D ec

-1 9

F eb

-2 0

A p

r- 20

Ju n

-2 0

A u

g -2

0

O ct

-2 0

D ec

-2 0

F eb

-2 1

A p

r- 21

Ju n

-2 1

A u

g -2

1

O ct

-2 1

D ec

-2 1

F eb

-2 2

A p

r- 22

Ju n

-2 2

A u

g -2

2

O ct

-2 2

D ec

-2 2

income weights first group Income weights fifth group Exp weights first group Exp weights fifht group

Figure 5. Inflation impact on the first and the fifth (by expenditure and income) groups of households. All-item

index. M/M-12 rate of change, January 2018 – December 2022

The impact of inflation on the poorest group of households in Italy

13 UNECE CPI Expert Group meeting 2023, June 7-9

• The differences highlighted in the weights between the extreme groups considered

either in terms of equivalent expenditure or in terms of equivalent income, do not

produce gap in terms on impact of inflation between the first and the fifth group in the

years when inflation is relatively low (2018 – 2020)

• As soon as the price increase becomes heterogeneous across the different product

aggregates, with energy prices sharply growing at rate strongly higher than that of

other aggregates, followed by food products, the inflation gap between the first and the

fifth group (considered both in terms of expenditure and income) starts enlarging (end

2020)

• Given the differences in the structure of weights of the 2 extreme groups, the gap

between the two groups detected by expenditure data becomes gradually wider than

that between the two groups detected by income data (in December 2022 it is equal to

8 p.p. in the first case and to 3.8 p.p. in the second case)

Characteristics of the households in the extreme groups and comparison between their

distributions in the five groups by expenditure and by income

14 UNECE CPI Expert Group meeting 2023, June 7-9

• In 2021 in the first fifth of households’ expenditure group (but not of income):

✓ about 72% of the households range from 2 to 4 members

✓ in 28.8% the breadth of households is equal to 2, are mainly couples without children in

which the reference person is elderly (14.6%) and retired from work (95.0%)

✓ Households of 3 or 4 members are mainly couples with two children

• In 2021 in the first fifth of households’ income group (but not of expenditure):

✓ there are mainly one-component households (37.1%), followed by those households of a

size equal to 2 (22.9%)

✓ A fifth of the households (54.7% of all single-component) are elderly alone, retired from

work (49.8%) or inactive but in other condition (different from retired) (46.3%)

✓ With respect to what is observed in households belonging to the first fifth of expenditure but

not income, in the first fifth of households’ income group, the over-represented are mainly

single people (1.8 times) and single-parent families (1.3 times)

Characteristics of the households in the extreme groups and comparison between their

distributions in the five groups by expenditure and by income

15 UNECE CPI Expert Group meeting 2023, June 7-9

Table 1. Cross distribution of households by expenditure/income fifths. Absolute value and percentage points. 2021 Income fifths

Exp fifths 1 2 3 4 5 Total

1

2,692,049 1,375,267 699,709 297,137 137,507 5,201,670

10.35 5.29 2.69 1.14 0.53 20

51.75 26.44 13.45 5.71 2.64

51.75 26.44 13.45 5.71 2.64

2

1,219,493 1,518,011 1,261,158 785,691 415,794 5,200,147

4.69 5.84 4.85 3.02 1.6 19.99

23.45 29.19 24.25 15.11 8

23.44 29.19 24.24 15.1 7.99

3

710,222 1,149,206 1,363,005 1,182,498 798,416 5,203,348

2.73 4.42 5.24 4.55 3.07 20.01

13.65 22.09 26.19 22.73 15.34

13.65 22.1 26.2 22.73 15.35

4

417,855 796,851 1,127,863 1,523,926 1,334,976 5,201,472

1.61 3.06 4.34 5.86 5.13 20

8.03 15.32 21.68 29.3 25.67

8.03 15.32 21.68 29.3 25.67

5

162,527 361,463 750,225 1,412,514 2,514,584 5,201,313

0.62 1.39 2.88 5.43 9.67 20

3.12 6.95 14.42 27.16 48.35

3.12 6.95 14.42 27.15 48.35

Total 5,202,147 5,200,798 5,201,960 5,201,766 5,201,277 26,010,000

20 20 20 20 20 100

F ifth

s o f in

co m

e b y fifth

s o f exp

en d

itu re

Fifths of expenditure by fifths of income

Characteristics of the households in the extreme groups and comparison between their

distributions in the five groups by expenditure and by income

16 UNECE CPI Expert Group meeting 2023, June 7-9

• In 2021:

✓ 21.8% of the first fifth of households by expenditure is allocated in the last three fifths of

households by income

✓ 24.8% of the first fifth of households by income is allocated in the last three fifths of

households by expenditure

✓ 24.5% of the fifth fifth of households by expenditure is allocated in the first three fifths of

households by income

✓ 26.0% of the fifth fifth of households by income is allocated in the first three fifths of

households by expenditure

Characteristics of the households in the extreme groups and comparison between their

distributions in the five groups by expenditure and by income

17 UNECE CPI Expert Group meeting 2023, June 7-9

• The heterogeneity of the allocation between the two groups of households at the basis of the

differences in the structure of weights

• In the first fifth of households by income there are households belonging to groups of

households by expenditure from the second (23.44%) to the fifth (3.12%)

• In the fifth group of households by income there are households belonging to groups of

households by expenditure from the second (26.44%) to the fifth (2.64%) group

• It means that the breakdown of expenditures in the first group of households by income is

different from that of the first group of households by expenditure, bringing behavior of

consumption typical of households that spend wider amount and reducing the relative weight of

food and energy products and vice versa for the fifth group of households by income

• This brings closer in 2022 the lines of inflation that affect the two extreme groups by income

(specifically lowering that related to the poorest) with respect those that affect the two extreme

groups by expenditure

Focusing only on the weights to measure the actual impact of inflation on the poorest?

18 UNECE CPI Expert Group meeting 2023, June 7-9

• Till now, in Italy, the analysis to estimate the differentiated impacts of inflation on groups of

households broken down by their economic condition has focused on the structure of weights

• In 2022 the government supports to poor households (detected in the basis of their income or

other indicators of their economic conditions) related to energy products (in particular, electricity

and gas) have been wide and in the form of reduction of prices

• This was considered in the compilation of the energy product consumer price indices that are

the results of weighted mean (with weights given by the number of households that have

benefited of the government support) of different inflation profile

• The aggregate consumer price indices of energy products are considered as such to estimate

the impact of overall inflation on the poorest and on the richest groups of households

• Should we start considering different profiles of inflation in addition to different structure of

weights?

• Moreover, how should we consider the impact of government support to households on the

weights of different groups, given the traditional temporal lag in the weights’ estimation?

Some concluding remarks and perspectives

19 UNECE CPI Expert Group meeting 2023, June 7-9

• The capacity of the extreme groups of households by expenditure to be a proxy of the poorest

and of the wealthiest households is mitigated by these results

• The outcomes of the estimation of the impact of inflation on different groups of households are

interesting and encourage further analysis. Specifically:

✓ Of the socio-demographic and socio-economic characteristics of the households’ groups by income

to be further analyzed

✓ Of the relationship between income and expenditure in the different groups

✓ Considering different profile of inflation for the 2 extreme groups to complement the approach based

exclusively on the weights

✓ Refining further the work on the weights given the effects on the structure of expenditure of the

government support to poorest households on energy products

• Starting the dissemination of an experimental statistics to open the debate (in 2024?)

• The new frame regulation on the social statistics that will harmonize HBS in the EU under a

common legal umbrella since 2026 will enhance the possibility to use HBS data to compare

across the European countries the impact of inflation on the different groups of households

Thank you

Ilaria Arigoni, Istat (Italy) ([email protected])

Alessandro Brunetti, Istat (Italy) ([email protected])

Valeria de Martino, Istat (Italy)([email protected])

Federico Polidoro, Istat (Italy) ([email protected])

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

1

Penalties management strategy and customised return of statistical information

to enterprises involved in official economic surveys1 Authors: S. Binci, P. Bosso, S. Curatolo, F.Monetti, P.Papa

ISTAT - Italian National Statistical Institute

Directorate for Data Collection

Abstract

The increasing statistical burden that official statistical surveys impose on the Italian business system determines a decreasing motivation to actively collaborate in direct surveys. The lack of motivation often translates into decreasing participation rates and growing impatience. This situation leads the National statistical institutes to identify alternative strategies oriented to identify new statistical sources and to automate some phases of the data collection process. In the above mentioned framework Istat started to investigate, from a statistical point of view, the role of penalties in ensuring adequate participation of companies, notably the most influential ones, in official statistical surveys. The attention was also focused on the possible “side-effects” on other dimensions of the Total Survey Error and on the possible alternative tools (new sources, new technologies) and solutions (organizational, communication) to be adopted during data collection in order to ensure awareness among companies and to make the provision of data by users more “sustainable”. The analysis mainly concerned the economic surveys carried out during the years 2021 and 2022, distinctly in the contexts of the structural and short-term economic surveys, which provided for two substantially different systems of application. The indicators used for the analysis are the response rates of the various surveys and the penalty- rates applied. The context is that of the centralized data collection model, adopted by Istat. A complementary solution, already partially implemented by Istat for some years in the context of the Centralized Data Collection model and the Business Statistical Portal of companies, consists in returning to companies a set of useful information to understand the specific trend of their sector of belonging and their positioning in the markets, motivating them to participate in surveys. In this context, the objective of the paper is to define an optimal structural framework, based on the information available, of the sector of economic activity to which each company belongs, as well as information on the reference markets and on its competitive positioning, providing useful guidelines for planning of such systems. The framework will be defined on the basis of the experience already acquired in Istat and on similar experiences carried out in other NIS and may constitute a basis of reference and comparison with other institutions that intend to design and implement a return system.

1.1 Penalties management strategy and quality of economic official surveys

Since the year 2016 Istat introduced generalized and strict criteria for penalties management as a consequence of introduction of a centralized data collection model [7], [8], [9]. In the past, while respecting the regulatory requirements, penalties were managed by a “local approach” that adopted specific criteria for each direct survey. According to this approach each survey adopted a specific and autonomous DC solution, involving no or partial integration among processes. 1 Contributors: S. Binci paragraph 1.3; S. Curatolo paragraphs 1.4.1 and 1.4.2 F. Monetti paragraphs 1.2.1 and 1.2.2 ; P.

Papa paragraphs 1.1 and 1.5, P. Bosso paragraphs 2.1, 2.2, 2.3, 2.4.

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

2

The generalization involved some specific basic penalties management criteria that were applied to all the surveys: 1) Defined and strict deadlines for the transmission of data; 2) Predefined inclusion/exclusion criteria in the lists of units subject to penalties; 3) Definition of a penalties provision procedure shared among all involved stakeholders (DC, thematic, legal, technical experts). Within the scope of these general criteria, the application of penalties provides for two different procedures for the two types of structural and short-term surveys. In Italy the legal framework for penalties (articles 7 and 11 of the legislative decree n. 322/1989) is updated annually by means of a specific decree that identifies the surveys subject to penalties and the related penalty thresholds determined in terms of the number of employees or turnover. According to this legislation obligation and penalties do not coincide, as most of the official surveys have an obligation but do not provide for the application of penalties. Operational application criteria are set out in specific methodological notes published on the Istat website and in the information letters sent in the start-up phase of each survey. It is important to stress that, according to Italian legislation, the economic amount of the penalty is not commensurate with the "statistical damage" caused by the defaulting unit but it is fixed and equal for all defaults (amount about 1032 Euros). For instance, the same amount is applied to a company that has omitted the delivery of data for one month and one that has omitted 12 months. This documents describes the third step of the work carried out by ISTAT on the effectiveness of penalties application on business surveys. The first one [10] presented the features of the new penalty procedure designed for STS surveys and the preliminary effects on response rates. The second [11] included the effects of the new procedure, after the first year of real application, on response rates of the STS surveys, pointing out possible negative «side effects» in terms of quality of the information produced. The third step extends the analysis to first months of the year 2022, to structural business surveys and to the types of non-compliance underlying the penalties application.

1.2.1. Short-term economic surveys. New penalty management procedure

In 2021 Istat, for the first time, fully implemented the strict criteria provided by the new penalty procedure introduced for short-term surveys in 2018. Istat delayed the implementation of the new procedure firstly to give companies time to familiarize with the new criteria. Subsequently due to the regulatory measures issued by the government to lighten the burden of the Covid19 health emergency on businesses. The new criteria concerned the time articulation of penalties on an annual basis and the provision of the administrative penalties in the following cases [10]:

- Unit is non-responding for one or more periods (default A) - Unit provides the data beyond the days of tolerance with respect to each monthly or

quarterly deadline, varying from survey to survey ( default B)

- Unit provides the data beyond the annual cumulative tolerance (less than the sum of the tolerance of the single periods).

1.2.2. Short-term economic surveys. main results

Defaulting companies for the 2021 survey year were 3.428 out of 16.299 potentially subject to non- compliance assessment, specifically 2.280 for the monthly surveys and 1.148 for the quarterly surveys. No company was defaulting for providing data beyond the annual cumulative delay. With reference to monthly surveys, about 50% of companies were subject to penalties for having

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

3

provided the data lately (default B), 25% of these provided data after the set deadlines for a single period. Non-responding companies are about 27% (default A), 11% of these responding for 11 periods (Graph 1). Table 1 shows that Industrial production survey (IPI) and Industry turnover and orders survey (FATT) are the surveys with the highest percentage of enterprises defaulting for providing data beyond the monthly deadline, respectively about 64% and 62%. On the other hand, surveys on Industrial producer prices are those that recorded the largest number of non-responding companies, about 56% both for non-domestic market (PPID) and domestic market (PPID), compared to about 16% of companies that provided data after the deadlines. This trend is partly justified by the closure, for these surveys, of the acquisition systems a few days after the expiry of the tolerance period.

Figure 1 - Monthly short-term business surveys: companies subject to administrative penalties by type of violation (%), years 2021.

Table 1 - Companies subject to administrative penalties by monthly short-term business surveys and type of violation, years 2021

Survey

Defaulting

units

(number)

Type of violation (%)

Data delivered

beyond useful

period deadline (A)

Non-

responding

(B)

A + B

Retail trade (DETT) 213 48.8 19.7 31.5

Industry turnover and orders (FATT) 497 62.8 17.9 19.3

Industrial production (IPI) 643 64.7 18.8 16.5

Employment in large enterprises (OCC) 438 50.7 22.6 26.7

Industrial producer prices, domestic market (PPID) 187 16.6 56.1 27.3

Industrial producer prices, non domestic market

(PPIND)

143 16.8 55.9 27.3

Industrial import prices (PREIMP) 159 24.5 52.2 23.3

Referring to quarterly surveys, the largest percentage of dafaulting companies is concentrated in the type of default B (51.7%); the remaining companies subject to assessment fall into types A and A + B, respectively for 36.7 and 11.7%.

36,7

51,7

11,6

27,9

14,9

5,5

0

10

20

30

40

50

60

Data delivered beyond useful period deadline (defaultA)

Non-responding (Default B) A + B

At least one default (%)

Single default (%)

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

4

This trend is partly due to the closure of the acquisition system of Services producer prices survey (PPS) at the end of each quarterly tolerance period: the closure of the data acquisition system implies that all the non-compliant companies fall within the type of default B. Also the other two quarterly surveys recorded the highest number of defaulting enterprises in the type of default B: respectively 47.8% for Service Turnover survey (FAS) and 50.9% for Job Vacancies survey (VELA). Figure 2 - Quarterly short-term business surveys, companies subject to administrative penalties by type of violation (%), years 2021

Table 2 - Companies subject to administrative penalties by quarterly short-term business surveys and type of violation, years 2021.

1.3. Main results: impact on response rates

As shown in the following tables, the new system has undoubtedly led to significant increases in

response rates (rr) for the main short-term surveys, ensuring participation of companies on final

results, notably the most influential ones. The comparison was carried out at the end of the useful

periods, among year 2016 (last survey edition run before CDC introduction), year 2017 (first survey

edition run after CDC introduction), year 2018 (the first survey edition run after the introduction of

the new penalty organization) year 2021 (when the new management is now fully operational) and

years 2022 (to verify the effects of the new fully operational penalties system).

Considering only the enterprises virtually subject to penalty (Table 4), the response rate (rr)

increased of about 18 percentage points (pp) in 2018 starting from 72 percent in 2016. As the data

collection process in 2017 and 2018 is characterized by the same new tools and methodologies

50,4

27,1

22,5 25,0

10,8

3,9

0

10

20

30

40

50

60

Data delivered beyond useful period deadline (defaultA)

Non-responding (Default B) A + B

At least one default (%) Single default (%)

Survey

N. Units

penaltied

Type of violation (%)

Data provides beyond useful

period deadline (A)

Non-responding

(B)

A + B

Service Turnover (FAS) 716 40,4 47,8 11,9

Services producer prices (PPS) 63 0,0 100,0 0,0

Job Vacancy (VELA) 369 35,8 50,9 13,3

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

5

introduced with the CDC, by comparing 2017 and 2018 results it is possible to focus on the effect

due exclusively to the introduction of the new penalty procedure [10] It shows an average raise of

about 13 pp starting from 77 percent in 2017. The increase is consolidated over the following years,

reaching 90 and 93 percent, respectively in 2021 (in this year the threshold of employees is lowered

for some surveys, causing a slight decrease in their rr compared to 2018) and 2022 (the growth is

due to the effects of the penalties received by enterprises in 2021)

The introduction of the new penalty organization allowed the positive average variation of 18 pp in

rr in 2018 also for the surveys as a whole, including companies that were not virtually subject to

penalty as under the penalty threshold, meaning that the impact of the new management criteria

has positive effects also on smaller units (Table 3). The increase in terms of rr is also confirmed in

2021 and 2022 compared to 2018, respectively of 1 and 4 pp.

Table 3 - Short-term business surveys - Average response rates (%)

Survey Year 2016 Year 2017 Year 2018 Year 2021 Year 2022

Employment in large enterprices OCC1 68 71 87 84 92

Industrial producer prices PPI 78 83 93 94 92

Retail trade DETT 39 45 70 78 79

Industry turnover and orders FATT 74 77 87 90 90

Industrial production IPI 60 66 84 90 91

Service turnover (Q) FAS 64 71 75 79 94

Services producer prices (Q) PPS 80 79 91 81 90

Table 4 - Short-term business surveys - Average response rates (%) of enterprises virtually subject to penalties

Survey Year 2016 Year 2017 Year 2018 Year 2021 Year

2022*

Employment in large enterprices OCC1 68 71 87 84 92

Industrial producer prices PPI 75 86 96 97 95

Retail trade DETT* 63 68 87 85 90

Industry turnover and orders FATT 86 90 95 96 97

Industrial production IPI 63 72 91 94 95

Service turnover (Q) FAS* 73 77 87 90 96

Services producer prices (Q) PPS* 80 79 91 84 93 *Period Jan-Aug 2022/I-II trim 2022 **In 2021, the threshold for penalties lowers to 100 employees, except for PPS that introduces a threshold for the first time

1.4.1 Structural economic surveys: penalties management procedure The structural economic surveys have an annual or multi-year periodicity and provide for a single supply of the requested data within a defined collection period. Once the deadline for data transmission has expired, the units are considered non-compliant and subject to penalty, if over the threshold values provided for by the legislation in force. Generally, the threshold value is represented by the employees reported in the business registers adopted by the survey owner, whose number can vary from one survey to another. Only in some cases the volume of turnover is considered as a threshold value in addition to the number of employees. 1.4.2 Structural economic surveys: main results

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

6

Table 5 shows the main structural economic bussines surveys, only for the PMI (SBS small and medium companies), GVC (Global Value Chain) and COVID surveys it is not imposed the administrative penalty measure. PMI and COVID are the surveys with the lowest response rate, both are not subject to penalty. Considering only the units shared between PMI, CPUE (Economic Units Permanent Census carried out in 2019), COVID and SCI (SBS large companies), similar by type of eligible units and technique, it emerges that the provision of penalties allows to obtaine higher rr. In particular the rr of the joint units between the PMI and CPUE is respectively of 53.4 and 75.0 percent. Focusing the analysis on the size class more than 50 employees, the threshold value for the applicability of the penality in CPUE (Economic Units Permanent Census), the rr increased to 65.2% per PMI and 86.1% for CPUE, despite the difference between the two rr always remains about 21%. Moreover relating COVID e SCI (edition 2022), the rr of the joint units are higher for the survey subject to administrative penalties that recorded an increase of about 27 pp respect to COVID. Comparing the response rate for the years 2021 and 2022, it emerges that the penalty rate records similar levels in the two years. The use of the administrative penalties met the active collaboration of the involved enterprises, despite the Covid19 health emergency. Focusing on the years 2021/2022 respect the years 2017/2018, the variation over time of the penalty rate mainly depends on changes of the threshold value passed from 500 to 250 employees.

Table 5 - Structural surveys: avarage response and units subject to penalties rates

Survey

Year 2017/2018 Year 2021 Year 2022

Sampled

units

(number)

Average

response

rates

(%)

Defaulting

units (%)

Sampled

units

(number)

Average

response

rates

(%)

Defaulting

units (%)

Sampled

units

(number)

Average

response

rates

(%)

Defaulting

units (%)

CIS

Community

innovation

survey

32,018 68.1 0.7 - - - 39,534 62.3 0.6

SCI – SBS

large

companies

10,558 76.4 1.3 3.811 86.7 8.9 3,997 85.7 9.6

PMI-SBS

small and

medium

companies

74,207 43.5 - 82.022 48.8 - 77,611 43.3 -

RFI - - - 83 84.7 - 87 89.7 5.7

OUTWARD 6,326 69.8 0.2 5.899 68.2 1.1 5,982 67.3 0.8

INWARD 7,791 74.4 0.2 8.937 68.4 1.7 - - -

RCL-LCS - - - - - - 24,528 61.7 1.9

PRODCOM 39,799 56.2 0.1

R&D1 17,977 76.5 0.4 39.115 66.8 0.6 30,826 72.8 1.0

IULGI 10,536 80.4 0.8 8.473 83.0 2.3 8,222 82.2 2.2

ICT 32,255 67.0 0.2 32.929 63.5 0.6 33,992 63.0 0.6

GVC-

Global

Value Chain

- - - 35.969 64.7 - - - -

COVID

Survey - - - 90,470 46.0 - - - -

1.5. Conclusions

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

7

In official business surveys the obligation is a necessary component of the DC strategy as it ensures completeness and timeliness of the information collected. A DC based only on awareness and free collaboration is not enough, as companies increasingly experience statistical obligations as a direct cost and this issue can prevail on the awareness of importance to provide accurate statistical information for official surveys. In fact, completeness as represented by response rate is not the only issue to consider in order to evaluate the quality of survey results as equally important is the nonresponse bias. In other terms, considering the components of TSE (Total Survey Error) [1], [6] the improvements on the “Non- response error dimension” can be partly offset by the increase of the “measurement error” component and nonresponse bias. ISTAT receives more and more frequently communication of complains by companies involved in the surveys and several of them open legal disputes. Some companies declare that they, in order to meet deadlines, provide provisional data that are not validated and deliberately decide to pay the penalties without delivering data. Therefore, effective DC approach requires a balance between obligation and awareness. The only obligation can involve negative side effects that can impact on the quality of the survey results. The penalty procedure currently applied by ISTAT is very effective in the short period as it has increased substantially response rates and timeliness of the direct survey but it presents some sustainability risks in the medium term, as it is perceived as too rigid and oppressive by companies. For STS surveys the analyses carried out on defalting units pointed out a relevant share of penalties applied for defaults concerning just one period, so a possible compromise could concern the introduction of further flexibility in the form of a "bonus" for a single period, even if both the effects on the completeness of the data collected and legal feasibility must be carefully investigated. Building a two-way communication flow between NSI and the enterprises involved in business surveys by designing a return of customised statistical information is a possible balancing strategy.

2.1. Part 2. Design of a new system for statistical information return to companies involved in

economic surveys, as part of the Statistical Business Portal

The Italian companies are involved in many official statistical surveys. The information provided is

necessary so that the central and local administrations can understand in detail the structure of

Italian business market and analyze the short-time economic phenomena. For this reason, the

business units involved in statistical surveys spend a lot of resources in order to to retrieve, collect

and filling out the numerous information required in the questionnaires and thus fulfill the statistical

obligations. Moreover, participating in official surveys is not optional for companies; they are

obliged by law to complete these surveys and in Italy the penalty system for those who do not

comply with statistical obligations is very rigid.

The large number of surveys and the associated obligation are a burden for the business units. This

burden is the reason for the widespread reduction in response rates affecting the business surveys

in the last years. Furthermore, the response burden doesn’t only concern the decrease in response

rates but can also affect the quality of the data provided.

The management of the response burden is therefore an high priority in the production of statistical

information, underlined primarily by the Code of Conduct for European Statistics. It is an essential

condition for the quality of future statistical production, therefore it’s necessary for the Statistical

Institutes to study strategies to reduce the response burden or to compensate, at least partially, the

units involved in the surveys for the commitment that the statistical obligations require.

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

8

2.1 The idea of the project

The basic idea of the project is to return “customised information” to the business units involved in

the economic surveys to compensate the response burden. Customised information means

returning a benchmark, that is a structural framework about the activity sector to which the

business units belong and information about their reference markets. A customised information

requires a profiling preliminary activity of the enterprises. The profiling activity can be based on the

main study domains, as the size of the business units, the geographic localisation, the activity sector.

Then the identified business clusters are put in a reference benchmarking framework.

The aims of an informative service so designed are:

 balance the statistical burden building a two-way information flow, so that the enterprises

provide data required and Istat return personalised information;

 encourage the enterprises to collaborate to Istat surveys and improve the quality of data

provided;

 provide information more adapted to the real needs of the business units and already

structured for their internal use.

Istat mission is the production and free and timely dissemination of statistical data. The information

covers all relevant topics and it is disseminated through different services/products, but it is not

necessarily provided according to a business-oriented approach. For example the thematic

databases are information services which allow to the users to browse and use the data available

with a deep level of detail, useful for specific aims. Using the different information services however

requires time and skills to select and buid a reference information frameworks.

Customised information via Business Portal instead, at full capacity, provide information already

structured and personalised for the business units. The information will be available using the

following services:

 Selection - Companies can select the most relevant data for their business: short-term

indicators (production, turnover, export, sentiment indicators) and structural indicators

(productivity, profitability);

 Customization - The information is already filtered according to the relevant characteristics

of the enterprises (economic sector, size, location);

 Analysis and data visualization - The information is displayed by synthetic indicators and

effective

 Benchmarking - Positioning indicators, based on microdata and in compliance with

confidentiality, provide the competitive positioning of the enterprises in their reference

markets.

2.2 Phases and activities of the project

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

9

The project of a customised information return system in Istat (Italian National Statistical Institute)

took place in two phases. The first phase started in 2015 with the start up of the Business Portal and

it concluded in 2018. The second phase started in 2020, suspended due to Covid-19 emergency and

restarted in the current year 2022.

The first phase of the project (2015-2018)

The first phase involved the identification of the short-term and structural indicators to return to

enterprises and the IT implementation in the Business Portal. The Portal includes the section

Statistical Data dedicated to the return of personalised information to the companies involved in

the surveys. The first phase has been an experimental step with the aim to produce a prototypal

customised information service.

In the following Figures are reported some example of indicators provide in the first prototypal

release, in the section Statistical Data. The Figure 3 shows some examples of short-term indicators,

broken down by activity sector; the Figures 4 and 5 show some examples of information returned

about the foreign trade.

Figure 3 – Short-term indexes by sector of activity

Nace 30.9 -

Manufacture of

transport

equipment n.e.c.

Turnover index

Monthly data

Nace 30.91 -

Manufacture of

motorcycles

Index of

industrial

production

Monthly data

Nace 30.91 -

Manufacture of

motorcycles

Producer price

index

Monthly data

Figure 4 – Markets and reference products for the company export.

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

10

Outlet markets for the

products sold by the

company

Main types of products sold

by the company

Figure 5 – Performance of companies in the foreign markets

The

performance of

the company in

the relevant

foreign market

segments

The second phase of the project (2022)

The second phase involved the redesign of the system according to the new data collection

organization in Istat (2016) and the compliance with the current privacy legislation. In fact, a focus

point of the project concerns confidentiality and data protection assessment.

The redesign implied a new production process which includes four phases, as summarized in the

following figures.

Figure 6 – Phases of redesign

1. Expansion of structural and

short-term indicators

2. Definition of IT specifications

and presentation methods

3.

Implementation and testing

4. Release of a first set of

indicators and monitoring

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

11

The project will be developed in several steps that involve the progressive integration of information

over time. The process is circular. A first release of indicators will be produced and then,

subsequently, the system can be fed with further information that will emerge both from internal

analyses and from the feedback coming from the users. The last phase includes the collect of opinion

from the users about the information system to obtein suggestions for procedure improvement.

Each phase involves a set of activities, as following.

1. Expansion of

structural and short-

term indicators

2. Definition of technical

specifications and

dissemination methods

3. Implementation and

testing

4. Release of a first set of

indicators

 Find skills and resources

necessary for the

development of the

project;

 Recognition of the

internal information

available and usable for

the project;

 Identification of a first

set of coherent indicators

(absolute values, relative

values, indexes)

 Identification of profiling

variables (sector of

activity, geographic

localization, size, …);

 Check of legal aspects

(protection of personal

data and confidentiality

of respondents);

 Share the hypotheses of

indicators with internal

and external experts

(companies, sectorial

associations).

 Definition of technical

specifications and

(automatic) methods of

supplying sources;

 Definition of the methods

of dissemination of

information and navigation

in the system;

 Share of dissemination

hypotheses with internal

and external experts

(companies, sectoral

associations).

 System development and

implementation;

 Test the functionalities and

usability of the system;

 Design of a system for

monitoring the user access

to the indicators

implemented (paradata);

 Find access procedures for

all companies and

implementation.

 Activation of the first

release of new

indicators, in the

Statistics Portal, section

statistical data

 Collection the opinion of

business units

concerning the new

customized information

system;

 Analyze feed-back from

the users and planning

improvement actions.

The redesign process implied the solution of several problems, in particular:

 address the technical problem by designing and building a solution (technical specification

document outlines, IT implementation and testing);

 define a coherent set of indicators for the enterprises assessment; that implies different

skills and braimstorming activity with internal expert and stakeholders;

 guarantee the compliance with the confidentiality and privacy issues;

 define the strategy for effective information dissemination.

As far as the last point, the main issues to resolve are:

 Compliance with confidentiality requirements. The information will always be returned in

aggregate form in compliance with current legislation on the protection of the confidentiality

of respondents and any industrial secret.

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

12

 Compliance with dissemination standard procedures. The information will be produced and

returned with standard procedures, in line with the dissemination methods adopted by Istat,

ensuring the same level of detail and the same access possibilities for all Italian companies.

 Equal access to information for all users. To ensure equal treatment of users, all companies

must be able to access the "Statistical data" section and obtain the personalized information

available.

 Timeliness and Punctuality. To satisfied the needs of enterprises the information will

provided in compliance with the timeliness and punctuality dimensions. That will be possible

using automatic solutions for the information system update.

2.3 Conclusions

The building of a customised information system to the business units could be an opportunity to

reduce and balance the statistical burden and the rigidity of the Italian penalty system, but there

are still some problem to resolve and information to acquire. There are currently two areas in which

we are working towards the goal:

1. Find a shared solution with the legal office of Istat to ensure equal treatment of Italian

companies. The solution could be allow all companies access to the Business Portal. Even

companies currently not authorized to access, as they are not involved in Istat surveys, will

be able to request access credentials within the web Portal or directly from the institutional

website.

2. Studying a first set of benchmark indicators able to be update automatically. This issue is

necessary to ensure timeliness and punctuality in providing information to the users.

References

[1] Groves R.M. and Lyberg L. (2010), Total Survey Error: Past, Present, and Future Public Opinion Quarterly, Volume 74, Issue 5, 2010, Pages 849–879, https://doi.org/10.1093/poq/nfq065.

[2] Snijkers, G., Haraldsen, G., Jones, J., & Willimack, D. (2013). Designing and conducting business surveys. John Wiley & Sons.

[3] Istat (2016), Istat’s modernization programme https://www.istat.it/it/files//2011/04/IstatsModernistionProgramme_EN.pdf

[4] Bavdaž, M., Snijkers, G., Sakshaug, J. W., Brand, T., Haraldsen, G., Kurban, B., ... & Willimack, D. K. (2020). Business data collection methodology: Current state and future outlook. Statistical Journal of the IAOS, 36(3), 741-756.

[5] Consolini M., Eiffe F.F.(2022) Experimental design of the European Company Survey online-follow up 2020: Customised reports as incentive to participateBDCM 2022 – Sixth International Workshop on Business Data Collection Methodology13-15 June 2022, Oslo, Norway) Giugno 2022

[6] Biemer P. Total survey error design, implementation, and evaluation Public Opinion Quarterly, Volume 74, Issue 5, 2010, Pages 849–879. [7] L. Rivais, M St-Denis, S. Lensen (2013), Centralising data collection at Statistics Canada. Seminar on Statistical data collection. Unece - Conference of european statisticians.

UNECE Expert Meeting on Statistical Data Collection 2022 26-28 October 2022, Istat, Rome

13

[8] P. Saraiva dos Santos, A.Moreira (2013), Creating a data collection department: statistics portugal's experience. Seminar on Statistical data collection, Unece - Conference of European statisticians. [9] Istat (2016), Istat’s modernisation programme,https://www.istat.it/it/files//2011/04/IstatsModernistionProgramme_EN.pdf. [10] Binci S., Monetti F., Papa P., (2019) Centralised data collection: effects of a new administrative penalties provision procedure in business short-term surveys EESW19, 6th European Establishment Statistics Workshop, 24-27 Settembre 2019, Bilbao. [11] Binci S., Monetti F., Papa P., (2022) Data collection strategies in business short-term official surveys: a balance between legal obligation and awareness, Workshop BDCM 2022 – Sixth International Workshop on Business Data Collection Methodology13-15 June 2022, Oslo, Norway, Giugno 2022.

Consumer price indices for motor vehicle insurances: the new source and

methodology in the Italian experience

Current survey on motor vehicles’ insurances pricesCurrent survey on motor vehicles’ insurances prices

Methodological approach: Model prices of representative consumers’ profiles

Current approach is based on model prices. It defines different consumers’ profiles, which are expected to be representative of all

consumers. These profiles are designed on the main features which are likely to influence insurance prices, among which the most

relevant are: driver’s age, driving experience, city of residence, kind of motor vehicle, number of car accident throughout last years.

Survey’s technique: Price collection on the field, carried on by municipal data collector

In terms of data collection, Italian consumers prices survey consists in two main collection channels: one is called “centralised” and is

directly managed by Istat; the other is called “territorial” and is managed by Istat through statistical offices of different municipalities. In

the territorial data collection prices are observed directly on the field by municipal data collectors and this is the case also for motor

vehicles’ insurances. In concrete terms, municipal data collectors visit selected insurances’ broker agencies and ask the insurances’ prices

with corresponded to the designed consumers’ profiles.

CURRENT SURVEY

In methodological terms

A model price method, based on a limited and predefined profiles,

increasingly appears inadequate to properly represents a market

characterised by a broad supply of products.

In feasibility terms

Many times, municipal data collectors cannot collect prices, since often

insurances’ brokers are not able in establishing a certain prices without

having a concrete licence plate number.

DRAWBACKS of current survey Current approach has begun to show severe drawbacks during last years

A private-public partnership for innovation of motor vehicles’ insurances prices surveyA private-public partnership for innovation of motor vehicles’ insurances prices survey

In 2020 a partnership has been established between Istat and public and private stakeholders of insurance market. The purpose was to innovate the survey on prices of motor vehicles’ insurances,

in order to provide to citizens and to market’s stakeholders, statistical information on insurance which could be very representative, prompt, and complete.

In 2013 Ivass started a quarterly survey on motor vehicles’ insurance

prices, called “Iper”, which became increasingly comprehensive, until

being able to consider a wide and very representative sample (about 8

million cars, representative of more than 30% of the whole motor

vehicles’ insurances sold in the national market). In order to use “Iper”

data for the consumer price index, some adjustments have been

required and implemented, in terms of methodological issues, frequency

and timeliness.

Starting from 2024 data elaborated on actual prices, referred to the sample of about 8 million

cars, will become available to Istat monthly. With the transition from a profile-based approach

(without any information about the number of contracts) towards actual prices, it will be possible

to properly represent the dynamics of this complex market. As a matter of fact, new data coming

from Iper are transaction prices, that bring information on the quantities of contracts signed, and

not only on a profile-based theoretical price and, as such, they represent the real market

behavior without the constraints that come from adopting a limited number of fixed consumers’

profiles.

Current vs New methodologyCurrent vs New methodology

NEW SURVEYCURRENT SURVEY

Sample of 80 Provinces (Nuts 3) All 107 Italian Provinces (Nuts 3)

Antonietta D’Amore ([email protected]), Francesca Ribaldi ([email protected]), Francesco Santangelo ([email protected]) ISTAT, ITALIAN NATIONAL INSTITUTE OF STATISTICS

About 1,500 prices surveyed monthly

Greater territorial coverage

More than 600,000 prices surveyed monthly

4 insured profiles

defined on the basis of geographical area, driver’s

age and seniority of his driving license, technical

characteristic and age of motor vehicles, risk profile

All insurance’s contracts

there is no reference to predefined profiles but all

insurance contracts, duly stratified by variables

affecting the price level and evolution, can be

potentially included in the sample

Greater representativeness

Difficulties in data collection

Many times, municipal data collectors aren’t

able in observing prices without having

concrete license plates number

A central survey carried out directly by ISTAT

Observed prices are the actual transaction

prices

Overcoming the difficulties

of field surveying

Prince index computation: two alternative methodsPrince index computation: two alternative methods

There are 4,280 stratum (107 provinces x 5 driver’s age class x 5 types of vehicle x 2 risk class). Weighted means are estimated for each stratum. Starting from these data, two computing methods can be applied,

depending on the period of relevance assigned to insurance prices. Indeed, insurance service is bought in a given month, but it lasts for one year. Therefore, considering only the prices of insurances sold in that

month, it does not take into account all the other living contracts (which have been sold in the previous eleven months). It might lead to overestimate, as well as underestimate (more probable given the market

tendencies), the actual price of insurance services for each month.

Considering insurances IN FORCE in the reference month: PRICES OF TWELVE MONTHSConsidering insurances IN FORCE in the reference month: PRICES OF TWELVE MONTHS

Considering insurances STIPULATED in the reference month: PRICES OF ONE MONTHConsidering insurances STIPULATED in the reference month: PRICES OF ONE MONTHHypothesis 1

Hypothesis 2

Istat – National institute of statistics

Ania – Association of insurance enterprises

Ivass – Authority for supervision of insurance

STAKEHOLDERS OF PARTNERSHIP

Insurance Enterprises

Central Bank of Italy

Indices, both with current and new approaches, show a decreasing trend over a considered period of time (2019-2021). It is possible to observe that the new

approach has a larger variation, greater in the first hypothesis than in the second one. The latter, indeed, embodying contracts of twelve months (and not only

those of one single month), has a more smoothed pattern. The differences in terms of variation are underlined by comparing monthly variations.

Meeting of the Group of Experts on Consumer Price Indices 7-9 June 2023, Palais des Nations, Geneva

Measuring geographical and population coverage in CPI internet

price collection: An application with groceries web scraping in

Italy

Tiziana Lauretia, Luigi Palumboab∗ aUniversità degli Studi della Tuscia, Viterbo, Italy

bBanca d’Italia, Roma, Italy ∗

Abstract

Consumer price indices (CPIs) are instrumental in the development of monetary policy and in mon- itoring economic developments. Prices collection for CPI compilation has come a long way in the past 20 years. However, while ideally, the index should include expenditure made by all households, urban and rural, throughout the country, CPIs in various countries have limited geographic coverage both for price collection and consumption expenditures. The introduction of new data sources, such as web scraping and scanner data, have contributed to reduce price collection costs and increase the reach across national territories, thus allowing to enhance the accuracy and quality of the CPI. The aim of this paper is to suggest a finer measurement CPI geographical coverage based on geostatistical fuzzy indices that would be particularly useful in cases where prices vary substantially across space, as it is proven that consumers only travel within limited extents for their purchases and a sparse network of outlets may lead to biased measurements. To explore the potential of the suggested mea- sure we estimate relative price levels across regions for a time period and price changes over the period for each region region-time-dummy method. This analyses is further validated by referring to structural breaks in our coverage metric and in spatio-temporal CPIs. Using a dataset deriving from geo-localized groceries web scraping in Italy, we provide a practical application calculating coverage at a regional level adopting different functional forms. Our findings corroborate the robustness of the proposed coverage metric and allow to embed information on geographic coverage in price statistics.

Keywords: geographical coverage; geostatistics; fuzzy logic; prices; web scraping.

1 Introduction

Consumer price indices (CPIs) measure price changes of the goods and services purchased by households in their role as consumers. CPIs are instrumental in the development of monetary policy and in monitoring economic developments. As a result, many policy debates have arisen surrounding the accuracy and reliability of price indices. Over the last decades, substantial progress has been made in developing new data sources, price collection methods, and related index calculation methods with the aim of reducing CPIs biases and errors (Smith, 2021). Price collection is becoming increasingly multimodal with prices being web scraped from the internet or obtained from scanner data, as well as being traditionally collected by collectors visiting individual outlets for several goods and services. Due to the fact that it is impossible to regularly record all the prices of the universe, CPIs are a sample statistics that represent the change in prices over the target universe in the two periods (International Monetary Fund, 2020).

Consequently, sampling techniques are used to select a subset of prices that enter the CPI compilation. The sampling process occurs on geographical location, outlet type, products and time dimensions. Within each of the different sampling levels, the sampling approach can differ from country to country, reflecting different ad- ministrative arrangements and practical reasons. Either probability or non-probability sampling methods can be adopted in each dimension. The geographic or spatial dimension is a key component in assessing the methodolog- ical soundness of the CPI (Berry, Graf, Stanger, & Ylä-Jarkko, 2019). concerning both product price collection and elementary price aggregation (average prices calculated for all item transactions in the country, province or state, city, neighbourhood). Diewert (2021) underlined that there is not a clear consensus on what the optimal degree of spatial disaggregation should be. Therefore, each NSI can make its own judgements on this matter, taking into account the costs of data collection and the demands of users for a spatial dimension for the CPI.

∗The views expressed herein are those of the authors and do not necessarily represent the views of the Bank of Italy and/or the Eurosystem.

1

Several CPI sampling operations are effectively cut-off samples with parts of the population of interest excluded thus producing a resulting in coverage error (Smith, 2021). Unless it is possible to sample outlets directly from a national sampling frame such as a business register (which often cannot identify small outlets or the precise range of products available in them), the sampling of outlets generally needs to be done in two stages. In the first stage, a sample of locations such as regions, cities or shopping areas is drawn/selected throughout the country, and in the second stage outlets are sampled.

When sampling locations, two major requirements must be considered: representativity and cost effectiveness. Areas where the bulk of consumer purchases take place need to be covered with certainty or by a probability sample to make the sample representative. Location samples are generally fixed for a long period of time, as they determine the whole organisation of work for the statistical office. When a large country is divided into administrative areas (state, region, province, etc.), all areas are often included with certainty, after which there may be sampling of locations within each of them, thus leading to increased representativity where price movements may differ due to different climates and/or transport costs. It is also a necessary requirement if regional CPIs are disseminated. In this case expenditure weights should be used to aggregate the regional indexes into the national ‘all-items’ index. If regional indexes are not disseminated, a representative sample of geographic areas can be selected for price collection, but the index weights should be based on the expenditure of all households in the country. In small countries it is common to select a few of the larger cities for price collection. This leaves out smaller towns and rural areas, but as consumers living in areas close to city will go there for some of their shopping the effect of their exclusion will be smaller than might be inferred from population numbers, and a sufficient coverage may still be achieved. It is then important that the selected cities are such that their outlets are used by a large part of the population and that they are situated in different parts of the country for maximum coverage. Car-friendly shopping centres situated immediately outside a city should be included if they are significant. While geographic coverage of CPIs is an indicator of quality, as ideally the index should include expenditure made by all households - urban and rural - throughout the country, little attention has been devoted to this issue in literature (Diewert, 2021; Guerreiro, Baer, & Silungwe, 2022; Hawkes & Piotrowski, 2003).

Many countries have CPIs with limited geographic coverage — capital city, including few of the largest areas (such as large and medium-sized cities) and prices are collected in urban areas only because their movements are considered to be representative of the price movements in rural areas. The geographical dimension, which is related to the scope of the index, becomes more important the smaller the region to which the index relates. The aim of this paper is to suggest a finer measurement of CPI geographical coverage to provide price statisticians with better insights on the actual reach of data collection. Since CPI compilation is becomes more important in economic planning and inflation monitoring, efforts should be made to expand the CPI to cover more geographic areas including all urban and rural areas. This would be particularly useful in cases where prices vary substantially across space, as it is proven that consumers only travel within limited extents for their purchases and a sparse network of outlets may lead to biased measurements.

The popularity and availability of new data sources for the compilation of the CPI, such as web-scraping and scanner data, has increased over the past twenty years and have contributed to reduce price collection costs and increased the reach across national territories, thus allowing to enhance the accuracy and quality of the CPI. Scanner data offer an opportunity to examine the effect of population exclusions, and Brunetti, Fatello, Polidoro, and Simone (2018) make such calculations for Italy, where sampling in the main CPI is restricted to the main provincial towns and uses only a sample of the most-sold products. They find only some differences, mainly due to sampling towns only, and concentrated in the south of Italy. Web-scraped data collection has been increasingly used by NSIs recent years and many countries are developing web-scraping tools tailored to specific CPIs requirements that allow to quickly collect large numbers of prices for a wide variety of online products and cover new consumpetion segments (Eurostat, 2020). Although the compilation of price indices from such large datasets is not straightforward, these new sources of data have proved to be of benefit to CPIs thanks to the detailed information available for individual products and the wide coverage both in terms of product groups and territorial areas.

We suggest a geostatistical fuzzy index (Zadeh, 1977; Zimmermann, 2011) to measure the reach of data collection in terms of geographical and population coverage of outlets where prices are collected. This index may be used to evaluate the degree of coverage for price data collection, both in the context of probabilistic and non- probabilistic outlet selection. An advantage of the fuzzy set theory approach is to overcome the limits of discrete classifications of data, preserving a higher degree of information for analysis. A properly designed membership functions may enable us to achieve a better classification of the data, smoothing distortions caused by outliers while still including them into the analysis. Using a fuzzy membership function it is possible to calculate the coverage value for each municipality, inversely proportional the driving distance in minutes from the closest outlet where prices have been collected. Total coverage value for a given territory is calculated as an average over municipalities coverage values, either using unweighted or weighted formulae.

Using a dataset deriving from geo-localized groceries web scraping in Italy, we provide a practical application calculating coverage at a regional level and comparing results from two different functional forms – linear and non-linear – as well as a set of different parameters for spatial decay of coverage. Our findings corroborate the robustness of the coverage index, as rank correlations amongst different parameters and functional values are close to 1 and statistically significant. In addition, with the aim of emphasising the importance of measuring and

Page 2 of 15

monitoring the degree of coverage, we carry out spatial-temporal panel comparisons using the Time-interaction- Region Product Dummy (TiRPD) method to assess what happens to consumer price indices when there is an abrupt change in coverage. TiRPD is a natural extension of the well known time-dummy and region-dummy methods which have widely been used in literature for constructing consumer price indices (Corrado & Ukhaneva, 2016). We conclude by illustrating our methodology using web scraped data from 616 online supermarkets belonging to 23 different retail chains in Italy from November 2020 to February 2023. Abrupt change in spatio- temporal CPIs and coverage are identified using a Bayesian estimator, in order to validate the link amongst the two phenomena.

The remainder of this paper is structured as follows. Section 2 illustrates the methodological approach. Section 3 describes the data and reports descriptive statistics. Section 4 presents the main results of the empirical analysis. Finally, Section 5 draws some conclusions.

2 Methodology

2.1 Fuzzy Coverage Index

The fundamental concept behind our proposed measure of coverage for price collection is that price information decays with space and travel time. Consumers may travel for certain distances and time to make purchases, thus providing an incentive for sellers to maintain competitive prices in different municipalities (see for example Kerr et al., 2012). However, consumers’ inclination to commit time and money for purchasing trips is directly connected to the expected economic benefit in terms of savings.

Given the average basket value for groceries shopping, it is reasonable to affirm that there are limits to shopping trips distances, even if those may vary between consumers because of different travel costs, cost-opportunity of travel time and other individual characteristics.

Empirical evidence of spatial effects underlying consumer price differences among geographical areas have been observed both at country and sub-national level (Aten, 1996; Biggeri, Laureti, & Polidoro, 2017; Montero, Laureti, Mı́nguez, & Fernández-Avilés, 2020; Rao, 2001). Therefore, we need to conclude that prices may be different between municipalities situated at a certain distance, and the information value of collected prices in a certain location will decay with space and travel time.

In order to appropriately model this decay we resort to Fuzzy Set theory, as it seems inappropriate to specify hard boundaries regarding the validity of price information in binary terms. We then propose two different membership functions to calculate the coverage value for each municipality. The first one is a simple linear function, where coverage is inversely proportional to travel distance.

lc(x) = max(1 − x

D , 0) (1)

Where x is the travel time by car in minutes between a municipality and the closest municipality where prices have been collected, and D is a parameter indicating at which travel time level the price information is considered no longer valid.

The second type of membership function is based on an inverse sigmoid modeling of price information decay. In fact, it is reasonable to assume that consumer willingness to travel for purchases is not linear, therefore price persistence in space is relatively stronger at short distances and weaker at longer ones. We can then propose a different membership function as follow:

c(x) = 1 − 1

1 + e−k(x−D 2 )

(2)

Where x is again the travel time by car in minutes between a municipality and the closest municipality where prices have been collected, D is a parameter indicating at which travel time level the price information is considered no longer valid (and D

2 is the midpoint of the inverse sigmoid), and k is a parameter indicating the

steepness of the inverse sigmoid curve. Once we calculate coverage values for all municipalities in a region, we need to synthesize a metric to indicate

the overall coverage for the region. In order to do so, we can aggregate individual municipalities as units or by weighting them according to their population.

If we chose to treat municipalities as individual units, the coverage for a given region could be expressed as a simple arithmetic mean as in (3).

Cmun =

∑n

i=1 ci

n (3)

On the other side, if we chose to weight coverage in each municipality by its population the overall Region coverage would be:

Cpop =

∑n

i=1 ci ∗ popi∑n

i=1 popi

(4)

Page 3 of 15

Formulas for coverage are applied at a regional level, since the main purpose is to provide a coverage metric for sub-national spatio-temporal price indices over space and time.

2.2 Spatio-temporal Price Indices

While there are several methodologies that could be applied to calculate regional SPIs (Laureti & Rao, 2018), for this work we selected a Time-interaction-Region Product Dummy (TiRPD) model which helps us reconcile aggregate SPIs across space and time and it has already been applied to price data from web scraping (Benedetti, Laureti, Palumbo, & Rose, 2022). This model was first proposed by Aizcorbe and Aten (2004), who referred to it as the Time-interaction-Country Product Dummy method. This model was designed as combination between the Country Product Dummy (CPD) model (Summers, 1973), which focuses on spatial price variation, and the Time Product Dummy model (TPD) (Aizcorbe, Corrado, & Doms, 2000; De Haan & Krsinich, 2014), which focuses on price variation over time. The specification of the TiRPD model is:

lnPijt =

N∑ i=1

βiDijt +

T∑ t=1

M∑ j=1

δjtRijTjt + ηijt (5)

Where lnPijt and Dijt are respectively the log-price and the dummy for product i in area j at time t (i = 1, 2, ..., N ; j = 1, 2, ...,M ; t = 1, 2, ..., T ). Rij and Tjt are dummy variables for each combination of area and time period. A price index for each region-period jt is obtained directly from the parameter of the dummy by exponentiation of the δjt coefficient, and it is possible to perform direct comparisons across regions or between time as TiRPD is a multilateral method. The TiCPD provides the same answers as separate CPD or TPD models, with the advantage that it normalizes the relationships on a single region and time period.

2.3 Structural breaks

In order to further validate the importance of measuring and monitoring coverage, we assess what happens when there are structural breaks. For this purpose, we use the Bayesian Estimator of Abrupt change, Seasonal change, and Trend (BEAST) proposed by Zhao et al. (2019), and implemented in the R package Rbeast.

The BEAST model, a Bayesian statistical model that performs time series decomposition into multiple trend and seasonal signals, provides us with the probability for each of the time series points to be a trend change point. The general form of the model is:

yi = S(ti; Θs) + T (ti; Θt) + εi (6)

where yi is the observed value at time ti, Θs and Θt are respectively the season and trend signals, and εi is noise with an assumed Gaussian distribution. Given the relative short length of our time series, we removed the seasonal component from the model, which is then formalized as:

yi = T (ti; Θt) + εi (7)

Trend change points are implicitly encoded in Θt, and the trend function is modeled as a piecewise linear function with m knots and m+ 1 segments. In each segment, the trend is built as:

T (t) = aj + bjt for τj ≤ t < τj+1, j = 0, ...,m (8)

where aj and bj are parameters for the linear trend in the j segment, which spans from τj to τj+1. Further details about the Bayesian formulation of BEAST, its Markov Chain Monte Carlo inference and

posterior inference of change points, seasonality, and trends can be found in Zhao et al. (2019). 1

For our purposes, once we obtain the probability of trend change for each time point in each Region for the coverage and spatio-temporal price index level, we first check the stationarity of both time series using the Kwiatkowski-Phillips-Schmidt-Shin (KPSS) test (Kwiatkowski, Phillips, Schmidt, & Shin, 1992), and then we calculate the Pearson correlation between the two series. A positive significant correlation would signal a potential effect on the estimated price level from the abrupt change in the coverage.

3 Data

Data used for the empirical validation of our proposed methodology has been scraped from 616 online supermarkets belonging to 23 different chains in 19 Italian Regions from November 2020 to February 20232. The portfolio of online supermarkets changed over time, as new sources were added and other became unavailable, due to failure in the scraping routines or anti-scraping measures implemented by the source website.

1By construction, the probability of being a trend change point is additive over time. In other words, the total probability of encountering a trend change point between time t and s equals the sum of all probabilities for time points between t and s.

2No data has been collected for the Trentino-Alto Adige region.

Page 4 of 15

Each online supermarket has been located with GPS coordinates and placed in a specific municipality using geographical merging functions. We collected prices for each supermarket using the “pick up” option for purchase delivery. Therefore, validity of price information is considered linked to its geographical position.

For exemplification purposes, we selected the Coffee category (ECOICOP code 01.2.1.1), which in 2021 ac- counted for an average Italian household monthly expenditure of 11.91 EUR. Weights for the 5 digit subclass Coffee ranges from 0.38% in 2020 to 0.43% in 2023. Our data collection for this category amounted to 5338 unique products and 1221755 total observations. In 2056 cases we were able to identify products using their Global Trade Item Number (GTIN), and therefore we could accurately match them across different retail chains. In the other cases, instead, each unique product was identified according to retailer-specific attributes such as product code or product name and could not be automatically matched across different retail chains.

We classified products according to their commercial category, which is not standard across retailer, and with filters based on inclusion (or exclusion) of specific terms in the product name when the commercial category was not sufficiently specific. It should be noted that product naming could vary substantially across retailer for the same product - for instance: the inclusion or exclusion of the category name, brand and size or the use of abbreviations - but at the same time different products in the same category may be quite similarly named.

Our products are overlapping across the different time periods and regions, as showed in Figures 1 and 2. The large number of common products is reassuring for the validity of our results, as if there is little overlap across time and space price levels are inherently difficult to compare (Hill & Timmer, 2006).

Figure 1: Common products across time periods.

507 507

2103

507

1996

2254

507

1675

1814

2384

507

1657

1771

2209

2320

505

1607

1718

2132

2221

2327

500

1573

1675

2085

2169

2246

2295

493

1529

1624

2023

2105

2171

2199

2305

494

1447

1539

1941

2022

2089

2102

2177

3308

490

1400

1494

1890

1969

2033

2048

2116

3192

3275

486

1394

1482

1882

1958

2023

2032

2101

3135

3182

3298

482

1425

1514

1906

1989

2051

2059

2104

3095

3119

3206

3406

483

1412

1501

1877

1957

2016

2025

2050

3025

3042

3109

3263

3397

480

1345

1429

1787

1856

1911

1914

1936

2795

2809

2865

2988

3086

3226

478

1312

1391

1663

1716

1731

1741

1758

2533

2537

2584

2698

2779

2859

2996

475

1251

1322

1593

1649

1664

1672

1688

2454

2461

2499

2616

2671

2723

2833

2931

460

1236

1290

1546

1603

1620

1626

1638

2382

2389

2427

2536

2586

2632

2733

2806

2880

459

1207

1260

1513

1565

1583

1591

1601

2289

2300

2348

2452

2488

2526

2622

2677

2727

2821

460

1189

1244

1497

1549

1565

1570

1581

2295

2303

2342

2433

2477

2511

2610

2663

2700

2741

2915

460

1160

1209

1464

1517

1528

1536

1547

2239

2246

2285

2388

2426

2450

2536

2584

2611

2628

2755

2864

347

880

920

1151

1200

1210

1217

1230

1904

1906

1935

2031

2052

2039

2070

2097

2122

2112

2173

2242

2296

336

641

679

902

950

960

968

993

1563

1567

1592

1653

1667

1654

1685

1707

1729

1727

1782

1832

1862

1914

327

631

669

889

938

949

955

982

1532

1538

1562

1614

1630

1613

1636

1661

1687

1683

1736

1786

1808

1845

1949

319

617

655

873

921

931

936

962

1475

1482

1506

1540

1550

1534

1560

1584

1604

1597

1651

1698

1718

1737

1828

1904

317

610

649

856

904

913

919

945

1428

1430

1453

1493

1508

1491

1518

1539

1558

1562

1612

1649

1677

1707

1788

1830

1882

324

614

653

860

904

913

919

946

1428

1429

1451

1491

1507

1492

1518

1541

1560

1560

1608

1646

1672

1698

1775

1811

1837

1887

321

584

623

825

869

876

884

907

1391

1392

1413

1453

1470

1451

1479

1499

1519

1517

1569

1612

1629

1651

1728

1766

1780

1814

1856

320

576

615

815

857

866

875

899

1379

1380

1399

1438

1453

1436

1463

1485

1505

1502

1556

1595

1611

1630

1701

1749

1761

1787

1811

1842

2020−11

2020−12

2021−01

2021−02

2021−03

2021−04

2021−05

2021−06

2021−07

2021−08

2021−09

2021−10

2021−11

2021−12

2022−01

2022−02

2022−03

2022−04

2022−05

2022−06

2022−07

2022−08

2022−09

2022−10

2022−11

2022−12

2023−01

2023−02

20 20

−1 1

20 20

−1 2

20 21

−0 1

20 21

−0 2

20 21

−0 3

20 21

−0 4

20 21

−0 5

20 21

−0 6

20 21

−0 7

20 21

−0 8

20 21

−0 9

20 21

−1 0

20 21

−1 1

20 21

−1 2

20 22

−0 1

20 22

−0 2

20 22

−0 3

20 22

−0 4

20 22

−0 5

20 22

−0 6

20 22

−0 7

20 22

−0 8

20 22

−0 9

20 22

−1 0

20 22

−1 1

20 22

−1 2

20 23

−0 1

20 23

−0 2

1000 2000 3000

Common products

Driving distances between municipalities have been obtained from a distance matrix published by the Italian National Institute of Statistics (Istat). Istat calculated the driving distance between centroids for all Italian mu- nicipalities in 2013 using a commercial road graph (Istat, 2019). We performed a basic elaboration in order to adjust for merging between small municipalities in the 2013-2021 period, also excluding minor islands and mu- nicipalities disconnected from the road graph3. The total amount of population living in excluding municipalities is marginal when compared to the relative Region population.

3Municipalities of Monte Isola (BS) and Campione d’Italia (CO) do not have any connection with the road graph used for distance calculation. Istat only provides distance from the closest municipality for them. For minor islands Istat provides a travel time by ferry to the closest port.

Page 5 of 15

Figure 2: Common products across regions.

947 316

357

326

276

916

392

317

644

1003

258

162

388

455

1087

89

75

112

100

182

194

589

199

608

710

589

145

2435

148

111

159

184

239

120

454

640

233

142

385

432

787

133

535

223

1392

175

90

174

174

262

176

233

143

200

316

310

119

168

188

181

76

255

105

173

161

329

172

115

182

205

318

129

278

245

254

164

139

391

324

266

298

361

225

72

273

145

206

141

157

155

466

159

120

188

207

259

119

680

421

231

158

122

223

168

1042

174

130

307

304

308

95

332

124

317

141

126

127

172

142

698

312

111

154

162

242

146

798

390

171

168

101

177

113

697

118

1018

545

171

535

631

507

113

956

191

452

182

168

230

230

208

290

409

1181

71

61

84

74

105

82

105

114

110

90

57

114

55

86

65

89

75

115

172

129

194

221

410

187

293

211

255

228

132

292

150

218

140

225

311

103

733

Abruzzo

Basilicata

Calabria

Campania

Emilia−Romagna

Friuli−Venezia Giulia

Lazio

Liguria

Lombardia

Marche

Molise

Piemonte

Puglia

Sardegna

Sicilia

Toscana

Umbria

Valle D’aosta

Veneto

Abr uz

zo

Bas ilic

at a

C al ab

ria

C am

pa ni a

Em ilia

−R om

ag na

Friu li−

Ve ne

zi a

G iu lia

La zi o

Li gu

ria

Lo m

ba rd

ia

M ar

ch e

M ol is e

Pie m

on te

Pug lia

Sar de

gn a

Sic ilia

To sc

an a

U m

br ia

Val le D

’a os

ta

Ve ne

to

500 100015002000

Common products

4 Results

In table 1 we report the results for coverage in December 2021 by Region calculated according to the different membership functions presented in (1) and (2), using as weight both individual municipalities, according to (3), and their population according to (4) in order to calculate the overall regional coverage and utilizing different D parameters for distance. Figures 3 to 6 are graphical representations of municipalities’ coverage values according to the above mentioned membership functions at selected values of D in December 2021.

In order to evaluate the stability and consistence of our coverage metrics, we performed a series of measurement leveraging the Spearman Rank Correlation non-parametric test on the coverage values calculated for each region in December 2021, as presented in Table 1. Results are presented in Table 2, and we can appreciate that rank correlations are very strong and significant in all cases, indicating that our proposed indicator can deliver robust and consistent results irrespective of the parameters chosen. The link is somehow less strong between population-weighted indexes and municipalities-weighted ones, but within each methodology seems fairly stable and consistent. Results for other months deliver a substantially identical picture and are available under request.

We complete the illustration in Table 3 pairing our spatio-temporal price indices for Coffee, computed with the TiPRD equation as in (5) taking as normalization prices in Lazio region in December 2021 with coverage values for each month and region. We can note a marked upward trend in prices starting in 2022, clearly showed in Figure 7.

The spatio-temporal indexes calculated in our exercise reflect the specific composition of retailers in our web scraping operations, as well as the addition and termination of data sources. In our specific example, different retailers may have very different positioning and geographical presence. The main purpose of the illustration in Table 3 is to show the pattern of structural breaks in each Region when there is a structural break in the coverage index

Finally, we calculated the correlation between the trend change point probability for the coverage and spatio- temporal price index level time series for each region. The KPSS test failed to reject the null hypothesis of stationarity in all cases. Results from the Pearson correlation test are presented in Table 5.

Page 6 of 15

T ab

le 1:

C ov

er ag

e va

lu e

in D

ec em

b er

2 0 2 1

b y

R eg

io n

a cc

o rd

in g

to d

iff er

en t

sp ec

ifi ca

ti o n

s a n

d p

a ra

m et

er s.

M u

n ic

ip al

it ie

s P

o p

u la

ti o n

S ig

m oi

d L

in ea

r S

ig m

o id

L in

ea r

R eg

io n

20 30

40 50

20 3 0

4 0

5 0

2 0

3 0

4 0

5 0

2 0

3 0

4 0

5 0

A b

ru zz

o 0.

70 5

0. 56

8 0.

41 6

0. 26

2 0.

63 3

0 .5

4 6

0 .4

1 8

0 .2

5 0

0 .9

1 0

0 .8

4 5

0 .7

5 4

0 .6

4 8

0 .8

4 3

0 .8

0 5

0 .7

4 4

0 .6

4 1

B as

il ic

at a

0. 16

6 0.

11 9

0. 08

0 0.

05 0

0. 18

8 0 .1

3 0

0 .0

8 3

0 .0

4 6

0 .3

5 5

0 .2

9 7

0 .2

4 9

0 .2

0 7

0 .3

7 6

0 .3

0 7

0 .2

5 3

0 .2

0 3

C al

ab ri

a 0.

48 2

0. 36

4 0.

25 2

0. 14

9 0.

46 6

0 .3

7 5

0 .2

5 8

0 .1

4 1

0 .6

3 1

0 .5

4 9

0 .4

7 0

0 .3

9 7

0 .6

1 9

0 .5

5 5

0 .4

7 4

0 .3

9 2

C am

p an

ia 0.

59 0

0. 49

2 0.

38 3

0. 26

4 0.

55 7

0 .4

7 5

0 .3

7 3

0 .2

6 1

0 .8

6 1

0 .7

9 6

0 .7

0 3

0 .5

8 0

0 .7

9 6

0 .7

5 0

0 .6

8 1

0 .5

8 0

E m

il ia

-R om

ag n

a 0.

58 5

0. 47

7 0.

36 4

0. 25

2 0.

55 3

0 .4

6 8

0 .3

6 5

0 .2

4 4

0 .7

5 1

0 .6

7 4

0 .5

8 8

0 .5

0 3

0 .7

2 4

0 .6

6 4

0 .5

8 5

0 .4

9 7

F ri

u li

-V en

ez ia

G iu

li a

0. 74

8 0.

62 9

0. 48

4 0.

32 2

0. 66

6 0 .5

8 8

0 .4

7 6

0 .3

1 4

0 .6

9 8

0 .6

0 1

0 .4

8 1

0 .3

3 6

0 .6

3 7

0 .5

5 1

0 .4

6 8

0 .3

3 1

L az

io 0.

52 6

0. 39

4 0.

27 1

0. 16

6 0.

52 0

0 .4

1 3

0 .2

8 1

0 .1

5 6

0 .8

7 4

0 .8

3 1

0 .7

8 5

0 .7

4 1

0 .8

6 0

0 .8

2 8

0 .7

8 6

0 .7

3 8

L ig

u ri

a 0.

43 1

0. 31

2 0.

20 4

0. 11

3 0.

44 3

0 .3

3 3

0 .2

1 2

0 .1

0 4

0 .7

3 6

0 .6

5 7

0 .5

8 4

0 .5

2 2

0 .7

3 0

0 .6

6 7

0 .5

9 3

0 .5

1 5

L om

b ar

d ia

0. 60

0 0.

47 9

0. 34

8 0.

21 7

0. 56

2 0 .4

6 6

0 .3

4 6

0 .2

0 8

0 .8

1 3

0 .7

1 3

0 .5

8 6

0 .4

4 5

0 .7

3 6

0 .6

7 3

0 .5

7 9

0 .4

3 7

M ar

ch e

0. 51

6 0.

38 7

0. 26

1 0.

14 7

0. 49

6 0 .3

9 3

0 .2

6 9

0 .1

3 7

0 .7

2 8

0 .6

0 4

0 .4

6 5

0 .3

3 4

0 .6

5 8

0 .5

8 3

0 .4

7 4

0 .3

1 6

M ol

is e

0. 26

5 0.

18 8

0. 11

9 0.

06 2

0. 27

5 0 .2

0 2

0 .1

2 4

0 .0

5 7

0 .3

3 1

0 .2

8 5

0 .2

4 3

0 .2

0 4

0 .3

4 3

0 .2

8 9

0 .2

4 4

0 .2

0 2

P ie

m on

te 0.

20 7

0. 14

8 0.

09 4

0. 04

7 0.

21 6

0 .1

5 6

0 .0

9 7

0 .0

4 3

0 .4

7 2

0 .4

1 9

0 .3

6 0

0 .2

9 9

0 .4

5 8

0 .4

0 9

0 .3

5 7

0 .2

9 5

P u

gl ia

0. 31

8 0.

23 5

0. 16

2 0.

10 2

0. 33

7 0 .2

5 0

0 .1

6 8

0 .0

9 7

0 .5

3 6

0 .4

5 7

0 .3

8 5

0 .3

2 3

0 .5

4 3

0 .4

7 1

0 .3

8 7

0 .3

1 8

S ar

d eg

n a

0. 44

2 0.

35 6

0. 26

7 0.

18 2

0. 43

7 0 .3

5 2

0 .2

6 5

0 .1

7 6

0 .6

5 9

0 .6

0 2

0 .5

3 8

0 .4

7 3

0 .6

4 6

0 .5

9 2

0 .5

3 4

0 .4

6 8

S ic

il ia

0. 26

8 0.

19 2

0. 12

5 0.

06 7

0. 27

8 0 .2

0 6

0 .1

3 1

0 .0

6 4

0 .5

0 3

0 .4

4 8

0 .3

9 2

0 .3

3 9

0 .4

9 1

0 .4

4 8

0 .3

9 5

0 .3

3 5

T os

ca n

a 0.

44 5

0. 33

6 0.

23 3

0. 14

5 0.

44 6

0 .3

5 1

0 .2

3 9

0 .1

3 7

0 .7

1 8

0 .6

3 7

0 .5

5 1

0 .4

7 0

0 .6

9 9

0 .6

3 6

0 .5

5 2

0 .4

6 4

U m

b ri

a 0.

77 6

0. 65

7 0.

51 4

0. 36

4 0.

69 8

0 .6

2 3

0 .5

0 8

0 .3

5 2

0 .9

4 6

0 .9

0 7

0 .8

5 1

0 .7

8 8

0 .9

0 6

0 .8

8 2

0 .8

4 5

0 .7

8 2

V al

le d

’A os

ta 0.

45 3

0. 33

7 0.

23 0

0. 13

4 0.

46 6

0 .3

5 3

0 .2

3 3

0 .1

2 7

0 .4

9 6

0 .3

4 7

0 .2

2 9

0 .1

2 3

0 .5

1 0

0 .3

9 4

0 .2

3 0

0 .1

2 0

V en

et o

0. 70

0 0.

59 0

0. 45

5 0.

30 9

0. 62

9 0 .5

5 0

0 .4

5 0

0 .2

9 8

0 .7

4 3

0 .6

2 2

0 .4

8 0

0 .3

2 8

0 .6

6 9

0 .5

8 9

0 .4

7 4

0 .3

1 6

Page 7 of 15

Figure 3: Coverage representation in December 2021 - Linear membership function - D: 20 min.

Figure 4: Coverage representation in December 2021 - Linear membership function - D: 50 min.

Page 8 of 15

Figure 5: Coverage representation in December 2021 - Sigmoid membership function - D: 20 min.

Figure 6: Coverage representation in December 2021 - Sigmoid membership function - D: 50 min.

Page 9 of 15

T ab

le 2:

S p

ea rm

an R

an k

C o rr

el a ti

o n

T es

t b

et w

ee n

p a ir

s o f

m em

b er

sh ip

fu n

ct io

n s.

W M

u n ic ip a li ti e s

P o p u la ti o n

F u n

S ig m o id

L in e a r

S ig m o id

L in e a r

M in

2 0

3 0

4 0

5 0

2 0

3 0

4 0

5 0

2 0

3 0

4 0

5 0

2 0

3 0

4 0

5 0

M u n

S ig

2 0

1 .0 0 0

0 .9 9 3

0 .9 8 1

0 .9 5 6

0 .9 9 8

0 .9 8 8

0 .9 7 4

0 .9 4 6

0 .6 0 4

0 .7 3 0

0 .7 3 7

0 .7 9 5

0 .5 8 9

0 .6 7 0

0 .6 9 8

0 .7 3 2

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 3 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 4 )

(0 .0 0 1 )

(0 .0 0 0 )

(0 .0 0 0 )

3 0

0 .9 9 3

1 .0 0 0

0 .9 9 1

0 .9 7 4

0 .9 8 9

0 .9 9 8

0 .9 8 6

0 .9 6 5

0 .6 1 1

0 .7 4 4

0 .7 6 3

0 .8 2 3

0 .5 9 3

0 .6 8 8

0 .7 1 8

0 .7 5 6

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 3 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 4 )

(0 .0 0 1 )

(0 .0 0 0 )

(0 .0 0 0 )

4 0

0 .9 8 1

0 .9 9 1

1 .0 0 0

0 .9 9 1

0 .9 7 4

0 .9 9 5

0 .9 9 6

0 .9 8 6

0 .5 7 0

0 .7 1 1

0 .7 4 7

0 .8 1 8

0 .5 4 9

0 .6 5 8

0 .6 9 6

0 .7 4 6

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 5 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 7 )

(0 .0 0 1 )

(0 .0 0 0 )

(0 .0 0 0 )

5 0

0 .9 5 6

0 .9 7 4

0 .9 9 1

1 .0 0 0

0 .9 5 1

0 .9 8 1

0 .9 9 1

0 .9 9 8

0 .5 5 8

0 .7 0 5

0 .7 5 3

0 .8 2 1

0 .5 3 7

0 .6 4 9

0 .6 9 5

0 .7 5 3

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 7 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 9 )

(0 .0 0 1 )

(0 .0 0 0 )

(0 .0 0 0 )

L in

2 0

0 .9 9 8

0 .9 8 9

0 .9 7 4

0 .9 5 1

1 .0 0 0

0 .9 8 2

0 .9 6 7

0 .9 4 0

0 .6 1 2

0 .7 3 9

0 .7 4 0

0 .7 9 3

0 .6 0 2

0 .6 7 5

0 .7 0 4

0 .7 3 5

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 3 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 3 )

(0 .0 0 1 )

(0 .0 0 0 )

(0 .0 0 0 )

3 0

0 .9 8 8

0 .9 9 8

0 .9 9 5

0 .9 8 1

0 .9 8 2

1 .0 0 0

0 .9 9 1

0 .9 7 4

0 .6 0 0

0 .7 3 7

0 .7 6 5

0 .8 2 8

0 .5 7 9

0 .6 8 4

0 .7 1 4

0 .7 5 8

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 3 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 5 )

(0 .0 0 1 )

(0 .0 0 0 )

(0 .0 0 0 )

4 0

0 .9 7 4

0 .9 8 6

0 .9 9 6

0 .9 9 1

0 .9 6 7

0 .9 9 1

1 .0 0 0

0 .9 8 8

0 .5 5 4

0 .6 9 5

0 .7 3 5

0 .8 0 9

0 .5 3 3

0 .6 4 2

0 .6 7 9

0 .7 3 5

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 7 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 9 )

(0 .0 0 2 )

(0 .0 0 1 )

(0 .0 0 0 )

5 0

0 .9 4 6

0 .9 6 5

0 .9 8 6

0 .9 9 8

0 .9 4 0

0 .9 7 4

0 .9 8 8

1 .0 0 0

0 .5 6 1

0 .7 0 9

0 .7 6 0

0 .8 2 8

0 .5 4 0

0 .6 5 6

0 .7 0 0

0 .7 6 1

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 6 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 8 )

(0 .0 0 1 )

(0 .0 0 0 )

(0 .0 0 0 )

P o p

S ig

2 0

0 .6 0 4

0 .6 1 1

0 .5 7 0

0 .5 5 8

0 .6 1 2

0 .6 0 0

0 .5 5 4

0 .5 6 1

1 .0 0 0

0 .9 6 1

0 .9 0 9

0 .8 6 0

0 .9 9 3

0 .9 6 3

0 .9 3 9

0 .8 9 3

(0 .0 0 3 )

(0 .0 0 3 )

(0 .0 0 5 )

(0 .0 0 7 )

(0 .0 0 3 )

(0 .0 0 3 )

(0 .0 0 7 )

(0 .0 0 6 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

3 0

0 .7 3 0

0 .7 4 4

0 .7 1 1

0 .7 0 5

0 .7 3 9

0 .7 3 7

0 .6 9 5

0 .7 0 9

0 .9 6 1

1 .0 0 0

0 .9 7 0

0 .9 4 2

0 .9 5 4

0 .9 8 2

0 .9 7 5

0 .9 5 4

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

4 0

0 .7 3 7

0 .7 6 3

0 .7 4 7

0 .7 5 3

0 .7 4 0

0 .7 6 5

0 .7 3 5

0 .7 6 0

0 .9 0 9

0 .9 7 0

1 .0 0 0

0 .9 8 8

0 .8 9 5

0 .9 7 7

0 .9 8 9

0 .9 9 1

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

5 0

0 .7 9 5

0 .8 2 3

0 .8 1 8

0 .8 2 1

0 .7 9 3

0 .8 2 8

0 .8 0 9

0 .8 2 8

0 .8 6 0

0 .9 4 2

0 .9 8 8

1 .0 0 0

0 .8 4 2

0 .9 4 7

0 .9 6 5

0 .9 8 2

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

L in

2 0

0 .5 8 9

0 .5 9 3

0 .5 4 9

0 .5 3 7

0 .6 0 2

0 .5 7 9

0 .5 3 3

0 .5 4 0

0 .9 9 3

0 .9 5 4

0 .8 9 5

0 .8 4 2

1 .0 0 0

0 .9 4 9

0 .9 2 8

0 .8 7 9

(0 .0 0 4 )

(0 .0 0 4 )

(0 .0 0 7 )

(0 .0 0 9 )

(0 .0 0 3 )

(0 .0 0 5 )

(0 .0 0 9 )

(0 .0 0 8 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

3 0

0 .6 7 0

0 .6 8 8

0 .6 5 8

0 .6 4 9

0 .6 7 5

0 .6 8 4

0 .6 4 2

0 .6 5 6

0 .9 6 3

0 .9 8 2

0 .9 7 7

0 .9 4 7

0 .9 4 9

1 .0 0 0

0 .9 8 8

0 .9 6 7

(0 .0 0 1 )

(0 .0 0 1 )

(0 .0 0 1 )

(0 .0 0 1 )

(0 .0 0 1 )

(0 .0 0 1 )

(0 .0 0 2 )

(0 .0 0 1 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

4 0

0 .6 9 8

0 .7 1 8

0 .6 9 6

0 .6 9 5

0 .7 0 4

0 .7 1 4

0 .6 7 9

0 .7 0 0

0 .9 3 9

0 .9 7 5

0 .9 8 9

0 .9 6 5

0 .9 2 8

0 .9 8 8

1 .0 0 0

0 .9 8 8

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 1 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

5 0

0 .7 3 2

0 .7 5 6

0 .7 4 6

0 .7 5 3

0 .7 3 5

0 .7 5 8

0 .7 3 5

0 .7 6 1

0 .8 9 3

0 .9 5 4

0 .9 9 1

0 .9 8 2

0 .8 7 9

0 .9 6 7

0 .9 8 8

1 .0 0 0

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

(0 .0 0 0 )

Page 10 of 15

T ab

le 3:

S p

at io

-t em

p o ra

l p ri

ce in

d ex

le ve

ls fo

r C

o ff

ee (E

C O

IC O

P 0 1 .2

.1 .1

). In

p ar

en th

es is

co ve

ra ge

va lu

es ca

lc u

la te

d w

it h

th e

S ig

m o id

fu n

ct io

n a n

d w

ei g h te

d b y

p o u

p la

ti o n

(D =

3 0

m in

)

M o n th

A b ru

z z o

B a si li c a ta

C a la b ri a

C a m p a n ia

E m il ia -R

o m a g n a

F ri u li -V

e n e z ia

G iu li a

L a z io

L ig u ri a

L o m b a rd

ia M

a rc h e

M o li se

2 0 2 0 -1 1

1 0 4 .5 5

1 0 2 .8 8

1 0 2 .0 0

1 0 1 .7 9

(0 .6 6 8 )

(0 .5 6 2 )

(0 .2 0 7 )

(0 .2 4 3 )

2 0 2 0 -1 2

1 0 5 .6 0

9 7 .0 0

9 7 .2 9

1 0 1 .4 3

1 0 1 .1 0

1 0 2 .5 2

9 9 .4 8

1 0 3 .5 6

1 0 5 .0 3

(0 .6 9 5 )

(0 .2 7 2 )

(0 .4 7 7 )

(0 .1 2 2 )

(0 .6 9 0 )

(0 .4 9 8 )

(0 .3 5 9 )

(0 .2 0 7 )

(0 .2 4 3 )

2 0 2 1 -0 1

1 0 5 .4 9

1 0 1 .4 8

1 0 0 .3 0

1 0 1 .9 2

1 0 1 .7 5

1 0 1 .6 4

9 9 .8 8

1 0 3 .4 6

1 0 4 .5 6

(0 .7 0 0 )

(0 .2 7 2 )

(0 .5 1 2 )

(0 .1 2 2 )

(0 .6 9 0 )

(0 .4 9 8 )

(0 .4 0 9 )

(0 .2 0 7 )

(0 .2 4 3 )

2 0 2 1 -0 2

1 0 4 .9 9

1 0 1 .8 6

1 0 0 .7 0

9 9 .6 8

1 0 1 .1 7

9 9 .4 4

1 0 1 .0 0

1 0 3 .3 4

1 0 0 .0 8

1 0 1 .1 1

1 0 4 .7 7

(0 .7 0 0 )

(0 .1 7 7 )

(0 .3 6 9 )

(0 .5 5 6 )

(0 .5 1 0 )

(0 .4 9 7 )

(0 .7 2 5 )

(0 .5 6 7 )

(0 .4 7 0 )

(0 .5 7 4 )

(0 .2 4 3 )

2 0 2 1 -0 3

1 0 4 .9 8

1 0 2 .2 1

1 0 0 .9 8

9 9 .8 0

1 0 0 .5 7

9 9 .7 9

1 0 0 .2 9

1 0 3 .1 1

9 9 .6 7

1 0 1 .5 7

1 0 4 .3 1

(0 .7 0 0 )

(0 .1 7 7 )

(0 .3 6 9 )

(0 .5 6 1 )

(0 .5 1 0 )

(0 .4 9 7 )

(0 .7 2 5 )

(0 .5 6 7 )

(0 .4 7 0 )

(0 .5 7 4 )

(0 .2 4 3 )

2 0 2 1 -0 4

1 0 4 .7 3

1 0 1 .6 2

1 0 1 .3 3

9 9 .7 3

1 0 1 .1 2

1 0 0 .0 0

1 0 1 .1 5

1 0 3 .2 5

1 0 0 .2 4

1 0 1 .7 5

1 0 4 .8 4

(0 .7 0 0 )

(0 .1 7 7 )

(0 .3 6 9 )

(0 .5 6 1 )

(0 .5 2 2 )

(0 .4 9 7 )

(0 .7 3 9 )

(0 .5 6 7 )

(0 .4 7 0 )

(0 .5 7 4 )

(0 .2 4 3 )

2 0 2 1 -0 5

1 0 4 .9 6

1 0 1 .8 6

1 0 1 .7 9

1 0 0 .0 8

1 0 1 .1 9

9 9 .7 2

1 0 0 .9 6

1 0 2 .4 7

1 0 0 .2 8

1 0 1 .9 0

1 0 3 .7 3

(0 .6 7 4 )

(0 .1 7 7 )

(0 .3 6 9 )

(0 .5 2 1 )

(0 .5 2 3 )

(0 .4 9 7 )

(0 .7 3 9 )

(0 .5 8 4 )

(0 .4 7 0 )

(0 .5 7 4 )

(0 .2 4 3 )

2 0 2 1 -0 6

1 0 4 .8 7

1 0 2 .3 3

1 0 1 .2 0

1 0 0 .7 2

1 0 1 .6 7

1 0 0 .3 0

1 0 0 .8 0

1 0 2 .8 0

9 9 .9 3

1 0 1 .9 1

1 0 3 .0 2

(0 .6 7 4 )

(0 .3 0 7 )

(0 .3 6 9 )

(0 .5 2 5 )

(0 .4 8 8 )

(0 .5 0 9 )

(0 .7 3 9 )

(0 .5 8 4 )

(0 .4 6 5 )

(0 .5 7 4 )

(0 .2 4 3 )

2 0 2 1 -0 7

1 0 2 .4 5

1 0 0 .3 1

1 0 0 .5 2

9 7 .7 2

9 8 .5 0

9 9 .8 0

1 0 0 .1 0

1 0 1 .4 9

9 9 .4 9

1 0 1 .8 7

1 0 2 .9 7

(0 .7 5 8 )

(0 .4 2 5 )

(0 .4 7 0 )

(0 .7 0 3 )

(0 .6 7 8 )

(0 .5 0 9 )

(0 .8 0 1 )

(0 .5 8 4 )

(0 .6 1 7 )

(0 .5 7 4 )

(0 .2 4 3 )

2 0 2 1 -0 8

1 0 2 .5 2

1 0 0 .7 6

1 0 0 .1 6

9 7 .5 7

9 8 .4 9

9 9 .2 9

1 0 0 .0 3

1 0 1 .8 8

9 9 .9 1

1 0 1 .4 1

1 0 2 .9 7

(0 .7 5 8 )

(0 .4 2 5 )

(0 .4 7 0 )

(0 .7 0 3 )

(0 .6 7 8 )

(0 .5 0 9 )

(0 .7 9 1 )

(0 .5 8 4 )

(0 .6 1 7 )

(0 .5 7 4 )

(0 .2 4 3 )

2 0 2 1 -0 9

1 0 2 .7 2

1 0 0 .5 9

1 0 0 .4 6

9 6 .8 6

9 7 .7 3

1 0 0 .1 4

9 9 .8 6

1 0 2 .6 2

9 9 .2 7

1 0 1 .7 6

1 0 2 .9 7

(0 .7 4 3 )

(0 .4 2 5 )

(0 .4 7 0 )

(0 .7 0 3 )

(0 .6 7 8 )

(0 .5 0 9 )

(0 .7 9 1 )

(0 .5 8 4 )

(0 .6 1 7 )

(0 .5 7 4 )

(0 .2 4 3 )

2 0 2 1 -1 0

1 0 2 .6 7

1 0 1 .0 8

9 9 .3 3

9 7 .2 6

9 7 .6 1

9 8 .9 8

9 9 .8 7

1 0 2 .2 6

9 9 .7 3

1 0 1 .2 0

1 0 2 .9 7

(0 .7 5 4 )

(0 .4 2 5 )

(0 .4 7 0 )

(0 .7 0 3 )

(0 .6 7 8 )

(0 .5 0 9 )

(0 .7 9 1 )

(0 .5 8 4 )

(0 .6 1 7 )

(0 .5 7 4 )

(0 .2 4 3 )

2 0 2 1 -1 1

1 0 2 .9 9

1 0 0 .5 8

1 0 0 .0 1

9 7 .5 5

9 7 .0 4

1 0 0 .3 6

1 0 0 .2 6

1 0 3 .0 2

9 8 .7 8

1 0 1 .9 1

1 0 3 .1 2

(0 .7 5 4 )

(0 .2 6 3 )

(0 .4 7 0 )

(0 .7 0 3 )

(0 .6 7 8 )

(0 .5 0 9 )

(0 .7 9 1 )

(0 .5 8 4 )

(0 .6 1 7 )

(0 .5 7 4 )

(0 .2 4 3 )

2 0 2 1 -1 2

1 0 2 .8 6

1 0 0 .4 6

9 9 .4 2

9 6 .9 1

1 0 1 .1 8

1 0 0 .5 5

1 0 0 .0 0

1 0 3 .3 7

9 9 .6 8

1 0 2 .4 4

1 0 4 .1 1

(0 .7 5 4 )

(0 .2 4 9 )

(0 .4 7 0 )

(0 .7 0 3 )

(0 .5 8 8 )

(0 .4 8 1 )

(0 .7 8 5 )

(0 .5 8 4 )

(0 .5 8 6 )

(0 .4 6 5 )

(0 .2 4 3 )

2 0 2 2 -0 1

1 0 3 .4 6

1 0 0 .7 9

1 0 0 .5 5

9 7 .9 9

1 0 0 .5 9

1 0 1 .3 3

1 0 1 .5 6

9 9 .8 1

1 0 5 .0 9

(0 .7 2 8 )

(0 .3 5 1 )

(0 .4 1 4 )

(0 .6 6 0 )

(0 .2 7 5 )

(0 .7 6 8 )

(0 .4 9 8 )

(0 .5 8 6 )

(0 .2 4 3 )

2 0 2 2 -0 2

1 0 3 .6 8

1 0 2 .0 2

1 0 0 .6 9

9 9 .0 4

1 0 0 .7 4

1 0 1 .8 7

1 0 1 .7 4

1 0 0 .1 3

1 0 3 .5 2

(0 .6 6 2 )

(0 .3 6 9 )

(0 .4 1 4 )

(0 .6 8 7 )

(0 .2 7 5 )

(0 .7 6 8 )

(0 .4 9 8 )

(0 .5 8 6 )

(0 .2 4 3 )

2 0 2 2 -0 3

1 0 3 .7 8

1 0 2 .4 1

1 0 0 .7 1

9 9 .1 2

1 0 0 .8 1

1 0 2 .3 9

1 0 2 .7 0

1 0 0 .9 8

1 0 3 .5 2

(0 .5 6 7 )

(0 .4 3 0 )

(0 .4 1 4 )

(0 .7 1 0 )

(0 .2 7 5 )

(0 .7 6 8 )

(0 .4 9 8 )

(0 .5 8 6 )

(0 .2 4 3 )

2 0 2 2 -0 4

1 0 3 .7 6

1 0 2 .2 0

1 0 1 .3 3

1 0 0 .1 8

1 0 1 .2 9

1 0 3 .1 6

1 0 2 .6 4

1 0 1 .1 8

(0 .4 5 8 )

(0 .4 3 0 )

(0 .4 1 4 )

(0 .7 1 0 )

(0 .2 7 5 )

(0 .7 6 8 )

(0 .4 9 8 )

(0 .5 8 6 )

2 0 2 2 -0 5

1 0 4 .5 2

1 0 3 .2 6

1 0 1 .6 2

1 0 0 .7 3

1 0 1 .6 1

1 0 3 .5 9

1 0 2 .5 5

1 0 1 .8 5

(0 .4 5 8 )

(0 .3 1 2 )

(0 .4 1 4 )

(0 .7 1 0 )

(0 .2 7 5 )

(0 .7 6 8 )

(0 .4 9 8 )

(0 .5 8 6 )

2 0 2 2 -0 6

1 0 4 .7 1

1 0 2 .7 9

1 0 1 .8 2

1 0 0 .4 7

1 0 2 .0 0

1 0 3 .9 5

1 0 1 .5 7

1 0 2 .5 9

(0 .4 5 8 )

(0 .4 3 0 )

(0 .4 1 4 )

(0 .7 0 2 )

(0 .2 7 5 )

(0 .7 6 8 )

(0 .3 9 5 )

(0 .5 8 6 )

2 0 2 2 -0 7

1 0 3 .6 6

1 0 3 .5 8

1 0 1 .6 9

1 0 0 .5 9

1 0 2 .1 5

1 0 3 .1 0

1 0 2 .2 7

1 0 2 .6 2

(0 .3 8 7 )

(0 .1 8 8 )

(0 .4 1 4 )

(0 .6 6 0 )

(0 .2 7 5 )

(0 .7 4 9 )

(0 .3 9 5 )

(0 .5 8 6 )

2 0 2 2 -0 8

1 0 5 .4 3

1 0 1 .0 5

1 0 2 .4 4

1 0 3 .6 6

1 0 4 .3 7

1 0 3 .0 8

1 0 3 .3 9

(0 .1 5 8 )

(0 .3 7 8 )

(0 .6 3 5 )

(0 .2 7 5 )

(0 .7 4 2 )

(0 .3 9 5 )

(0 .5 8 6 )

2 0 2 2 -0 9

1 0 8 .5 6

1 0 2 .4 3

1 0 3 .0 1

1 0 3 .2 5

1 0 5 .2 5

1 0 3 .7 7

1 0 3 .1 7

(0 .1 5 8 )

(0 .3 7 8 )

(0 .6 3 5 )

(0 .2 7 5 )

(0 .7 4 2 )

(0 .3 9 5 )

(0 .5 8 6 )

2 0 2 2 -1 0

1 0 4 .0 2

1 0 1 .0 7

1 0 2 .3 8

1 0 5 .3 6

1 0 5 .1 2

1 0 4 .0 5

1 0 5 .2 3

(0 .1 5 8 )

(0 .3 7 8 )

(0 .6 3 5 )

(0 .2 7 5 )

(0 .7 4 2 )

(0 .3 9 5 )

(0 .5 8 6 )

2 0 2 2 -1 1

1 0 9 .4 1

1 0 3 .4 5

1 0 4 .5 6

1 0 6 .1 3

1 0 6 .4 3

1 0 4 .4 7

1 0 5 .9 0

(0 .1 5 8 )

(0 .3 7 8 )

(0 .6 3 5 )

(0 .2 7 5 )

(0 .7 4 2 )

(0 .3 9 5 )

(0 .5 8 6 )

2 0 2 2 -1 2

1 0 9 .3 4

1 0 2 .0 5

1 0 3 .3 3

1 0 5 .2 8

1 0 6 .3 3

1 0 4 .1 8

1 0 5 .4 4

(0 .1 5 8 )

(0 .3 7 8 )

(0 .6 3 5 )

(0 .2 7 5 )

(0 .7 4 2 )

(0 .3 9 5 )

(0 .5 8 6 )

2 0 2 3 -0 1

1 0 8 .2 3

1 0 3 .3 3

1 0 4 .4 7

1 0 6 .6 7

1 0 7 .4 7

1 0 6 .1 9

1 0 6 .5 2

(0 .1 5 8 )

(0 .3 7 2 )

(0 .6 1 7 )

(0 .1 8 7 )

(0 .7 2 8 )

(0 .3 9 5 )

(0 .5 6 0 )

2 0 2 3 -0 2

1 0 8 .6 6

1 0 3 .3 7

1 0 4 .8 2

1 0 4 .6 1

1 0 7 .7 7

1 0 7 .0 4

1 0 4 .4 5

(0 .1 5 8 )

(0 .3 7 2 )

(0 .6 1 7 )

(0 .1 8 7 )

(0 .7 2 8 )

(0 .3 9 5 )

(0 .5 6 0 )

Page 11 of 15

T a b

le 4 :

C o n ti

n u

e fr

o m

T a b

le 3 .

M o n th

P ie m o n te

P u g li a

S a rd

e g n a

S ic il ia

T o sc a n a

U m b ri a

V a ll e D ’A

o st a

V e n e to

2 0 2 0 -1 1

9 4 .8 3

(0 .1 0 6 )

2 0 2 0 -1 2

9 0 .1 9

1 0 4 .0 2

1 0 2 .8 0

1 0 0 .3 9

9 8 .0 9

1 0 7 .5 5

(0 .1 2 0 )

(0 .4 0 6 )

(0 .1 6 4 )

(0 .5 2 3 )

(0 .2 8 3 )

(0 .1 2 2 )

2 0 2 1 -0 1

9 1 .3 6

1 0 3 .4 2

1 0 4 .5 8

1 0 0 .0 6

1 0 0 .7 9

(0 .1 2 0 )

(0 .4 2 3 )

(0 .1 6 4 )

(0 .5 2 3 )

(0 .2 8 3 )

(0 .1 2 2 )

2 0 2 1 -0 2

1 0 5 .3 3

9 4 .8 3

1 0 2 .1 4

1 0 2 .7 3

1 0 0 .1 1

9 9 .4 7

1 1 0 .7 0

9 8 .1 3

(0 .4 0 3 )

(0 .1 2 0 )

(0 .5 4 2 )

(0 .3 9 7 )

(0 .5 6 3 )

(0 .2 8 3 )

(0 .6 5 3 )

(0 .4 9 9 )

2 0 2 1 -0 3

1 0 5 .4 7

9 5 .4 5

1 0 2 .1 2

1 0 3 .0 7

1 0 0 .5 2

9 8 .4 7

1 1 0 .7 6

9 8 .7 0

(0 .4 0 3 )

(0 .1 2 0 )

(0 .5 4 2 )

(0 .3 9 7 )

(0 .5 6 3 )

(0 .2 8 3 )

(0 .6 5 3 )

(0 .4 9 9 )

2 0 2 1 -0 4

1 0 5 .5 8

9 6 .3 9

1 0 2 .7 5

1 0 2 .6 5

1 0 0 .5 7

9 9 .8 7

1 1 0 .2 0

9 8 .8 3

(0 .4 0 3 )

(0 .1 2 0 )

(0 .5 4 2 )

(0 .3 9 7 )

(0 .5 6 3 )

(0 .2 8 3 )

(0 .6 5 3 )

(0 .4 9 9 )

2 0 2 1 -0 5

1 0 4 .9 2

9 6 .4 5

1 0 2 .6 9

1 0 1 .7 6

1 0 0 .4 7

9 9 .3 8

1 0 9 .8 0

9 8 .6 5

(0 .4 0 3 )

(0 .1 2 0 )

(0 .5 4 2 )

(0 .3 9 7 )

(0 .5 6 3 )

(0 .2 8 3 )

(0 .6 5 3 )

(0 .4 9 9 )

2 0 2 1 -0 6

1 0 5 .2 1

9 5 .2 9

1 0 2 .5 6

1 0 2 .2 7

1 0 0 .4 3

9 9 .1 0

1 1 1 .1 8

9 9 .6 6

(0 .4 0 3 )

(0 .1 0 6 )

(0 .5 4 2 )

(0 .3 6 1 )

(0 .5 6 3 )

(0 .2 8 3 )

(0 .6 5 3 )

(0 .4 8 7 )

2 0 2 1 -0 7

9 9 .7 9

9 4 .6 2

1 0 2 .1 7

1 0 2 .3 8

1 0 0 .0 2

9 8 .5 2

1 0 9 .8 0

9 4 .1 9

(0 .5 5 1 )

(0 .3 9 8 )

(0 .5 7 2 )

(0 .4 6 0 )

(0 .5 9 1 )

(0 .8 5 1 )

(0 .6 5 3 )

(0 .5 6 5 )

2 0 2 1 -0 8

9 8 .3 1

9 4 .3 3

1 0 2 .2 8

1 0 2 .8 6

9 9 .7 2

9 8 .4 2

1 1 0 .3 1

9 7 .1 0

(0 .5 5 4 )

(0 .3 6 2 )

(0 .5 7 2 )

(0 .4 6 0 )

(0 .5 9 1 )

(0 .8 5 1 )

(0 .6 5 3 )

(0 .5 4 1 )

2 0 2 1 -0 9

1 0 5 .7 6

9 4 .2 6

1 0 1 .3 9

1 0 2 .2 3

9 9 .8 8

9 8 .6 7

1 1 0 .4 3

9 6 .8 2

(0 .4 0 3 )

(0 .3 6 2 )

(0 .5 6 1 )

(0 .4 6 0 )

(0 .5 9 1 )

(0 .8 5 1 )

(0 .6 5 3 )

(0 .5 4 1 )

2 0 2 1 -1 0

1 0 4 .9 6

9 3 .7 2

1 0 2 .3 1

1 0 2 .3 6

1 0 0 .0 0

9 8 .9 6

1 0 9 .4 6

9 6 .4 5

(0 .4 0 3 )

(0 .3 6 2 )

(0 .5 7 2 )

(0 .4 6 0 )

(0 .5 9 1 )

(0 .8 5 1 )

(0 .6 5 3 )

(0 .5 4 1 )

2 0 2 1 -1 1

1 0 5 .2 4

9 5 .0 3

1 0 2 .7 2

1 0 3 .2 3

1 0 0 .0 4

9 9 .6 9

1 1 0 .2 6

9 7 .2 1

(0 .4 0 3 )

(0 .3 8 5 )

(0 .5 7 2 )

(0 .4 6 0 )

(0 .5 9 1 )

(0 .8 5 1 )

(0 .6 5 3 )

(0 .5 4 1 )

2 0 2 1 -1 2

1 0 5 .1 9

9 5 .5 3

1 0 3 .2 2

1 0 2 .2 3

9 9 .9 6

9 9 .7 7

1 0 7 .2 1

9 9 .9 1

(0 .3 6 0 )

(0 .3 8 5 )

(0 .5 3 8 )

(0 .3 9 2 )

(0 .5 5 1 )

(0 .8 5 1 )

(0 .2 2 9 )

(0 .4 8 0 )

2 0 2 2 -0 1

9 5 .5 5

1 0 4 .3 1

1 0 4 .9 7

1 0 1 .1 9

1 0 1 .7 7

(0 .2 9 2 )

(0 .4 2 3 )

(0 .3 9 9 )

(0 .5 4 4 )

(0 .8 3 9 )

2 0 2 2 -0 2

9 6 .5 3

1 0 3 .9 6

1 0 6 .2 4

1 0 0 .5 3

1 0 1 .7 5

(0 .2 6 9 )

(0 .4 2 3 )

(0 .4 2 8 )

(0 .5 2 3 )

(0 .8 3 9 )

2 0 2 2 -0 3

9 7 .4 3

1 0 5 .4 0

1 0 7 .6 5

1 0 2 .0 9

1 0 2 .0 6

(0 .2 6 9 )

(0 .4 2 3 )

(0 .4 2 8 )

(0 .5 2 3 )

(0 .8 3 9 )

2 0 2 2 -0 4

9 7 .9 0

1 0 6 .8 0

1 0 8 .8 2

1 0 3 .9 4

1 0 2 .3 0

(0 .2 6 9 )

(0 .4 2 3 )

(0 .4 2 8 )

(0 .5 2 3 )

(0 .8 3 9 )

2 0 2 2 -0 5

9 8 .4 2

1 0 7 .5 9

1 0 9 .2 9

1 0 3 .1 1

1 0 1 .9 2

(0 .2 6 9 )

(0 .4 2 3 )

(0 .4 2 8 )

(0 .4 0 4 )

(0 .8 3 9 )

2 0 2 2 -0 6

9 8 .3 4

1 0 6 .7 3

1 0 9 .1 9

1 0 0 .6 8

1 0 1 .5 7

(0 .2 6 9 )

(0 .4 2 3 )

(0 .4 2 8 )

(0 .1 3 8 )

(0 .8 3 9 )

2 0 2 2 -0 7

1 0 1 .2 2

1 0 9 .4 0

1 0 1 .6 2

(0 .2 5 4 )

(0 .4 2 8 )

(0 .8 3 9 )

2 0 2 2 -0 8

1 0 9 .2 7

1 0 3 .5 4

(0 .4 2 8 )

(0 .8 3 9 )

2 0 2 2 -0 9

1 0 9 .3 5

1 0 5 .2 3

(0 .4 2 8 )

(0 .8 3 9 )

2 0 2 2 -1 0

1 0 9 .9 8

1 0 4 .7 9

(0 .4 2 8 )

(0 .8 3 9 )

2 0 2 2 -1 1

1 1 0 .8 0

1 0 6 .3 4

(0 .4 2 8 )

(0 .8 3 9 )

2 0 2 2 -1 2

1 1 0 .0 7

1 0 5 .6 8

(0 .4 2 8 )

(0 .8 3 9 )

2 0 2 3 -0 1

1 1 1 .6 0

1 0 7 .2 0

(0 .4 2 8 )

(0 .8 3 9 )

2 0 2 3 -0 2

1 1 3 .0 2

(0 .4 2 8 )

Page 12 of 15

Figure 7: Spatio-temporal price index levels for Coffee (ECOICOP 01.2.1.1).

20 20

-1 1

20 20

-1 2

20 21

-0 1

20 21

-0 2

20 21

-0 3

20 21

-0 4

20 21

-0 5

20 21

-0 6

20 21

-0 7

20 21

-0 8

20 21

-0 9

20 21

-1 0

20 21

-1 1

20 21

-1 2

20 22

-0 1

20 22

-0 2

20 22

-0 3

20 22

-0 4

20 22

-0 5

20 22

-0 6

20 22

-0 7

20 22

-0 8

20 22

-0 9

20 22

-1 0

20 22

-1 1

20 22

-1 2

20 23

-0 1

20 23

-0 2

Date

90

95

100

105

110

Pr ice

In de x

Regions Abruzzo Basilicata Calabria

Campania Emilia-Romagna Friuli-Venezia Giulia

Lazio Liguria Lombardia

Marche Molise Piemonte

Puglia Sardegna Sicilia

Toscana Umbria

Valle D'Aosta Veneto

Table 5: Correlation between structural breaks in coverage and spatio-temporal price index level time series.

Region Correlation p-value Abruzzo 0.359 (0.061) Basilicata 0.399 (0.101) Calabria 0.142 (0.481) Campania 0.735 (0.000) Emilia-Romagna 0.010 (0.961) Friuli-Venezia Giulia 1.000 (0.000) Lazio -0.078 (0.695) Liguria 0.410 (0.034) Lombardia -0.048 (0.813) Marche 0.815 (0.000) Molise 0.993 (0.000) Piemonte 0.994 (0.000) Puglia 0.300 (0.186) Sardegna -0.047 (0.848) Sicilia -0.033 (0.868) Toscana 0.954 (0.000) Umbria 0.554 (0.003) Valle d’Aosta 0.999 (0.000) Veneto 0.919 (0.000)

We can see that in 10 cases out of 19 we are able to identify a significant and strong positive correlation between the two time series. Therefore, we can maintain that structural breaks in coverage may effectively impact the stability of the measured price index level.

Page 13 of 15

5 Conclusions and future research

We believe coverage information is a relevant metric for price statistics. Modelling accurately where price collection takes place considering consumer purchasing habits and travel distance. Embedding this information in CPIs can provide tremendous insights at several levels in the price statistics compilation and utilization process.

During selection and sampling of outlets for price collection it would be important to have an accurate view of geographical and population coverage in order to make sure that no dark spot is left systematically in price surveys and there is continuity and consistent overlap over time for the covered area. As demonstrated, substantial changes in coverage between different period may effectively impact the stability of price index measurement.

When using price statistics this coverage view would be equally important. Local dynamics in economic and social measures are object of a growing number of studies, and granular coverage information could help to better integrate price statistics in this stream of research.

One of the main points for future improvement is the ability to correctly identify and match product across multiple retailers when GTINs are not available. In our case, less than half of the unique products had a GTIN associated. It is quite likely that amongst the others there will be matching products, but the large number of unique products - even in the very limited perimeter we selected for this exercise - combined with the sheer similarity in naming across different products makes manual vetting a complex and time consuming task on the one side, and tricky for automatic matching algorithms on the other. In this area we plan to explore the use of Large Language Models, as those tools have extended language capabilities and the potential to leverage context information and prompting specific for the task.

Finally, other services may be explored for obtaining updated travel time calculations in the future, such as Google Distance Matrix API or TravelTime API. Furthermore, we foresee additional application for a measure of information decay over space or travel time as presented in this work for other fields beyond price statistics.

References

Aizcorbe, A., & Aten, B. (2004). An approach to pooled time and space comparisons.. Aizcorbe, A., Corrado, C., & Doms, M. (2000). Constructing price and quantity indexes for high tech-

nology goods. Industrial Output Section, Division of Research and Statistics, Board of Governors of the Federal Reserve System, July , 19 .

Aten, B. (1996). Evidence of spatial autocorreilation in international prices. Review of Income and Wealth, 42 (2), 149–163.

Benedetti, I., Laureti, T., Palumbo, L., & Rose, B. M. (2022). Computation of high-frequency sub- national spatial consumer price indexes using web scraping techniques. Economies, 10 (4), 95.

Berry, F., Graf, B., Stanger, M. M., & Ylä-Jarkko, M. (2019). Price statistics compilation in 196 economies: The relevance for policy analysis. International Monetary Fund Working Papers, 2019 . DOI: https://doi.org/10.5089/9781513508313.001

Biggeri, L., Laureti, T., & Polidoro, F. (2017). Computing sub-national PPPs with CPI data: an empirical analysis on Italian data using country product dummy models. Social Indicators Research, 131 (1), 93–121.

Brunetti, A., Fatello, S., Polidoro, F., & Simone, A. (2018). Improvements in Italian CPI/HICP deriving from the use of scanner data. In 50th scientific meeting of the italian statistical soci- ety. Retrieved from http://meetings3.sis-statistica.org/index.php/sis2018/50th/paper/

viewFile/1484/32

Corrado, C., & Ukhaneva, O. (2016). Hedonic prices for fixed broadband services: estimation across oecd countries. OECD Science, Technology and Industry Working Papers(2016/07). DOI: https://doi.org/https://doi.org/10.1787/5jlpl4sgc9hj-en”

De Haan, J., & Krsinich, F. (2014). Scanner data and the treatment of quality change in nonrevisable price indexes. Journal of Business & Economic Statistics, 32 (3), 341–358.

Diewert, W. (2021). Elementary indexes. Consumer Price Index Theory . Eurostat. (2020). Practical Guidelines on Web Scraping for the HICP. Retrieved 2022-04-20, from

https://ec.europa.eu/eurostat/documents/272892/12032198/Guidelines-web-scraping

-HICP-11-2020.pdf

Guerreiro, V., Baer, M. A., & Silungwe, A. (2022). The availability, methodological soundness, and scope of consumer price statistics in 2020. International Monetary Fund.

Hawkes, W. J., & Piotrowski, F. W. (2003). Using scanner data to improve the quality of measurement in the consumer price index. In Scanner data and price indexes (pp. 17–38). University of Chicago Press.

Hill, R. J., & Timmer, M. P. (2006). Standard errors as weights in multilateral price indexes. Journal of Business & Economic Statistics, 24 (3), 366–377.

Page 14 of 15

International Monetary Fund. (2020). Consumer price index manual. London, England: International Monetary Fund.

Istat. (2019). Matrici di contiguità, distanza e pendolarismo. Retrieved 2022-04-20, from https://

www.istat.it/it/archivio/157423

Kerr, J., Lawrence, F., Sallis, J. F., Saelens, B., Glanz, K., & Chapman, J. (2012). Predictors of trips to food destinations. International Journal of Behavioral Nutrition and Physical Activity , 9 (58). DOI: 10.1186/1479-5868-9-58

Kwiatkowski, D., Phillips, P. C. B., Schmidt, P., & Shin, Y. (1992). Testing the null hypothesis of stationarity against the alternative of a unit root : How sure are we that economic time series have a unit root? Journal of Econometrics, 54 (1-3), 159-178.

Laureti, T., & Rao, D. P. (2018). Measuring spatial price level differences within a country: Current status and future developments. Studies of Applied Economics, 36 (1), 119–148.

Montero, J.-M., Laureti, T., Mı́nguez, R., & Fernández-Avilés, G. (2020). A stochastic model with penalized coefficients for spatial price comparisons: An application to regional price indexes in Italy. Review of Income and Wealth, 66 (3), 512–533.

Rao, D. S. P. (2001). Weighted EKS and generalised CPD methods for aggregation at basic heading level and above basic heading level. In Joint World Bank-OECD seminar on purchasing power parities, recent advances in methods and applications. Washington DC.

Smith, P. A. (2021). Estimating sampling errors in consumer price indices. International Statistical Review , 89 (3), 481–504.

Summers, R. (1973). International price comparisons based upon incomplete data. Review of Income and Wealth, 19 (1), 1–16.

Zadeh, L. A. (1977). Fuzzy sets and their application to pattern classification and clustering analysis. In Classification and clustering (pp. 251–299). Elsevier.

Zhao, K., Wulder, M. A., Hu, T., Bright, R., Wu, Q., Qin, H., . . . Brown, M. (2019). Detecting change- point, trend, and seasonality in satellite time series data to track abrupt changes and nonlinear dynamics: A Bayesian ensemble algorithm. Remote Sensing of Environment , 232 , 111181.

Zimmermann, H.-J. (2011). Fuzzy set theory—and its applications. Springer Science & Business Media.

Page 15 of 15

INTEGRATING SURVEY DATA AND BIG DATA. RESULTS BASED ON ISTAT’S WORK ABOUT GENDER STEREOTYPES.

Istat | DCDC

Geneva, 10-12 May 2023

Meeting of the UNECE Group of

Experts on Gender Statistics

FRANCESCO GOSETTI MARIA GIUSEPPINA MURATORE LUCILLA SCARNICCHIA

o Why studying gender-based stereotypes

o Survey on gender role stereotypes and the social image of violence

o Survey main results

o Experimenting Big Data analysis

o Conclusions

Contents

INTEGRATING SURVEY DATA AND BIG DATA. RESULTS BASED ON ISTAT’S WORK ABOUT GENDER STEREOTYPES2

➢ Gender-based stereotypes limit the access of women and girls to

education, work, career: prevent their full advancement

➢ Istanbul Convention focuses on stereotyping as a major cause of

Violence Against Women and Girls (VAW)

Understand the extend of stereotypes that

corresponds to a specific society, country

Monitor the effectiveness of education policies

Assess the tolerance of violence, to be correlated with

the results of prevalence survey

Why studying gender-based stereotypes

INTEGRATING SURVEY DATA AND BIG DATA. RESULTS BASED ON ISTAT’S WORK ABOUT GENDER STEREOTYPES3

Istanbul Convention

Article12

invites Parties “to promote

changes in the social and cultural

patterns of behaviour of women

and men with a view to

eradicating prejudices, customs,

traditions and all other practices

which are based on the idea of

the inferiority of women or on

stereotyped roles for women

Article 14

focuses on the role of education to

eliminate stereotypes

THEMATIC AREAS AND RESEARCH PURPOSES IN THE MODULE

The module on gender role stereotypes and the social image of violence

INTEGRATING SURVEY DATA AND BIG DATA. RESULTS BASED ON ISTAT’S WORK ABOUT GENDER STEREOTYPES4

Questions Information gathered

GENDER ROLES STEREOTYPES level of gender stereotyping among

the population

INTIMATE PARTNER VIOLENCE

Acceptability

level of tolerance of IPV

INTIMATE PARTNER VIOLENCE

Perceived prevalence and its causes

population's awareness, that might

affect attitudes

REACTIONS TO VIOLENCE familiarity with some services,

awareness of the complexity of the

pathway out of violence

SEXUAL VIOLENCE STEREOTYPES how is pervasive the culture of

violence

2013

Italy ratified Istanbul

Convention

→ National plan against VAW

2017

Agreement ISTAT -

National Department

Equal Opportunities

→ Integrated system of

information on VAW

2018

The ad hoc module

PREJUDICES ABOUT SEXUAL VIOLENCE

Year 2018. Percentage values.

THE MOST COMMON STEREOTYPES ABOUT GENDER ROLES

AMONG WOMEN AND MEN. Year 2018. Percentage values.

Survey: main results

INTEGRATING SURVEY DATA AND BIG DATA. RESULTS BASED ON ISTAT’S WORK ABOUT GENDER STEREOTYPES5

For the man, more than for the woman, it

is very important to be successful at work

Men are less suited to do housework

It is up to the man to provide for the

family’s financial needs

When jobs are scarce, employers should

give priority to men over women

It’s up to the man to take the most

important decisions about the family

32.5

27.9

16.1

8.8

31.5

Women who don’t want to have a sexual

intercourse are able to avoid it 39.3

Women can provoke sexual violence by how

they dress 23.9

If a woman suffers sexual violence when she

is affected by alcohol or drugs, she is at least

partially responsible

15.1

AGREE WITH AT LEAST ONE STEREOTYPE / BEHAVIOUR . Year 2018. Percentage values.

BEHAVIOURS ACCECTABLE AT LEAST UNDER CERTAIN

CIRCUMSTANCES. Year 2018. Percentage values.

Survey: main results

INTEGRATING SURVEY DATA AND BIG DATA. RESULTS BASED ON ISTAT’S WORK ABOUT GENDER STEREOTYPES6

A young man slaps his girlfriend because she

flirted with another man 7.4

In a couple’s relationship, it is normal that a

slap might occasionally occur 6.2

A man habitually control his wife’s/partner’s

mobile phone and activities on social media

(Facebook, chats, etc.).

17.7

control is acceptable

28.8% among aged 18-29

GENDER ROLES STEREOTYPE

58.8

SEXUAL VIOLENCE STEREOTYPE

54.6

INTIMATE PARTNER VIOLENCE

ACCEPTABILITY 25.4

AGE and

EDUCATION matter

POSSIBLE CAUSES OF INTIMATE PARTNER VIOLENCE. Year 2018. Percentage values.

INTEGRATING SURVEY DATA AND BIG DATA. RESULTS BASED ON ISTAT’S WORK ABOUT GENDER STEREOTYPES

Survey: main results

7

Men Women

70.4 84.9 Considering women to be property

74.0 77.0 Abuse of drugs or alcohol

68.5 81.3 Need to feel stronger than one’s partner/wife

66.4 74.6 Difficulty in managing anger

60.1 67.1 Having negative experiences of family violence as a child

55,2 69.9 Not standing women’s empowerment

33.5 34.0 Religious reasons

Tell her to file a report 64.5

Advise her to leave her

husband/partner 33.2

Direct her to anti-violence centres

helping women 20.4

Direct her to other services or

professionals 18.2

Advise her to try talking with her

husband/partner 3.8

Not know what to do/advise 2.6

Tell her to call 1522 helpline 2.0

Not offer advice because I don’t want

To interfere in family issues 1.1

IF YOU KNEW A WOMAN WHO SUFFERED VIOLENCE

BY HER PARTNER, YOU WOULD … - Year 2018 (%)

Survey: main results

INTEGRATING SURVEY DATA AND BIG DATA. RESULTS BASED ON ISTAT’S WORK ABOUT GENDER STEREOTYPES8

➢ NO STEREOTYPES

2 clusters, 62.0% of population

➢ STRONG STEREOTYPES

2 clusters, 36.3% of population

➢ ALOOF

1 cluster, 1.8% of population

strict link between prejudices

and the acceptability of

violence

ALOOF

INTEGRATING SURVEY DATA AND BIG DATA. RESULTS BASED ON ISTAT’S WORK ABOUT GENDER STEREOTYPES

Experimenting Big Data analysis

9

Experimental study

How Gender Based Violence and Gender Stereotypes are

represented and perceived in social media: Twitter,

Facebook, Instagram, posts on news websites

Analysis of social media contents:

➢ Sentiment analysis

➢ Emotion detection

…. Attitudes expressing awareness?

aggression?

… Body-shaming

…

➢ WEAKNESSES

No info about users: sex, age, education, geographical area

No profile of users

➢ STRENGHTS

Analysis from a new perspective

involve some groups excluded by the survey (under18)

can be used to know new expressions of stereotypes

intersectionality of discrimination grounds: studying language and stereotyped

opinions used also or additionally against other vulnerable groups

Experimenting Big Data analysis

INTEGRATING SURVEY DATA AND BIG DATA. RESULTS BASED ON ISTAT’S WORK ABOUT GENDER STEREOTYPES10

➢ Still long way to eradicate stereotypes

➢ Young and more educated are more open

➢ Population survey is a powerful tool

➢ Social media contents to be exploited

Conclusions

INTEGRATING SURVEY DATA AND BIG DATA. RESULTS BASED ON ISTAT’S WORK ABOUT GENDER STEREOTYPES11

• Regular basis – 2023 edition ongoing

• Developped new questions

• Planned survey for students aged 11-19

• Invest in prevention, education

• Monitor attitudes

• Orient policies

• New forms gender stereotypes over time

• Intersectionality of discrimination grounds

• In the survey, questions about the use of

social media

Thank you

https://www.istat.it/en/violence-against-women

[email protected]

*Prepared by Eugenia De Rosa (section II, section V, section VI), Valeria de Martino (Introduction, section III),

Francesca Scambia (section IV).

NOTE: The designations employed in this document do not imply the expression of any opinion whatsoever on the part

of the Secretariat of the United Nations concerning the legal status of any country, territory, city or area or of its

authorities, or concerning the delimitation of its frontiers or boundaries.

Economic Commission for Europe

Conference of European Statisticians

Group of Experts on Gender Statistics Geneva, Switzerland, 10-12 May 2023

Item D of the provisional agenda

Measuring sex and gender

Gender Identity Representation in Data Collection: New Approaches from Italy

Note by National Statistical Institute of Italy - Istat*

Abstract

Istat (National Statistical Institute of Italy), in collaboration with Unar (National Antidiscrimination Office), is

carrying out a project on "Labour discrimination against LGBT+ people and diversity policies implemented in

enterprises" which started in 2018. It is characterized by a mixed method (quantitative-qualitative) and

includes the direct collection of information from different target groups of LGBT+ people.

The project includes three CAWI surveys based on respondents’ self-identification as LGBT+ people, and

carried out by a web self-completed questionnaire:

a) in 2020-2021 a total survey of resident individuals (over 21,000) who, as of 1 January 2020, were or had

been in Civil Union. The main results were published in 2022 (Istat, 2022). In Italy the union between persons

of the same sex is regulated by a legal institution named Civil Union (since July 2016) which is different from

marriage which is for different-sex couples alone;

b) in 2022 a survey on LGB people who have never been in Civil Union (completed in May 2022). Istat

tested for the first time the snowball technique RDS (Respondent Driving Sample), which afterwards opened to a

convenience sample;

c) a survey on trans and non-binary persons which is currently in progress.

Specific questions on SOGIESC (sexual orientation, gender identity, gender expression and sex characteristics)

indicators were discussed, tested and analysed.

The aim of this article is to illustrate the Italian experience in surveying gender identity and gradually introducing

other

SOGIESC indicators in official statistics. It in depth illustrates indicators of sex and gender identity and tested with

reference to the different target groups.

Finally, it identifies the main challenges and offers some suggestions for improving- gender representation in data

collection, and developing indicators of gender identity to be introduced in official surveys targeted to the whole

population

Working paper 11 Rev1

Distr.: General

22 May 2023

English

Working paper 11 Rev1

2

I. Introduction

1. In recent years, a broad debate and comparison has developed in the field of official

statistics, both at international and European level, on the so-called SOGIESC indicators

(Sexual Orientation, Gender Identity and Expression, and Sex Characteristics Indicators).

The common intent of the National Statistical Institutes, international and research agencies

(e.g., United Nations Economic Commission for Europe-UNECE, European Union Agency

for Fundamental Rights- FRA, Praia Group on Governance Statistics-Praia Group) and

equality bodies is to produce more inclusive statistics and comparable data to monitor

inequalities and discrimination based on SOGIESC characteristics and those based on the

intersection with other relevant aspects/characteristics to define the identity and position of

individuals and groups in society (e.g., citizenship, age, social class…).

2. Academic and grey literature about LGBT+ issues, especially about trans and non-

conforming gender identities, is very rich and developed in some contexts, and represents a

useful point of reference.

3. In 2011, the National Statistical Institute of Italy (Istat) addressed for the first time issues

related to diversity in terms of sexual orientation and gender identity when, in “Survey on

Discriminations by Gender, Sexual Orientation and Ethnic Origin” collected information on

opinions and attitudes toward gender roles, homosexuality and immigration; also estimating

the number of discriminations’ victims at school and/or at work (Istat, 2013). The survey

included questions on sexual orientation which allowed to provide a first estimate of the

homosexual and bisexual population in Italy.

4. In 2018, Istat dealt with these issues again in the framework of a collaboration agreement

signed with Unar (National Antidiscrimination Office) to fill an information gap on LGBT+

populations. This agreement gave rise to the project, currently underway, on "Labour

discrimination against LGBT+ people and the diversity policies implemented in enterprises".

Making use of the available resources in terms of budget and timing, the Istat-Unar project

combined surveys both with standard and non-standard sampling techniques.

5. As part of this project, which involved carrying out surveys aimed at LGBT+ population

groups, questions on gender identity were tested and introduced for the first time, as well as

on sexual orientation, gender expression and the sexual characteristics. Definitions and

indicators provided in the questionnaires were discussed and shared with experts, academics,

associations and LGBT+ people.

6. The Istat-Unar project on labour discrimination constitutes therefore an important step

towards inclusive statistics that give a plural representation of gender and sexual identities.

At the same time they provide a basis from which testing various SOGIESC indicators also

in surveys addressed to the whole population.

7. This paper particularly focuses on indicators relating to sex and gender identity and is

structured as follows - Section II describes the methodological framework of the Istat-Unar

project and the ways in which SOGIESC questions were introduced in the various surveys

within the project; section III illustrates the ways in which indicators of sex and gender, were

developed and discussed; section IV focuses on SOGIESC indicators used in the Survey on

Labour Discrimination addressed to LGB people not in a Civil Union (2022); while section

V focuses on indicators designed for the survey aimed at trans and non-binary people. The

concluding section outlines considerations derived from the current experience together with

some and more general recommendations.

Working paper 11 Rev1

3

II. Istat-Unar project on “Labour Discrimination against LGBT+ people and diversity policies in enterprises” and SOGIESC indicators

8. The Istat-Unar project is characterized by a mixed quantitative-qualitative and multiple

perspective approach (LGBT+ people, employers, stakeholders) as well as the interaction of

experts, academics and LGBT+ associations (De Rosa, Inglese, 2018; Istat, 2020; De Rosa

et. al 2023). It includes the direct collection of information from LGBT+ people (first

macro-area) and from employers, particularly enterprises, and the main stakeholders (second

macro-area).

9. The project adopts a participatory approach. It includes the creation of various Working

Groups composed by associations and bodies of the «Permanent Consultation Table for the

promotion of LGBT rights and the protection of LGBT persons» established in 2018 to the

Presidency of the Council of Ministers. Also non-members of LGBT+ associations, experts

and academics have been involved in the design of the questionnaire and indicators.

10. The project was funded by EU funds.

11. This paper focuses on the first macro area of the project and in particular on gender identity

representation in data collection, on surveying sex and gender identity.

12. The first macro-area of the project aimed at providing an insight on the condition of labour

discrimination against LGBT+ people in Italy by means of different surveys collecting

information from various target groups of LGBT+ people.

13. Statistical representative surveys on the LGBT+ population is strongly biased by the lack of

knowledge of these populations such as to have theoretical frames for the construction of

probabilistic samples. Theoretical and statistical representativeness of the various groups

included in the acronym is difficult to achieve due to their relatively low incidence in the

population.

14. Three CAWI surveys based on a web self-completed questionnaire were planned. The main

investigated phenomena are: coming out, experience of discrimination while looking for a

job or while working and discrimination in other areas of social life (e.g., at school),

microaggressions, aggressions, hate speech. The first two surveys mainly focused on aspects

related to sexual orientation; the third focuses on gender identity issues.

15. Self-identification of respondent’s as LGBT+ was a key principle adopted. SOGIESC

indicators were gradually tested and introduced into the three surveys.

16. The first survey on labour discrimination was carried out in 2020-2021. It was a total survey

of the resident individuals (over 21,000) who, as of 1 January 2020, were or had been in

Civil Union (same-sex couples). Since July 2016 in Italy the union of same-sex persons over

18 has been regulated by a special institution named Civil Union. It differs from marriage

which is only for different-sex couples.

17. The main results of the “Survey on Labour Discrimination against LGBT+ people (in Civil

Union or formerly in union)” were published in 2022 (Istat, 2022). Same-sex couples in civil

union represent a specific group of LGBT+ population living in Italy. They indeed

evidenced some specific features for being in their majority homosexuals and bisexuals,

men, older people (43.6% of homosexuals and bisexuals are >50), opened and well

integrated in the labour market.

Working paper 11 Rev1

4

18. The survey included questions about sex and sexual orientation. Gender identity was not

disclosed. 95.2% people in Civil Union or formerly in union who live in Italy declared a

homosexual or bisexual orientation. Main results were analysed by the following profiles:

gay, lesbians, bisexual males and bisexual females.

19. A second survey, complementary to the first one, was addressed to LGB people who have

never been in Civil Union. The “Survey on Labour Discrimination against LGBT+ people

(not in Civil Union)” was carried out in 2022 (January-May) and included questions to

detect sex, sexual orientation and gender identity. The target population was limited to

homosexual and bisexual cisgender and non-binary persons.

20. With this survey Istat implemented for the first time the snowball technique Respondent

Driving Sample (RDS) in its web version (WebRDS). This technique is based on social

relationships in order to reach out the so-called hidden and hard-to-reach population (De

Rosa et. al 2020).

21. RDS consists of combining the snowball technique, in which the sample is constructed using

the names provided by initial recruiters, with a mathematical model that formalizes, under

certain conditions, the recruitment process as a Markov chain, or as a probabilistic process.

The data collected during the sampling process are used to make inferences about the

structure of the social network and obtain from this unbiased estimate on the population of

interest. It requires respondents to play an active role in recruiting new respondents who

belong to their network of relationships.

22. About fifty LGBT+ association agreed in facilitating the survey and, after signing an

agreement with Istat on privacy protection, were formed to identify the “seeds” of the

network chain. After an established time passed from the beginning of the survey, even-

though some questionnaires were filled-in, however there was evidence that the RDS

snowball technique was not working properly due to different reasons. In order to go on with

the work, the option of a convenience sample was considered since it could anyway provide

interesting and qualitative information on the target population of homosexual and bisexual

persons.

23. The final results of the “Survey on Labour Discrimination against LGBT+ people (not in

Civil Union)” are representative of the people who decided to participate in the survey alone.

24. As a whole more than a thousand LGB people were interviewed; differently from the first

survey more young people and women were reached out. The main results of this survey

have been published in May 2023 (Istat).

25. Respondents were asked to provide information on their sexual orientation, sex (currently

registered in the civil registry and at birth). For the first time a question on gender identity

was introduced. In line with the UNECE review on measuring gender identity (2019), a two-

step approach (a question that measures the sex assigned at birth and a question to assess the

current gender identity) was adopted.

26. Main results were analysed by the following profiles already used in the previous survey:

gay, lesbians, bisexual males and bisexual females. At the same time the introduction of

questions about gender identity allows to give a more inclusive gender representation

beyond the information that a question on legal sex can provide. Though the target

population was very specific, when cross-tabulated, the two-step measures provide counts of

cisgender women and men, and non-binary people.

Working paper 11 Rev1

5

27. A survey on “Labour Discrimination against Trans and Non-binary people”, based on a

convenience sampling, is currently in progress. This survey mainly focuses on gender

identity and gender expression issues with particular reference to the work experience.

28. Various SOGIESC indicators have been developed and included in the questionnaire

addressed to trans and non-binary people (e.g., sex, gender identity, gender expression,

intersexuality, gender affirming process) following a more updated theoretical and

methodological debate. Two-step measures of gender (sex assigned at birth and current

gender identity) are adopted to count trans and non-binary persons. The main results of this

survey are expected by the end of 2023.

III. Developing indicators on sex and gender identity

29. For the questionnaire design and for developing indicators relating to sex and gender (and

other SOGIESC) in all the surveys, a review of the international and national literature in the

field of official statistics and social research was carried out.

30. In order to take into account the specificity of the Italian context and get shared indicators,

various workshops, auditions and informal meetings were carried out with academics,

experts and LGBT+ organizations of the «Permanent Consultation Table for the promotion

of LGBT rights and the protection of LGBT persons» managed by Unar.

31. LGBT+ organizations were involved at different steps of the project; various working

groups were set up on a voluntary basis and considering the specific expertise.

32. Starting from some good practices implemented in other countries, an initial exploratory

workshop with associations was held in 2019 to discuss the main information needs

regarding the LGBT+ population in Italy and the conceptualization and definition of

SOGIESC characteristics.

33. After that, several meetings were held to discuss SOGIESC indicators, to keep up with

terminology changing and developing. Ad hoc documents were used to collect the LGBT+

associations’ feedback.

34. Among the main aspects that emerged during these meetings are the plurality of terms that

can be used to describe own sexual and gender identity and can change over a person's

lifetime and depending on the context; the importance of giving representation to all subjects

that fall within the acronym LGBT+ (e.g., bisexuals, intersexual) including non-binary

identities and overcoming of a medicalized vision.

35. On the other hand, the implementation time of surveys for official statistics are on average

longer because of quality control operations and compliance with certain official procedures;

moreover, the objective of statistics is to simplify and reduce complexity, but at the same

time to provide an accurate picture of the observed phenomenon and monitor it over time.

36. Definitions and indicators were also discussed with academics.

37. At the same time Istat discussed and shared this experience within the “Equality Data

Subgroup” of the EC “High-Level Group on Non-Discrimination, Equality and Diversity”

and within the “Task Team on Non-discrimination and Equality” of the Praia Group. The

Group published in 2022 the Handbook on Governance Statistics for National Statistical

Offices (2022) and non-discrimination and equality is one of the key dimensions of

governance.

Working paper 11 Rev1

6

38. The Istat-Unar project on “Labour Discrimination against LGBT+ people and diversity

policies in enterprises” has been included in the “Compendium of promising practices for

equality data collection” (FRA, 2019). Istat also participates in the “Task Team on Non-

discrimination and Equality” of the Praia Group which has the mandate to support the

development of international statistical guidance, standards and instruments for measuring

Non-Discrimination & Equality.

IV. Istat-Unar “Survey on Labour Discrimination against LGB people (not in Civil Union)”

39. As already mentioned, a first survey of the project addressed to people who chose Civil

Union as an official recognition of their relationship was carried out by Istat in 2021. It

studied a segment of the LGBT+ population that could be reached out by the municipal lists

of civilly united persons. The survey included questions about sex and sexual orientation.

Gender identity was not disclosed.

40. On the other hand the survey on Labour Discrimination against LGB people not in Civil

Union, carried out in 2022, had the original goal of detecting comparable and

complementary features to the previous survey.

41. The target population was made of LGB cisgender and non-binary people who have never

been in Civil Union. The “T” population, of any sexual orientation, was not included having

a third survey specifically focused on the topic of gender identity. This second survey was

also an experiment to test Web Respondent Driven Sampling (WebRDS) technique. This

technique did not work properly; therefore, the investigation was then opened to anyone of

the target population.

42. The beginning of the questionnaire was aimed at selecting the eligible respondents. A part

from being aged 18 and over, and living in Italy, they had to answer a question on their

current sexual orientation as “Homosexual; Bisexual; Other; Prefer not to say” in the last

two cases the respondent was directed out of the questionnaire. This because the target

population was limited to homosexual and bisexual persons.

43. In the questionnaire the terms homosexual and bisexual persons were defined as follows -

the former as attracted by same-sex persons and the latter by both-sex persons. This kind of

definition, which may not match with the contemporary approach and theoretical

understandings, was suitable for this kind of survey that wanted to be very specific and clear

in its target. The definition was based on sex and not on gender, in order to keep in line with

the first survey on individuals currently or formerly in Civil Union, which followed the

approach used by the Italian law ruling the Civil Unions.

44. In the first part of the survey respondents were asked to provide information on their sex

currently registered in the civil registry “Female; Male”. This question, combined with the

response to the question on sexual orientation, enabled to depict and study various profiles,

and the data analysis was based on them: lesbians; gays; bisexual females; bisexual males.

These profiles allowed to analyse discriminations and inequalities based on gender and

sexual orientation in the Italian labour market.

45. Though the target population was very specific for the first time a question on gender

identity was introduced by means of a two-step approach (Fig.1). This kind of approach is

suggested by international guidelines, and this was a way of testing it in the Italian context.

In addition to the above-mentioned question on the sex currently registered in the civil

registry, and in a different section a second question was asked on the sex assigned at birth,

Working paper 11 Rev1

7

and the possible items were “female/male/prefer not to say”. The answers on sex were also

combined with a question on their gender identity.

46. The question on gender identity was “How do you currently identify yourself?” The answer

items – “Woman/girl; Man/boy; Trans woman/girl; Trans man/boy; Non-binary or other;

Prefer not to say” adopted an inclusive approach albeit “Trans woman/girl” and “Trans man

/boy” were not a possible option to enter the target population. The answers were agreed,

also for this question with the associations and the other actors involved. The items were

purposely only a few in the aim of limiting the very many possible options and expressions.

Differently we would run the risk of having very little and not readable numbers.

47. This was a way to test this question and to begin including such a topic in a survey. It made

it possible to have data on binary and non-binary identities and various profiles also on the

basis of the target population and gender identity declared by the respondents - cisgender

homosexual men, cisgender homosexual women, cisgender bisexual women, cisgender

bisexual men; homosexual non-binary people and bisexual non-binary people. Very few

respondents preferred not to answer. It is worth noticing that the term “cisgender” was not

used in the questionnaire in agreement with the LGBT+ associations. It is indeed considered

a term for scholars and not well known by the possible respondents. In this survey gender

was inferred by combining their answers.

V. Istat-Unar “Survey on Discrimination against Trans a Non- Binary People”

48. A further step in the introduction of SOGIESC indicators was carried out with the design of

the third survey of the Istat-Unar project. The experience of previous surveys, the most

recent academic and terminological debate and new consultations with associations and

experts represented a starting point. This survey that should be on the field in June 2023 is

addressed to individuals aged 18 and over, who usually live in Italy, and whose gender

identity does not correspond to the sex they were assigned at birth.

49. The survey, conducted with a self-completed web questionnaire, is based on the voluntary

participation of people belonging to the target population. The initial dissemination of the

survey participation link goes through multiple channels, both by associations and by

individual respondents. The questionnaire investigates thematic areas and phenomena

already included in previous surveys but declined in such a way as to explore issues related

to gender identity and expression (e.g., coming out and visibility of one's gender identity,

discrimination) as well as some ad hoc phenomena (e.g., microaggressions against trans and

non-binary people, the process of gender affirmation).

50. Regardless of the number of people we will be able to reach out with this third survey, the

very design of questions on the topics of gender identity and expression, together with the

reflection undertaken not to give for granted a binary representation, are themselves

meaningful results of the project.

51. As with the previous one, this survey is based on the principle of self-identification whereby

the first questions of the questionnaire regarding gender identity are functional in identifying

people as belonging or not to the target population (binary and non-binary identities). An

inclusive and non-binary language was adopted through the use of the asterisk.

52. Gender identity is investigated by means of a two-step approach (Fig.1). A first question

asks the sex at birth (registered in the birth certificate) and the possible items were

female/male. Sex is an assignment process and in Italy there are only two legal categories.

Working paper 11 Rev1

8

The second question is about gender identity: “Thinking about your gender identity, how do

you currently identify yourself?” The answer items are: “1. Woman/girl; 2. Man/boy; 3.

Trans woman/trans girl; 4. Trans man/trans boy; 5. Non-binary gender identity; 6. Prefer not

to say”.

53. Although it is always preferable to include the item “other: specify” in this survey it was not

possible as questions on gender identity are functional to identify the target population.

54. By combining the answers to the questions on sex at birth and gender identity, the target

population can be identified. Those who declare a gender identity opposite to their sex at

birth and those who prefer not to answer the question on gender identity are not included in

the target population, the same happens for those who identify themselves as a man or

woman and afterwards declare themselves trans man and trans woman. If sex at birth is

different from gender identity, then respondents are labelled as “trans people” in the

questionnaire (e.g., sex at birth female and man as gender identity); those who declare a non-

binary gender identity are labelled as “non-binary people” in the questionnaire.

55. All the target population was labelled as “trans and non-binary persons” as not all the people

with a non-binary identity identify themselves in the umbrella label trans or transgender.

56. Compared to the question tested in the previous survey, the non-binary term is used without

adding the item “other”. Non-binary is used as an umbrella term, which includes all people

whose gender experience lies outside the female/male binary gender, thus referring to: a

presence of more than one gender (e.g., bigender, pangender), a fluctuation between

different genders (e.g., genderfluid), identification with a neutral gender within the

male/female spectrum or outside of it (e.g. genderqueer, neutral gender, third gender) or to a

partial identification with being a man or a woman (e.g., demiboy or demigirl), not to have a

gender identity (e.g., agender). Tooltips inside the questionnaire specify this.

57. The following open question was then included to capture the complexity of issues related to

identity: “How would you currently define your gender identity?” The aim of this question

was to give the interviewed people the opportunity to express their feelings and story, being

able to account for progressive coming out, for example as trans and then as non-binary, and

the coexistence of terms that according to some categorical approaches may apparently seem

contradictory (e.g., non-binary woman where woman is used as a political subjectivity).

58. Other questions have been included to capture gender as a process, more specifically the

following indicators of coming out milestones have been developed: “At what age did you

become aware of your current gender identity (trans and non-binary)?”; “At what age did

you begin using the terms trans, non-binary, or other to define your identity?” and “At what

age did you first come out as a trans or non-binary person?”. For all the three questions, the

item “I prefer not to answer” was inserted possible answer.

59. Gender is also investigated by means of indicators of gender expression. In detail two

questions have been included: “How would you currently describe yourself on the basis of

your appearance, your clothing, the way you move, the way you speak?” and “A person's

appearance, clothing, mannerism or the way of speaking can influence how others describe

them. How do you think others would currently describe you?”. For both the answers items

are: “1. Very feminine; 2. Feminine; 3. Rather feminine; 4. Neither feminine nor masculine;

5. Rather masculine; 6. Masculine; 7. Very masculine; 8. I don't know; 9. I prefer not to

say”.

60. For the first time, questions were developed to get to know intersex people, people which are

born with variations in sexual characteristics (such as sexual anatomy, reproductive organs

and/or chromosomal arrangements) that do not strictly belong to male or female categories

Working paper 11 Rev1

9

or belong simultaneously to both. This phenomena is not well known in Italy. Two different

dimensions, medicalization and perception, are investigated with the following two

questions: “Have you ever been diagnosed with an "intersex condition," either at birth or

later? and “Would you describe yourself as an intersex person?”. The answer items are: “1.

Yes; 2. No; 3. I prefer not to say”.

61. Another important aspect is the gender affirming process. According to the most recent

approach against medicalization, we do not refer to transition rather to gender affirming

process. So we ask: “To date, have you taken/performed any of the following actions? The

answer items are: “1. Dressing according to your gender identity; 2. Using a name that is

consistent with your gender identity (without registry adjustment); 3. Using a pronoun

consistent with your gender identity; 4. Using neutral nouns/neutral pronouns; 5. Taking a

psychological course aimed at the diagnosis of gender dysphoria; 6. Taking hormone

therapies; 7. Changing your master data; 8. Sex reassignment; 9. Gender affirmation surgery;

10. Other (Specify)”; 11. I prefer not to say.

62. In order to collect data about timing and difficulties in the procedure for change master data

and sex reassignment, these items have been included in the question about the gender

affirmation process.

63. The language aspect is also very relevant; so a question asks: “What pronouns do you want

others to refer to you with?” with the following options: 1. Masculine; 2. Feminine; 3. With

a u or schwa; 4. No pronouns in particular; 5. Other (Specify)”. Item 3 is specific for the

Italian language which marks with the final letter masculine and feminine.

64. Not limiting gender to a binary representation also implies a rethinking and reflection on

other definitions and methods of classification, such as, for example, the definition of sexual

orientation.

VI. Lessons learned and future prospects

65. The ongoing project that Istat launched in 2018 in collaboration with Unar on the topic of

labour discrimination against LGBT+ people and diversity policies offered the opportunity

to reflect as official statistics on the so-called SOGIESC indicators, to select and test

theoretical and operational definitions of these characteristics.

66. The introduction of these indicators gradually took place and was functional to the project's

research design, which provided for several surveys addressed to different LGBT+

population targets, also having the general objective of getting as close as possible to

probabilistic survey approaches.

67. An initial difficulty was to combine the multiplicity and mutability of terms used by people

to define complex aspects such as their (gender) identity and the need for official statistics to

classify and categorize, often according to a dichotomous perspective.

68. The discussion and collaboration by LGBT+ associations and stakeholders were crucial in

order to understand the meanings people give to certain terms. However, in order to avoid a

squashed vision of this reality, the involvement of LGBT+ people who do not belong to the

world of activism/associations is a key issue. The same happens for the share with academic

research, which has been studying these issues for a longer time, knows the specificities of

the Italian context and can observe the fast evolution of the terminological debate and

changes in society, such as those related to gender representations.

Working paper 11 Rev1

10

69. In addition to the indicators on sexual orientation, already included in the 2011 Istat survey,

for the first time in Italian official statistics indicators were introduced to detect the

respondents’ gender identity, not taking for granted the equivalence between biological sex

and gender, and giving visibility to a plurality of gender representations, not only the binary

one. This is an exploratory operation in a context such as the Italian one in which the public

debate on gender identity is rather new, as well as the systematic development of the

academic studies on Gender, Intersex, Feminist, Transfeminist and Sexuality.

70. The internationally recommended two-step approach to surveying gender identity was

implemented. The design of the survey on labour discrimination against trans and non-

binary people allowed the issue of gender to be addressed in its complexity and multi-facets

going beyond legislative and medical definitions and classifications. It emphasises the

importance of providing a range of indicators that are not limited to the registered sex and

self-identification by gender identity, but covers, for example, gender expression,

intersexuality, actions to affirm one's own gender and the interaction between gender and

other characteristics.

71. Finding the right trade-off between number of indicators and the questionnaire burden is

necessary, especially when shifting from data collection on a specific topic or population

(e.g. LGBT+ people) to general surveys on the whole population.

72. Further experimentation should be carried out using survey techniques other than the self-

administered web-based questionnaire and in surveys targeting the whole population (e.g.

the Istat Pilot Survey on Discrimination, 2022-2023).

73. Although the surveys were carried out on LGBT+ population groups, this experience

enabled the exploration and field-test of terminology choices, indicators and questions, and

their improvement also with a view to their inclusion in surveys addressed to the whole

population. Furthermore, the introduction of the SOGIESC indicators occurred following an

intersectional approach. An attempt was made on the one hand to detect as many

characteristics as possible through self-identification, in order to facilitate an intersectional

analysis of the data, and on the other hand to develop indicators to capture the specific

intersectional condition of certain groups/subjects (De Rosa, 2022).

74. Since the start of the project to date, the theoretical and methodological debate on the

SOGIESC indicators (and on the issue of discrimination) has developed and enriched. The

same has happened for the exchange of experiences at an institutional, European and

international level (UNECE, 2009; FRA, 2019; EC, 2021). The agreement on certain

defining aspects (sexual orientation defined with reference to gender and not sex, the

inappropriateness of using the term transsexual, the shift from the concept of gender

transition to that of gender affirmation) and the growing attention to the comparability of

data are a starting point and an important input for more inclusive institutional research.

75. In operational terms, the greatest effort consists in drawing on subjective experience and get

to a synthesis and a choice of terms in order to conceive definitions and indicators to achieve

reliable, valid and comparable SOGIESC data.

76. Thematic surveys or surveys on specific population groups are suitable for exploring issues

or aspects of society that are still little known such as trans and non-binary experiences.

However the long-term objective should be the introduction of SOGIESC self-identification

indicators in surveys targeted to the whole population in order to investigate and monitor

multiple and intersectional discrimination and inequalities through a combination of

“objective or outcome indicators”, “subjective indicators about experiences of

discrimination” and “group-specific indicators of discrimination” (De Rosa, 2022).

Working paper 11 Rev1

11

77. The focus on representing the different ways of performing gender in data collection leads to

understand statistics beyond an exclusively binary key. Non-binarism certainly represents a

great challenge but also gives a critical perspective to the processes of reasoning, valuing,

measuring, and comparing through definitions, classifications and numbers.

Figure 1

Questions on gender identity adopted in the Istat-Unar project on "Labour discrimination against

LGBT+ people and diversity policies implemented in enterprises" (2018-2023)

VII. References

78. De Rosa, E., de Martino, E., Scambia, F., Nur, N. (2022). Perspectives on LGBT+ working

lives: stakeholders, employers and LGBT+ people. RIEDS - Rivista Italiana di Economia,

Working paper 11 Rev1

12

Demografia e Statistica - The Italian Journal of Economic, Demographic and Statistical

Studies, SIEDS Societa' Italiana di Economia Demografia e Statistica, Vol. LXXVI-2 April-

June 2022, pp. 4-12. ISSN: 0035-6832.

79. De Rosa, E. (2022). Intersezionalità e discriminazioni LGBT+: paradigmi, concetti e

indicatori. AG AboutGender, vol. 11, no. 22, pp. 306-336, DOI: 10.15167/2279-

5057/AG2022.11.22.2024. ISSN 2279-5057.

80. De Rosa, E., De Vitiis, C., Inglese, F., Vitalini, A. (2020). Il Web-Respondent Driven

Sampling per lo studio della popolazione LGBT+. RIEDS - Rivista Italiana di Economia,

Demografia e Statistica, vol. LXXIV, no. 1 Gennaio-Marzo 2020, pp. 73-84. ISSN: 0035-

6832.

81. De Rosa, E., Inglese, F. (2018). Diseguaglianze e discriminazioni nei confronti delle persone

LGBT: quale contributo della statistica ufficiale?. RIEDS - Rivista Italiana di Economia,

Demografia e Statistica, vol. LXXII-4 Ottobre-Dicembre 2018, pp. 77-88. ISSN: 0035-6832.

82. European Commission, Directorate-General for Justice and Consumers. (2021). Guidelines

on improving the collection and use of equality data, Publications Office,

https://data.europa.eu/doi/10.2838/9725

83. FRA. (2019). Compendium of promising practices for equality data collection;

https://fra.europa.eu/en/promising-practices-list

84. Istat. (2013). Stereotipi, rinunce, discriminazioni di genere. Anno 2011;

https://www.istat.it/it/archivio/106599

85. Istat (2020). Diversity Management for LGBT+ Diversities in Enterprises and Desirable

Actions to Improve Inclusiveness at Work. Year 2019;

https://www.istat.it/it/files//2021/01/LGBT-Report.pdf

86. Istat (2022). Survey on Labour Discrimination toward LGBT+ People (in Civil Union or

formerly in union). Year 2020-2021;

https://www.istat.it/it/files//2022/05/REPORTDISCRIMINAZIONILGBT_2022_en.pdf

87. Istat (2023). L’Indagine Istat-Unar sulle discriminazioni lavorative nei confronti delle

persone LGBT+ (non in unione civile o già in unione). Anno 2022;

https://www.istat.it/it/files//2023/05/report-discriminazioni-15maggio.pdf

88. PRAIA Group. (2020). The Handbook on Governance Statistics;

https://unstats.un.org/unsd/statcom/51st-

session/documents/Handbook_on_GovernanceStatistics-Draft_for_global_consultation-

E.pdf

89. UNECE. (2019). In-Depth Review of Measuring Gender Identity. Conference of European

Statisticians, Paris; https://unece.org/sites/default/files/2021-01/In-

depth%20review%20of%20Measuring%20Gender%20Identity%20for%20bureau.pdf

* Подготовлена Эудженией Де Роса (раздел II , раздел V , раздел VI ), Валерией де Мартино (Введение,

раздел III) и Франческой Скамбиа (раздел IV).

ПРИМЕЧАНИЕ: Обозначения, используемые в настоящем документе, не подразумевают выражения какого-

либо мнения со стороны Секретариата Организации Объединенных Наций относительно правового статуса той

или иной страны, территории, города или района или их властей, или относительно делимитации их границ или

рубежей.

Европейская экономическая комиссия

Конференция европейских статистиков

Группа экспертов по гендерной статистике Женева, Швейцария, 10–12 мая 2023 года

Пункт D предварительной повестки дня Измерение показателей пола и гендера

Репрезентация гендерной идентичности при сборе данных: новые подходы, применяемые в Италии

Записка Национального института статистики Италии - ИСТАТа*

Резюме

ИСТАТ (Национальный статистический институт Италии) в сотрудничестве с ЮНАР (Национальное

управление по борьбе с дискриминацией) реализует проект «Трудовая дискриминация в отношении

представителей ЛГБТ+-сообщества и политика диверсификации на предприятиях», который начался в

2018 году. Он характеризуется использованием смешанного (количественно-качественного) метода и

предусматривает прямой сбор информации от разных целевых групп ЛГБТ+-сообщества.

Проект включает в себя три обследования по методу CAWI, основанные на самоидентификации

респондентов как ЛГБТ+ лиц и проведенные с помощью веб-анкеты для самостоятельного заполнения:

а) в 2020-2021 годах было проведено сплошное обследование физических лиц-резидентов (более 21 000

человек), которые по состоянию на 1 января 2020 года или ранее состояли в гражданском союзе. Основные

результаты были опубликованы в 2022 году (Istat, 2022). В Италии союз между лицами одного пола

регулируется правовым институтом под названием «гражданский союз» (с июля 2016 года), который

отличается от брака, который могут заключать только разнополые пары;

б) в 2022 году было проведено обследование ЛГБ-лиц, никогда не состоявших в гражданском союзе

(завершено в мае 2022 года). ИСТАТ впервые протестировал метод «снежного кома» в модификации RDS

(управляемая респондентом выборка), который впоследствии был заменен удобной выборкой;

c) обследование трансгендерных и небинарных лиц, которое проводится в настоящее время.

Были обсуждены, протестированы и проанализированы конкретные вопросы, касающиеся показателей

СОГИГСПП (сексуальная ориентация, гендерная идентичность, гендерное самовыражение и половые

признаки).

Цель настоящей статьи состоит в том, чтобы проиллюстрировать итальянский опыт в области проведения

обследований гендерной идентичности и постепенного внедрения других показателей СОГИГСПП в

официальную статистику. В ней подробно проиллюстрированы показатели пола и гендерной

идентичности, которые протестированы применительно к различным целевым группам.

Рабочий документ 8

Distr.: General

10 мая 2023

English

Рабочий документ

2

В заключение, в ней определены основные проблемные аспекты и предложены некоторые рекомендации

по улучшению гендерной репрезентации при сборе данных, а также по разработке показателей гендерной

идентичности, которые предлагается включить в официальные обследования, ориентированные на все

население.

I. Введение

1. В последние годы в области официальной статистики, как на международном, так и на

европейском уровне, развернулись широкие дебаты и сравнительные оценки на тему

так называемых показателей СОГИГСПП (показатели сексуальной ориентации,

гендерной идентичности и самовыражения и половых признаков). Общее намерение

национальных статистических институтов, международных и исследовательских

учреждений (например, Европейской экономической комиссии Организации

Объединенных Наций – ЕЭК ООН, Агентства Европейского союза по основным

правам – АОП, Прайской группы по статистике государственного управления –

Прайская группа) и органов по обеспечению равенства состоит в том, чтобы

производить более всеобъемлющие статистические данные и сопоставимые данные

для мониторинга неравенства и дискриминации по признаку СОГИГСПП, а также на

основе пересечения с другими соответствующими аспектами/признаками,

используемыми для определения идентичности и положения отдельных лиц и групп в

обществе (например, гражданство, возраст, социальный класс…).

2. Академическая и внеиздательская литература по вопросам ЛГБТ+, особенно на тему

трансгендерной и гендерно-неконформной идентичности, весьма богата и развита в

некоторых контекстах и представляет собой полезную исходную базу.

3. В 2011 году Национальный статистический институт Италии (ИСТАТ) впервые

рассмотрел вопросы, связанные с многообразием в плане сексуальной ориентации и

гендерной идентичности, когда в рамках «Обследования в области дискриминации по

признаку пола, сексуальной ориентации и этнического происхождения» была собрана

информация о мнениях и отношении к гендерным ролям, гомосексуализму и

иммиграции; а также проведена оценка числа жертв дискриминации в школе и/или на

работе (Istat, 2013). Обследование включало вопросы о сексуальной ориентации, что

позволило получить первую оценку численности гомосексуального и бисексуального

населения Италии.

4. В 2018 году ИСТАТ снова занялся изучением этих вопросов в рамках соглашения о

сотрудничестве, подписанного с ЮНАР (Национальное управление по борьбе с

дискриминацией), чтобы восполнить пробел в информации о ЛГБТ+-сообществе. Это

соглашение положило начало реализуемому в настоящее время проекту «Трудовая

дискриминация в отношении представителей ЛГБТ+-сообщества и политика

диверсификации на предприятиях». В рамках имеющихся бюджетных и временных

Рабочий документ

3

ресурсов в ходе проекта ИСТАТ-ЮНАР были проведены обследования с

применением как стандартных, так и нестандартных методов выборки.

5. В рамках этого проекта, который включал проведение обследований,

ориентированных на группы ЛГБТ+-сообщества, впервые были протестированы и

включены вопросы о гендерной идентичности, а также о сексуальной ориентации,

гендерном самовыражении и половых признаках. Определения и показатели,

представленные в анкетах, обсуждались с экспертами, учеными, ассоциациями и

представителями ЛГБТ+- сообщества и доводились до их сведения.

6. Проект ИСТАТ-ЮНАР по трудовой дискриминации представляет собой важный шаг

на пути к подготовке всеобъемлющей статистике, обеспечивающей представленность

множественных вариантов гендерной и сексуальной идентичности. В то же время они

обеспечивают основу для тестирования различных показателей СОГИГСПП и в

обследованиях, ориентированных на все население в целом.

7. Настоящий документ сосредоточен прежде всего на показателях, касающихся пола и

гендерной идентичности, и структурирован следующим образом: в разделе II

описывается методологическая основа проекта ИСТАТ-ЮНАР и способы включения

вопросов по СОГИГСПП в различные обследования в рамках проекта; в разделе III

показано, как разрабатывались и обсуждались показатели пола и гендера; в разделе IV

основное внимание уделяется показателям СОГИГСПП, использованным в

Обследовании трудовой дискриминации, ориентированном на представителей ЛГБ-

сообщества, не состоящим в гражданском союзе (2022 год); в то время как раздел V

посвящен показателям, разработанным для обследования, ориентированного на

трансгендерных и небинарных лиц. В заключительном разделе излагаются

соображения, вытекающие из нынешнего опыта, наряду с некоторыми более общими

рекомендациями.

II. Проект ИСТАТ-ЮНАР «Трудовая дискриминация в отношении представителей ЛГБТ+-сообщества и политика диверсификации на предприятиях» и показатели СОГИГСПП

8. Проект ИСТАТ-ЮНАР характеризуется использованием смешанного количественно-

качественного и мультиперспективного подхода (ЛГБТ+ лица, работодатели,

заинтересованные стороны), а также взаимодействием между экспертами, учеными и

ЛГБТ+- ассоциациями (De Rosa, Inglese, 2018; Istat, 2020 ; De Роза et. al 2023). Он

предусматривает прямой сбор информации от представителей ЛГБТ+-сообщества

(первая макрообласть), от работодателей, особенно предприятий, и от основных

заинтересованных сторон (вторая макрообласть).

9. В проекте используется подход, основанный на широком участии. Он

предусматривает создание различных рабочих групп, в состав которых входят

ассоциации и органы Постоянного консультативного стола по продвижению прав

ЛГБТ и защите ЛГБТ-лиц, созданного в 2018 году при Президиуме Совета министров.

К разработке опросной анкеты и показателей также привлекались лица, не

являющиеся членами ассоциаций ЛГБТ+, эксперты и представители научных кругов,

10. Проект финансировался из фондов ЕС.

Рабочий документ

4

11. В настоящем документе основное внимание уделяется первой макрообласти проекта

и, в частности, репрезентации гендерной идентичности в процессе сбора данных при

проведении обследований на тему пола и гендерной идентичности.

12. Целью первой макрообласти проекта было составить представление о ситуации с

трудовой дискриминацией в отношении представителей ЛГБТ+-сообщества в Италии

с помощью различных обследований, проводимых с целью сбора информации от

различных целевых групп ЛГБТ+-сообщества.

13. В статистических репрезентативных обследованиях населения ЛГБТ+ отмечается

значительная необъективность по причине отсутствия знаний об этих группах

населения, например, отсутствия теоретических основ для конструирования

вероятностных выборок. Теоретическую и статистическую репрезентативность

различных групп, охватываемых этой аббревиатурой, трудно обеспечить из-за их

относительно низкой доли в популяции.

14. Были запланированы три обследования по методу CAWI на основе веб-анкеты для

самостоятельного заполнения. Основными объектами исследования являются:

каминг-аут, опыт дискриминации при поиске работы или на работе, а также

дискриминация в других сферах общественной жизни (например, в школе),

микроагрессии, агрессии, ненавистнических высказываниц. Первые два обследования

в основном были сосредоточены на аспектах, связанных с сексуальной ориентацией;

третье посвящено вопросам гендерной идентичности.

15. Ключевым принятым принципом была самоидентификация респондентов как ЛГБТ+.

Показатели СОГИГСПП постепенно тестировались и включались в три обследования.

16. Первое обследование по проблеме трудовой дискриминации было проведено в 2020-

2021 годах. Оно представляло собой сплошное обследование физических лиц-

резидентов (более 21 000 человек), которые по состоянию на 1 января 2020 года или

ранее состояли в гражданском союзе (однополые пары). С июля 2016 года в Италии

союз однополых лиц старше 18 лет регулируется специальным институтом под

названием «гражданский союз». Он отличается от брака, который могут заключать

только разнополые пары.

17. Основные результаты «Обследования трудовой дискриминации в отношении

представителей ЛГБТ+-сообщества (состоящих или ранее состоявших в гражданском

союзе)» были опубликованы в 2022 году (Istat, 2022). Однополые пары, состоящие в

гражданском союзе, представляют собой особую группу ЛГБТ+-населения,

проживающего в Италии. Среди них действительно были выявлены некоторые

конкретные характеристики: в своем большинстве это были гомосексуальные и

бисексуальные лица, мужчины, лица старшего возраста (43,6% гомосексуальных и

бисексуальных лиц были старше 50 лет), не скрывающими свою ориентацию и

хорошо интегрированные в рынок труда.

18. Обследование включало вопросы о поле и сексуальной ориентации. Указать свою

гендерную идентичность не предлагалось. 95,2% людей, состоящих или ранее

состоявших в гражданском союзе, проживающих в Италии, заявили о своей

гомосексуальной или бисексуальной ориентации. Основные результаты были

проанализированы по следующим профилям: геи, лесбиянки, бисексуальные

мужчины и бисексуальные женщины.

19. Второе обследование, дополняющее первое, было ориентировано на представителей

ЛГБ-сообщества, которые никогда не состояли в гражданском союзе. «Обследование

трудовой дискриминации в отношении представителей ЛГБ-сообщества (не

Рабочий документ

5

состоящих в гражданском союзе)» было проведено в 2022 году и включало вопросы

для определения пола, сексуальной ориентации и гендерной идентичности. Целевая

группа населения была ограничена гомосексуальными и бисексуальными

цисгендерными и небинарными лицами.

20. В рамках этого обследования ИСТАТ впервые применил метод «снежного кома» в

модификации RDS (управляемая респондентом выборка) с использованием его веб-

версии (WebRDS). Этот метод основан на социальных отношениях, которые

используются для охвата так называемых скрытых и труднодоступных групп

населения (De Rosa et. al 2020).

21. RDS сочетает в себе метод «снежного кома», в соответствии с которым выборка

конструируется с использованием имен, предоставленных первоначальными

рекрутерами, с математической моделью, которая формализует, с соблюдением

определенных условий, процесс рекрутирования в виде цепи Маркова или

вероятностного процесса. Данные, собранные в процессе формирования выборки,

используются для того, чтобы сделать выводы о структуре социальной сети и на этой

основе получить несмещенную оценку исследуемой группы населения. Для этого

необходимо, чтобы респонденты играли активную роль в рекрутировании новых

респондентов, которые принадлежат к их сети взаимоотношений.

22. Около пятидесяти ЛГБТ+-ассоциаций согласились оказать содействие в проведении

обследования и после подписания с ИСТАТом соглашения о защите

конфиденциальности были привлечены для определения первичных («посевных»)

участников сетевой цепочки. По прошествии установленного времени с начала

обследования, несмотря на наличие некоторого количества заполненных анкет,

появились основания полагать, что методика «снежного кома» RDS не работает

должным образом по разным причинам. Для продолжения работы был рассмотрен

вариант удобной выборки, поскольку она в любом случае могла бы предоставить

интересную качественную информацию о целевой популяции гомосексуальных и

бисексуальных лиц.

23. Окончательные результаты «Обследования трудовой дискриминации в отношении

представителей ЛГБ-сообщества (не состоящих в гражданском союзе)» являются

репрезентативными для людей, которые решили участвовать в обследовании

самостоятельно.

24. Всего было опрошено более тысячи ЛГБ-лиц; в отличие от первого обследования

удалось охватить больше молодых людей и женщин. Основные результаты этого

обследования будут опубликованы к концу мая.

25. Респондентов просили предоставить информацию о своей сексуальной ориентации и

поле (зарегистрированном в гражданском реестре на данный момент и при рождении).

Впервые был включен вопрос о гендерной идентичности. В соответствии с обзором

ЕЭК ООН по измерению показателей гендерной идентичности (2019 год) был принят

двухступенчатый подход (вопрос для измерения показателей пола,

зарегистрированного при рождении, и вопрос для оценки текущей гендерной

идентичности.

26. Основные результаты были проанализированы по следующим профилям, которые уже

использовались в предыдущем обследовании: геи, лесбиянки, бисексуальные

мужчины и бисексуальные женщины. В то же время включение вопросов о гендерной

идентичности обеспечивает более всеобъемлющую гендерную репрезентацию,

выходящую за рамки той информации, которую позволяет получить вопрос о поле,

Рабочий документ

6

зарегистрированном в юридических документах. Несмотря на то, что целевая группа

была очень специфичной, при сведении в комбинационные таблицы эти

двуступенчатые количественные показатели позволяют получить информацию о

количестве цисгендерных женщин и мужчин, а также небинарных лиц.

27. В настоящее время проводится обследование «Трудовая дискриминация в сфере туда

в отношении трансгендерных и небинарных лиц», основанное на удобной выборке.

Это обследование в основном фокусируется на вопросах гендерной идентичности и

гендерного самовыражения с особым упором на опыт в сфере труда.

28. Различные показатели СОГИГСПП были разработаны и включены в анкету,

предназначенную для трансгендерных и небинарных лиц (например, пол, гендерная

идентичность, гендерное самовыражение, интерсексуальность, процесс

подтверждения гендерной идентичности), после теоретических и методологических

дебатов на основе актуализированной информации. Для подсчета количества

трансгендерных и небинарных лиц применяются двухступенчатые количественные

показатели для измерения пола (пол, зарегистрированный при рождении, и текущая

гендерная идентичность). Ожидается, что основные результаты этого обследования

будут готовы к концу 2023 года.

III. Разработка показателей для измерения пола и гендерной идентичности

29. Для дизайна анкеты и разработки показателей для измерения пола и гендера (и других

показателей СОГИГСПП) во всех обследованиях был проведен обзор международных

и национальных литературных источников в области официальной статистики и

социальных исследований.

30. Чтобы учесть специфику итальянского контекста и получить общие показатели, были

проведены различные семинары, собеседования и неформальные встречи с

представителями научных кругов, экспертами и ЛГБТ+-организациями, входящими в

состав Постоянного консультативного стола по продвижению прав ЛГБТ и защите

ЛГБТ-лиц, действующего под руководством ЮНАР.

31. ЛГБТ+-организации участвовали в проекте на разных его этапах; различные рабочие

группы создавались на добровольной основе и с учетом конкретного экспертного

опыта.

32. На основе некоторых передовых практик, реализованных в других странах, в 2019

году был проведен первоначальный ознакомительный семинар с участием

ассоциаций для обсуждения основных потребностей в информации, касающейся

ЛГБТ+-населения в Италии, а также концептуализации и определения признаков

СОГИГСПП.

33. После этого было проведено несколько встреч для обсуждения показателей

СОГИГСПП, чтобы идти в ногу с изменением и развитием терминологии. Для

получения обратной связи от ЛГБТ+-ассоциаций использовались специальные

документы.

34. Среди основных аспектов, выявленных в ходе этих встреч, можно отметить

множественность терминов, которые могут использоваться для описания собственной

сексуальной и гендерной идентичности и могут меняться на протяжении жизни

Рабочий документ

7

человека и в зависимости от контекста; важность обеспечения репрезентации всех

субъектов, подпадающих под аббревиатуру ЛГБТ+ (например, бисексуальных и

интерсексуальных лиц), включая небинарные идентичности, а также преодоление

медикализированного видения.

35. С другой стороны, проведение обследований с целью получения официальной

статистики в среднем занимает больше времени больше из-за осуществления

операций по контролю качества и необходимости соблюдения определенных

официальных процедур; кроме того, цель статистики состоит в том, чтобы упростить

и уменьшить степень сложности, но в то же время дать точную картину наблюдаемого

явления и отслеживать его динамику с течением времени.

36. Определения и показатели также обсуждались с представителями научных кругов.

37. В то же время ИСТАТ обсудил и поделился этим опытом в рамках Подгруппы по

данным о положении в области равенства Группы высокого уровня по

недискриминации, равенству и разнообразию Евросоюза и в рамках Целевой группы

по вопросам недискриминации и равенства Прайской группы. В 2022 году Группа

опубликовала «Справочник по статистике государственного управления для

национальных статистических управлений» (2022 год), где недискриминация и

равенство рассматриваются в качестве одного из ключевых аспектов

государственного управления.

38. Проект ИСТАТ-ЮНАР «Трудовая дискриминация в отношении представителей

ЛГБТ+-сообщества и политика диверсификации на предприятиях» был включен в

«Сборник перспективных практических методов сбора данных о положении в области

равенства» (FRA, 2019). ИСТАТ также участвует в работе Целевой группы по

вопросам недискриминации и равенства Прайской группы, уполномоченной

оказывать поддержку в разработке международных статистических руководств,

стандартов и инструментов для измерения показателей недискриминации и

равенства.

IV. Обследование ИСТАТ-ЮНАР по проблеме трудовой дискриминации в отношении представителей ЛГБ- сообщества (не состоящих в гражданском союзе)»

39. Как уже упоминалось, первое обследование в рамках проекта, ориентированное на

людей, выбравшим гражданский союз в качестве официального признания своих

отношений, было проведено ИСТАТом в 2021 году. Был изучен сегмент ЛГБТ+-

населения, который можно было охватить только с использованием муниципальных

списков лиц, состоящих в гражданских союзах. Обследование включало вопросы о

поле и сексуальной ориентации. Гендерная идентичность не указывалась.

40. С другой стороны, проведенное в 2022 году Обследование трудовой дискриминации в

отношении представителей ЛГБ-сообщества, не состоящих в гражданском союзе,

первоначально ставило целью выявить характеристики, которые были бы

сопоставимы и дополняли бы результаты предыдущего обследования.

41. Целевая группа населения состояла из цисгендерных и небинарных ЛГБ-лиц, которые

никогда не состояли в гражданском союзе . Популяция «Т» любой сексуальной

ориентации не была включена в третье обследование, специально посвященное теме

гендерной идентичности. Это второе обследование также представляло собой

эксперимент по тестированию метода веб-выборки, управляемой респондентами

Рабочий документ

8

(WebRDS). Этот метод не дал должных результатов, в связи с чем обследование было

открыто для любого лица из целевой группы населения.

42. Начальные вопросы анкеты были направлены на отбор подходящих респондентов.

Помимо того, что им должно было быть от 18 лет и старше и они должны были

проживать в Италии, они должны были ответить на вопрос о своей нынешней

сексуальной ориентации как «Гомосексуальня; Бисексуальная; Другая; Предпочитаю

не отвечать», причем в последних двух случаях респондент направлялся на выход из

опроса. Это было связано с тем, что целевая группа населения была ограничена

гомосексуальными и бисексуальными лицами.

43. В опросной анкете термины «гомосексуальные лица» и «бисексуальные лица» были

определены следующим образом: первые определялись как испытывающие влечение

к лицам одного с ними пола, а вторые – к лицам обоих полов. Такое определение,

которое может не соответствовать современному подходу и теоретическому

пониманию, подходило для обследования такого рода, цель которого была очень

конкретной и ясной. Определение было основано на поле, а не на гендере, чтобы

обеспечить согласованность с первым обследованием лиц, в настоящее время

состоящих или ранее состоявших в гражданском союзе, которое следовало подходу,

используемому в итальянском законодательстве, регулирующем гражданские союзы.

44. В первой части обследования респондентов просили указать сведения о своем поле,

зарегистрированном на данный момент в гражданском реестре: «Женский; Мужской».

Этот вопрос в сочетании с ответом на вопрос о сексуальной ориентации позволил

описать и изучить различные профили, на основе которых был проведен анализ

данных: гомосексуальные женщины; гомосексуальные мужчины; бисексуальные

женщины; бисексуальные мужчины. Эти профили позволили проанализировать

дискриминацию и неравенство по гендерному признаку и признаку сексуальной

ориентации на итальянском рынке труда.

45. Несмотря на то, что целевая группа населения была очень специфичной, в

обследование впервые был включен вопрос о гендерной идентичности с

использованием двухступенчатого подхода (Рисунок 1). Такой подход предлагается

международными руководствами, и обследование дало возможность протестировать

его в итальянском контексте. В дополнение к вышеупомянутому вопросу о поле,

зарегистрированном на данный момент в гражданском реестре, в другом разделе был

задан второй вопрос о поле, зарегистрированном при рождении, с возможными

вариантами ответов: «женский/ мужской/ предпочитаю не указывать». Ответы на

вопросы, касающиеся пола, также были объединены с вопросом о гендерной

идентичности.

46. Вопрос о гендерной идентичности был сформулирован следующим образом: «Как вы

себя идентифицируете в настоящее время?» Варианты ответов: «Женщина/девочка;

мужчина/мальчик; Транс-женщина/девочка; Транc-мужчина/мальчик; Небинарное

лицо или другой вариант; Предпочитаю не указывать» были сформулированы с

применением инклюзивного подхода, хотя «транс-женщина/девочка» и «транс-

мужчина/мальчик» не являлись возможными вариантами для охвата целевой группы

населения. Ответы, в том числе и на этот вопрос, были согласованы с ассоциациями и

другими заинтересованными сторонами. Вариантов ответов преднамеренно было

немного, чтобы ограничить очень большое количество возможных ответов и

формулировок. В противном случае мы рисковали бы получить очень низкие и

нечитаемые количественные показатели.

Рабочий документ

9

47. Это дало возможность протестировать этот вопрос и начать включать эту тему в

обследования. Обследование позволило получить данные о бинарных и небинарных

идентичностях и различных профилях в том числе и на основе целевой группы

населения и заявленной респондентами гендерной идентичности - цисгендерные

гомосексуальные мужчины, цисгендерные гомосексуальные женщины, цисгендерные

бисексуальные женщины, цисгендерные бисексуальные мужчины; гомосексуальные

небинарные лица и бисексуальные небинарные лица. Лишь очень немногие

респонденты предпочли не отвечать. Стоит отметить, что термин «цисгендерный» в

анкете не использовался по согласованию с ЛГБТ+-ассоциациями. Он, несомненно,

считается термином для ученых и не очень хорошо известен возможным

респондентам. В этом обследовании гендер определялся путем совмещения ответов

респондентов.

V. Обследование ИСТАТ-ЮНАР по проблеме дискриминации в отношении небинарных транс-лиц

48. Дальнейший шаг по внедрению показателей СОГИГСПП был сделан при разработке

третьего обследования в рамках проекта ИСТАТа-ЮНАР. Отправной точкой стал

опыт, полученный по итогам предыдущих обследований, последних академических и

терминологических дебатов и новых консультаций с ассоциациями и экспертами. Это

обследование, которое должно быть проведено в июне 2023 года, ориентировано на

лиц в возрасте 18 лет и старше, которые обычно проживают в Италии и чья гендерная

идентичность не соответствует полу, зарегистрированному при рождении.

49. Обследование, проводимое с помощью самостоятельно заполняемой веб-анкеты,

основано на добровольном участии людей, принадлежащих к целевой группе

населения. Первоначальное распространение ссылки для участия в обследовании

осуществляется по нескольким каналам как ассоциациями, так и отдельными

респондентами. Анкета предназначена для изучения тематических областей и

явлений, которые уже включались в предыдущие обследования, но построена таким

образом, чтобы исследовать вопросы, связанные с гендерной идентичностью и

самовыражением (например, каминг-аут и проявление своей гендерной идентичности,

дискриминация), а также некоторые индивидуализированные явления (например,

микроагрессия в отношении трансгендерных и небинарных лиц, процесс

подтверждения гендерной идентичности).

50. Независимо от количества людей, которых мы сможем охватить с помощью этого

третьего обследования, сами формулировки вопросов по темам гендерной

идентичности и самовыражения вместе с теоретической работой, проведенной для

того, чтобы избежать бинарной репрезентации как само собой разумеющейся,

являются значимыми результатами проекта.

51. Как и в предыдущем случае, это обследование основано на принципе

самоидентификации, в соответствии с которым первые вопросы анкеты, касающиеся

гендерной идентичности, служат для идентификации людей как принадлежащих или

не принадлежащих к целевой группе населения (бинарные и небинарные

идентичности). Инклюзивный и небинарный характер используемой в обследовании

лексики был обеспечен с помощью знака «*».

52. Гендерная идентичность исследуется с помощью двухступечатого подхода

(Рисунок 1). Первый вопрос касается пола, зарегистрированного при рождении

(зарегистрированного в свидетельстве о рождении), и возможные варианты ответов

Рабочий документ

10

были «женский/мужской». Пол регистрируется в соответствии с определенной

процедурой, и в Италии существуют только две юридические категории. Второй

вопрос касается гендерной идентичности: «Как Вы себя идентифицируете себя на

сегодняшний день в плане гендерной идентичности?» Варианты ответа: «1.

Женщина/девочка; 2. Мужчина/мальчик; 3. Транс-женщина/транс-девочка; 4. Транс-

мужчина/транс-мальчик; 5. Небинарная гендерная идентичность; 6. Предпочитаю не

отвечать».

53. Несмотря на то, что всегда предпочтительнее включать в такое обследование вариант

ответа «другой вариант: уточните», сделать это было невозможно, поскольку вопросы

о гендерной идентичности служат для определения целевой группы населения.

54. Совместив ответы на вопросы о поле, зарегистрированном при рождении, и гендерной

идентичности, можно определить целевую группу населения. Те, кто указывает

гендерную идентичность, противоположной своему полу при рождении, и те, кто

предпочитает не отвечать на вопрос о гендерной идентичности, не включаются в

целевую группу; то же самое происходит с теми, кто идентифицирует себя как

мужчина или женщина, а затем называет себя транс-мужчиной и транс-женщиной.

Если пол, зарегистрированный при рождении, отличается от гендерной идентичности,

то респонденты обозначаются в анкете как «трансгендерные лица» (например, пол при

рождении указан как женский, а в качестве гендерной идентичности указано

«мужчина»); а те, кто заявляет о небинарной гендерной идентичности, помечаются в

анкете как «небинарные лица».

55. Вся целевая группа населения была обозначена как «трансгендерные и небинарные

лица», поскольку не все люди с небинарной идентичностью идентифицируют себя с

использованием общего обозначения «транс» или «трансгендерные»

56. По сравнению с вопросом, протестированным в рамках предыдущего обследования,

термин «небинарный» используется без добавления варианта ответа «другое». Термин

«небинарный» используется в качестве общего термина, который охватывает всех

лиц, чей гендерный опыт выходит за рамки бинарной гендерной системы

«мужчина/женщина», что означает: принадлежность к более чем одному полу

(например, бигендер, пангендер), изменчивую гендерную самоидентификацию

(например, гендерфлюид), идентификацию себя как гендерно-нейтрального лица в

пределах или за пределами мужского/женского спектра (например, гендерквир,

нейтральный гендер, третий гендер) или частичную идентификацию как мужчина или

женщина (например, демибой или демигерл), отсутствие гендерной идентичности

(например, агендер). Это указано в оперативных подсказках внутри анкеты.

57. Затем в обследование был включен следующий открытый вопрос, чтобы отразить

сложность вопросов, связанных с идентичностью: «Как бы вы в настоящее время

определили свою гендерную идентичность?» Цель этого вопроса состояла в том,

чтобы дать опрашиваемым людям возможность выразить свои чувства и раскрыть

свою историю, объяснив последовательность своего каминг-аута, например, в

качестве трансгендера, а затем в качестве небинарного лица, и сосуществование

терминов, которые согласно некоторым категориальным подходам могут показаться

противоречивыми (например, небинарная женщина, где слово «женщина»

используется как политическая субъективность).

58. В обследование были включены и другие вопросы, цель которых состояла в том,

чтобы отразить гендер как процесс; в частности, были разработаны следующие

индикаторы этапов камин-аута: «В каком возрасте Вы осознали свою нынешнюю

гендерную идентичность (транс и небинарность)?»; «В каком возрасте Вы начали

Рабочий документ

11

использовать термины «транс», «небинарный» и т. д. для определения своей

идентичности? и «В каком возрасте Вы впервые совершили каминг-аут в качестве

трансгендерного или небинарного лица?». Для всех трех вопросов был предусмотрен

возможный ответ «Предпочитаю не отвечать».

59. Гендер также исследуется с помощью индикаторов гендерного самовыражения. В

частности, в анкету были включены два вопроса: «Как бы Вы описали себя в

настоящее время на основе Вашей внешности, Вашей одежды, манеры двигаться,

манеры говорить?» и «Внешний вид, одежда, манера поведения или речи человека

могут влиять на то, как его описывают другие. Как Вы думаете, как бы другие люди

описали Вас в настоящее время?». Варианты ответов для обоих вопросов: «1. Очень

женственный(ая); 2. Женственный(ая); 3. Довольно женственный(ая); 4. Ни

женственный(ая), ни мужественный(ая); 5. Скорее мужественный(ая); 6.

Мужественный(ая); 7. Очень мужественный(ая); 8. Не знаю; 9. Предпочитаю не

отвечать».

60. Впервые были разработаны вопросы для получения информации о интерсексуалах –

людях, которые рождаются с вариациями половых признаков (таких как половая

анатомия, репродуктивные органы и/или набор хромосом), которые не относятся

строго к категории мужских или женских или относятся одновременно к обеим

категориям. Это явление мало известно в Италии. Два разных аспекта –

медикализация и восприятие – исследуются с помощью следующих двух вопросов:

«Была ли у Вас когда-либо диагностирована «интерсекс-вариация» при рождении или

позже? и «Могли бы Вы назвать себя интерсексуалом?». Варианты ответа: «1. Да; 2.

Нет; 3. Предпочитаю не отвечать».

61. Еще одним важным аспектом является процесс подтверждения гендерной

идентичности. В соответствии с самым современным подходом, не приветствующим

медикализацию, мы говорим не о переходе, а о процессе подтверждения гендерной

идентичности. Поэтому мы задаем следующий вопрос: «На сегодняшний день

предпринимали/совершали ли Вы какие-либо из следующих действий? Пункты

ответа: «1. Ношение одежды в соответствии с Вашей гендерной идентичностью; 2.

Использование имени, соответствующего Вашей гендерной идентичности (без

изменения имени в гражданском реестре); 3. Использование местоимения,

соответствующего Вашей гендерной идентичности; 4. Использование нейтральных

существительных/нейтральных местоимений; 5. Прохождение психологического

курса, направленного на диагностику гендерной дисфории; 6. Прохождение

гормональной терапии; 7. Изменение Ваших основных данных; 8. Коррекция пола; 9.

Хирургическое вмешательство для подтверждения гендерной идентичности; 10.

Другое (указать); 11. Предпочитаю не отвечать.

62. С целью сбора данных о сроках и сложности процедуры изменения основных данных

и коррекции пола эти пункты были включены в вопрос о процессе подтверждения

гендерной идентичности.

63. Языковой аспект также очень актуален; поэтому вопрос сформулирован следующим

образом: «Какие местоимения Вы хотите, чтобы другие люди использовали,

обращаясь к Вам?» со следующими вариантами ответов: 1. Мужского рода; 2.

Женского рода; 3. С “u” или нейтральным гласным звуком; 4. Никаких конкретных

местоимений; 5. Другое (указать). Вариант 3 относится конкретно к итальянскому

языку, в котором мужской и женский род обозначается последней буквой слова.

Рабочий документ

12

64. Отказ от ограничения гендерной идентичности бинарной репрезентацией также

предполагает переосмысление и анализ других определений и методов

классификации, таких, например, как определение сексуальной ориентации.

VI. Полученные уроки и перспективы на будущее

65. Текущий проект, который ИСТАТ запустил в 2018 году в сотрудничестве с ЮНАР на

тему трудовой дискриминации в отношении представителей ЛГБT+-сообщества и

политики диверсификации, дал возможность отразить так называемые показатели

СОГИГСПП в официальной статистике, отобрать и протестировать теоретические и

рабочие определения этих признаков.

66. Включение этих показателей происходило постепенно и соответствовало дизайну

исследования в рамках проекта, который предусматривал проведение нескольких

обследований, ориентированных на различные целевые группы ЛГБТ+-сообщества, а

также преследовал общую цель максимально приблизиться к вероятностным методам

обследования.

67. Первоначальные трудности заключались в том, чтобы совместить множественность и

изменчивость терминов, используемых людьми для определения сложных аспектов,

таких как собственная (гендерная) идентичность, с необходимостью получения

официальных статистических данных для классификации и категоризации, зачастую

на основе дихотомического подхода.

68. Обсуждение и сотрудничество с ЛГБТ+-ассоциациями и заинтересованными

сторонами имели решающее значение для понимания смысла, который люди

вкладывают в определенные термины. Однако, чтобы избежать размытого видения

этих реалий, одной из ключевых задач является вовлечение ЛГБТ+-лиц, не

принадлежащих к миру активизма/ассоциаций. То же самое относится и к обмену

информацией с представителями научно-исследовательских кругов, которые изучают

эти вопросы в течение более длительного времени, знают особенности итальянского

контекста и могут наблюдать быструю эволюцию терминологических дебатов и

изменения в обществе, например, связанные с гендерной репрезентацией.

69. В дополнение к показателям сексуальной ориентации, которые уже включались в

обследование ИСТАТа 2011 года, впервые в официальной статистике Италии были

использованы показатели для выявления гендерной идентичности респондентов,

которые не опирались на презумпцию эквивалентности биологического пола и

гендерной идентичности человека, а обеспечивали возможность выявления множества

гендерных репрезентаций, не ограничивающихся бинарной системой. В контексте,

подобном итальянскому, в котором общественная дискуссия на тему гендерной

идентичности, так же как и систематическое развитие академических исследований на

тему гендера, интерсексуальности, феминизма, трансфеминизма и сексуальности

начались лишь недавно, это обследование представляет собой исследовательскую

работу.

70. Для проведения обследования применялся рекомендованный на международном

уровне двухступенчатый подход к исследованию гендерной идентичности. Дизайн

обследования по проблеме трудовой дискриминации трансгендерных и небинарных

лиц позволил рассмотреть проблему гендера во всей ее сложности и многогранности,

с выходом за рамки законодательных и медицинских определений и классификаций.

Подчеркивается важность наличия ряда показателей, которые не ограничиваются

Рабочий документ

13

зарегистрированным полом и самоидентификацией по признаку гендерной

идентичности, а охватывают, к примеру, гендерное самовыражение,

интерсексуальность, действия по утверждению собственной гендерной идентичности

и взаимосвязь между гендером и другими характеристиками.

71. Несмотря на то, что обследования проводились среди групп населения ЛГБТ+, этот

опыт позволил изучить и апробировать на практике варианты терминологии,

индикаторы и вопросы, а также оптимизировать их с целью включения в

обследования, предназначенные для всего населения. Кроме того, показатели

СОГИГСПП были введены на основе интерсекционального подхода. Была

предпринята попытка, с одной стороны, выявить как можно больше характеристик

посредством самоидентификации, чтобы облегчить интерсекциональный анализ

данных, а с другой стороны, разработать показатели, отражающие конкретное

интерсекциональное состояние определенных групп/субъектов (De Rosa, 2022).

72. С момента начала проекта по настоящее время теоретические и методологические

дискуссии по показателям СОГИГСПП (и по проблеме дискриминации) развивались и

обогащались. То же самое произошло и с обменом опытом на институциональном,

европейском и международном уровнях (UNECE, 2009; FRA, 2019; EC, 2021).

Согласие по некоторым определяющим аспектам (сексуальная ориентация,

определяемая по гендеру, а не по полу; нецелесообразность использования термина

«транссексуал», переход от концепции гендерного перехода к концепции

подтверждения гендерной идентичности) и растущее внимание к сопоставимости

данных являются отправной точкой и важным вкладом в более всеобъемлющие

институциональные исследования.

73. В оперативном плане наибольшие усилия потребовались для того, чтобы, опираясь на

субъективный опыт, перейти к синтезу и выбору терминов, чтобы сформулировать

определения и показатели для получения надежных, достоверных и сопоставимых

данных СОГИГСПП.

74. Тематические обследования или обследования конкретных групп населения подходят

для изучения проблем или аспектов общества, которые еще мало известны, например,

опыт трансгендерных и небинарных людей. Однако долгосрочной целью должно быть

включение индикаторов самоидентификации по признакам СОГИГСПП в

обследования, охватывающие все население, с целью изучения и мониторинга

множественной и перекрестной дискриминации и неравенства с помощью

комбинации «объективных или итоговых показателей», «субъективных показателей,

касающихся опыта дискриминации» и «показателей дискриминации, относящихся к

конкретным группам» (De Rosa, 2022).

75. Сосредоточение внимания на представлении различных способов учета гендерных

аспектов в процессе сбора данных помогает понять статистические данные,

выходящие за рамки исключительно бинарной модели. Небинарная модель,

безусловно, представляет собой серьезный вызов, но она также позволяет критически

анализировать процессы рассуждения, оценки, измерения и сравнения посредством

определений, классификаций и количественных показателей.

Рабочий документ

14

Рисунок 1

Вопросы, касающиеся гендерной идентичности, использованные в рамках проекта

ИСТАТ-ЮНАР «Трудовая дискриминация в отношении представителей ЛГБТ+-

сообщества и политика диверсификации на предприятиях» (2018-2023 годы)

2ое ОБСЛЕДОВАНИЕ ТРУДОВОЙ ДИСКРИМИНАЦИИ В ОТНОШЕНИИ

ПРЕДСТАВИТЕЛЕЙ ЛГБ-СООБЩЕСТВА (НЕ СОСТОЯЩИХ В

ГРАЖДАНСКОМ СОЮЗЕ)

Гендерная идентичность – двухступенчатый подход (а + b)

a. Ваш пол, зарегистрированный при рождении. Один ответ

1. Женский

2. Мужской

3. Предпочитаю не отвечать

b. Как Вы себя идентифицируете в настоящее время? Один ответ

1. Женщина/девочка

2. Мужчина/мальчик

3. Транс-женщина/девочка

4. Транс-мужчина/мальчик

5. Небинарное лицо или другой вариант

6. Предпочитаю не отвечать

3ье ОБСЛЕДОВАНИЕ ТРУДОВОЙ ДИСКРИМИНАЦИИ В ОТНОШЕНИИ

ТРАНС- И НЕБИНАРНЫХ ЛИЦ (в стадии осуществления)

a. Пол (зарегистрированный в свидетельстве о рождении)

1. Женский

2. Мужской

b. Как Вы себя идентифицируете себя на сегодняшний день в плане гендерной

идентичности?

1. Женщина/девочка

2. Мужчина/мальчик

3. Транс-женщина/девочка

4. Транс-мужчина/мальчик

5. Небинарная гендерная идентичность

6. Предпочитаю не отвечать

c. Как Вы определяете свою гендерную идентичность на сегодняшний день?

Произвольный текст

Рабочий документ

15

__________________________________________________________________________

VII. Литература

76. De Rosa, E., de Martino, E., Scambia, F., Nur, N. (2022). Perspectives on LGBT+ working

lives: stakeholders, employers and LGBT+ people. RIEDS - Rivista Italiana di Economia,

Demografia e Statistica - The Italian Journal of Economic, Demographic and Statistical

Studies, SIEDS Societa' Italiana di Economia Demografia e Statistica, Vol. LXXVI-2 April-

June 2022, pp. 4-12. ISSN: 0035-6832.

77. De Rosa, E. (2022). Intersezionalità e discriminazioni LGBT+: paradigmi, concetti e

indicatori. AG AboutGender, vol. 11, no. 22, pp. 306-336, DOI: 10.15167/2279-

5057/AG2022.11.22.2024. ISSN 2279-5057.

78. De Rosa, E., De Vitiis, C., Inglese, F., Vitalini, A. (2020). Il Web-Respondent Driven

Sampling per lo studio della popolazione LGBT+. RIEDS - Rivista Italiana di Economia,

Demografia e Statistica, vol. LXXIV, no. 1 Gennaio-Marzo 2020, pp. 73-84. ISSN: 0035-

6832.

79. De Rosa, E., Inglese, F. (2018). Diseguaglianze e discriminazioni nei confronti delle persone

LGBT: quale contributo della statistica ufficiale?. RIEDS - Rivista Italiana di Economia,

Demografia e Statistica, vol. LXXII-4 Ottobre-Dicembre 2018, pp. 77-88. ISSN: 0035-6832.

80. European Commission, Directorate-General for Justice and Consumers. (2021). Guidelines

on improving the collection and use of equality data, Publications Office,

https://data.europa.eu/doi/10.2838/9725

81. FRA. (2019). Compendium of promising practices for equality data collection;

https://fra.europa.eu/en/promising-practices-list

82. Istat. (2013). Stereotipi, rinunce, discriminazioni di genere. Anno 2011;

https://www.istat.it/it/archivio/106599

83. Istat (2020). Diversity Management for LGBT+ Diversities in Enterprises and Desirable

Actions to Improve Inclusiveness at Work. Year 2019;

https://www.istat.it/it/files//2021/01/LGBT-Report.pdf

84. Istat (2022). Survey on Labour Discrimination toward LGBT+ People (in Civil Union or

formerly in union). Year 2020-2021;

https://www.istat.it/it/files//2022/05/REPORTDISCRIMINAZIONILGBT_2022_en.pdf

85. PRAIA Group. (2020). The Handbook on Governance Statistics;

https://unstats.un.org/unsd/statcom/51st-

session/documents/Handbook_on_GovernanceStatistics-Draft_for_global_consultation-

E.pdf

86. UNECE. (2019). In-Depth Review of Measuring Gender Identity. Conference of European

Statisticians, Paris; https://unece.org/sites/default/files/2021-01/In-

depth%20review%20of%20Measuring%20Gender%20Identity%20for%20bureau.pdf

Рабочий документ

16

Gender Identity Representation in Data Collection: New Approaches from Italy National Statistical Institute Italy (ISTAT)

EUGENIA DE ROSA , VALERIA de MARTINO, FRANCESCA SCAMBIA

10 - 12 May 2023

UNECE - Group of Experts on Gender Statistics

Outline

❑ISTAT-UNAR project “Labour discrimination against LGBT+ people and diversity policies in enterprises”

❑Mixed method (quantitative-qualitative, probabilistic and non-probabilistic samples)

❑Different surveys and different target groups of LGBT+ people based on respondents’ self-

identification

❑Development and collection of SOGIESC indicators and gender representation

beyond the binary

❑Participatory approach

Inter-institutional cooperation A multiperspective approach

Labour discrimination against LGBT+ people and diversity policies in enterprises: a multiperspective approach

3

✓ A collaboration agreement

between the Italian National

Statistical Institute (ISTAT)

and the National Anti-

Discrimination Office (UNAR)

✓ EU Funds

✓ Working Groups and

“Permanent consultation

Board for the promotion of

the rights and protection of

LGBT+ people”

Stakeholders

In-depth interviews

LGBT+ people CAWI surveys

addressed to different targets

Enterprises

Ad hoc module on Diversity Management in national surveys on

enterprises

Main results available to date:

https://www.istat.it/en/archivi

o/252737

Main results available to date:

https://www.istat.it/en/archivio/270

626

•

•

Provide insights on labour discrimination against LGBT+ people

3 CAWI surveys based on a self-amministrative web questionnaire – respondents’ self-identification as

LGBT+ people

Stepwise introduction of SOGIESC indicators

Different surveys for different targets within the LGBT+ population

•

Surveys targeted at LGBT+ people

Survey of individuals who

were or had been in a Civil Union

(same- sex couples, over 21,000 people)

carried out in 2020- 2021

Survey of LGB people who have never been in a

civil union, through an experimentation of an

advanced snowball sampling technique (RDS

- respondent driven sampling) and convenience

sample

carried out in 2022 (January-

May)

Survey with a non- probabilistic

sample, on trans and non-binary

people

currently in progress

4

❑Developing and testing SOGIESC (sexual orientation, gender identity, gender expression

and sex characteristics) indicators in official statistics

❑ Meetings with LGBT+ associations of the “Permanent consultation table for the promotion of

the rights and protection of LGBT+ people”

❑ Different “Working Groups” of the project set up by some LGBT+ associations of

the Permanent Table

❑ Involvement of experts, academics and other stakeholders

❑ Respondents’ remarks in open questions

❑ Istat discussed and shared this experience within the “Equality Data Subgroup” of the EC

“High-Level Group on Non-Discrimination, Equality and Diversity” and within the “Task Team

on Non-discrimination and Equality” of the Praia Group

…. conceptions and terminological debate rapidly evolve5

SOGIESC indicators

1st- SURVEY ON LABOUR DISCRIMINATION AGAINST LGBT+ PEOPLE

(currently or formerly in Civil Union) – 2020/2021

Target population: all resident individuals (over 21,000) who, as of 1 January 2020, were or had been in

Civil Union (same-sex couples - Law 76/20 May 2016)

✓ 95.2% people in civil union or formerly in union who live in Italy declare a homosexual or bisexual orientation. As for

the remaining 4.8%: 0.2% asexual orientation, 1.3% another orientation and the remaining prefer not to answer

✓ Among those who declare a homosexual or bisexual orientation - 65.2% gay, 28.9% lesbians, 4.2% bisexual women

and 1.7% bisexual men

➢ Question to identify the “core” population:

What is your current sexual orientation?

Single response

• Homosexual

• Bisexual

• Asexual

• Heterosexual

• Other

• Prefer not to say

Being a CAWI a tooltip was made available to specify the meaning of the response’ items In order to be consistent with the Italian law on Civil Unions the definition provided for sexual orientation was based on sex and not on gender. A question on gender identity was not included

Sex (currently in the civil registry): Single response

1. Female

2. Male

✓ The target population has specific features: over 50 (43.6%), majority men, outed, a stable working life

✓ Data are not representative of the entire LGBT+ population

1st- SURVEY ON LABOUR DISCRIMINATION AGAINST LGBT+ PEOPLE

(currently or formerly in Civil Union) – 2020/2021

7

✓ Gender segregation in employment

✓ Multiple discrimination, more often against lesbian, bisexual women and young people

0

10

20

30

40

50

60

Lesbian Gay Bisexual (female)

Bisexual (male)

SEXUAL ORIENTATION Total

18-34 years

35-49 years

50 and over

▪ Target population: LGB people who have never been in Civil Union

✓ The RDS snowball technique was not working properly and we shifted to a convenience sample

✓ Results cannot be taken as complementary to the survey on people in Civil Union, and they exclusively

refer to LGB people who decided to take part in this second survey

✓ More than a thousand of LGB people were interviewed. Main results will be published on 15 May 2023

8

2nd - SURVEY ON LABOUR DISCRIMINATION AGAINST LGB PEOPLE

(NOT in Civil Union) - 2022

➢ Question to identify the target population:

What is your current sexual orientation?

Single response

• Homosexual

• Bisexual

• Other

• Prefer not to say

Response items are reduced because survey intended to:

- capture a complementary segment to the previous survey on homosexual and bisexual persons in Civil Union

- facilitate the implementation of the snowball technique

2nd - SURVEY ON LABOUR DISCRIMINATION AGAINST LGB PEOPLE

(NOT in Civil Union) - 2022

9

Sex (currently in the civil registry):

Single response

1. Female

2. Male

a. What sex were you assigned at birth?:

Single response

1. Female

2. Male

3. Prefer not to say

b. How do you currently identify yourself?

Single response

1. Woman/girl

2. Man/boy

3. Trans woman/girl

4. Trans man/boy

5. Non-binary or other

6. Prefer not to say

✓ Data could be analyzed, not only for gay, lesbian and bisexual people, but also for cisgender

men/women and non-binary people (question's test albeit on a very specific target)

Gender identity – two step approach (a+b)

10

Target population: individuals aged 18 and over, who usually live in Italy, and whose gender

identity does not correspond to the sex they were assigned at birth

✓ All the target population was labelled as “trans and non-binary persons”: as not all the non-

binary people identity themselves in the umbrella label trans or transgender

✓ Not only indicators to survey sex and gender identity, also complementary open questions to

better understand

✓ Gender expression, coming out/visibility milestones, gender affirmation, use of pronouns

✓ Going beyond a binary logic means to include open questions, to set sexual orientation

indicators not only sticking to a definition based on gender.

✓ Definitions cannot be given for granted

3rd - SURVEY ON LABOUR DISCRIMINATION AGAINST TRANS AND NON-BINARY

PEOPLE (in progress)

11

➢ Question to identify the target population:

GENDER IDENTITY

Sex (registered in the birth certificate)

1. Female

2. Male

Thinking about your gender identity, how do you currently identify yourself?

1. Woman/girl

2. Man/boy

3. Trans woman/Trans girl

4. Trans man/Trans boy

5. Non-binary gender identity

6. Prefer not to say (out of questionnaire)

How would you currently define your gender identity? ______

3rd - SURVEY ON LABOUR DISCRIMINATION AGAINST TRANS AND NON-BINARY

PEOPLE (in progress)

Sex at birth= gender identity then out of questionnaire Sex at birth ≠ gender identity then labelled trans people

Labelled non-binary people: according to the questionnaire tooltip

Sex at birth ≠ gender identity then labelled trans people

A question with fixed items necessary to define the target population is followed by an open question to give a chance of freely describing co-existing/fluctuant dimentions and progressive coming out (e.g. Non-binary woman)

A - How would you currently describe yourself on the basis of your appearance, your clothing, the way you move, the way you speak?

B- A person's appearance, clothing, mannerism or the way of speaking can influence how others

describe them. How do you think others would currently describe you?

1. Very feminine

2. Feminine

3. Rather feminine

4. Neither feminine nor masculine

5. Rather masculine

6. Masculine

7. Very masculine

8. I don't know

9. I prefer not to say

12

GENDER EXPRESSION

Important indicator because discrimination is often based on gender expression. Associations agreed on its relevance

To date, have you taken/performed any of the following actions? Mark all that apply

1. Dressing according to your gender identity

2. Using a name that is consistent with your gender identity (without registry adjustment)

3. Using a pronoun consistent with your gender identity

4. Using neutral nouns/neutral pronouns

5. Taking a psychological course aimed at the diagnosis of gender dysphoria

6. Taking hormone therapies

7. Changing your master data

8. Sex reassignment

9. Gender affirmation surgery

10. Other (Specify )

11. I prefer not to say

What pronouns do you want others to refer you with? Mark all that apply

1. Masculine

2. Feminine

3. With a u or schwa (ə)

4. No pronouns in particular

5. Other (Specify…………)

13

GENDER AFFIRMING PROCESS

According to the most recent approach against medicalization, we do not refer to transition rather to gender affirming process

Item 3 is specific for the italian language which marks with the final letter masculine and feminine

How would you currently define yourself with reference to your sexual orientation?

1. Heterosexual

2. Gay

3. Lesbian

4. Bisexual

5. Pansexual

6. Queer

7. Asexual

8. Other (Specify ___________)

9. I don’t know

10. I prefer not to say

14

SEXUAL ORIENTATION

Identity dimension is followed by a question on attraction They are aimed at exploring different profiles Sexual orientation is referred to gender and not to sex

Some people are born with variations in sex characteristics, sexual characteristics (such as sexual

anatomy, reproductive organs and/or chromosomal arrangements) that do not strictly belong to male or

female categories or belong simultaneously to both. This condition is known as "intersexuality"

Have you ever been diagnosed with an "intersex condition," either at birth or later?

1. Yes

2. No

3. I prefer not to say

Would you describe yourself as an intersex person?

1. Yes

2. No

3. I prefer not to say

15

SEX CARACTHERISTICS

Two different dimensions, medicalization and perception; this phenomena is not well known in Italy

➢ The Istat-Unar project - an opportunity to include SOGIESC indicators in the official statistics

➢ Initial challenge - the need to combine the multiplicity and mutability of terms used by

people to define complex aspects such as their (gender) identity, and the need for official

statistics to synthesize and classify

➢ Exchange with LGBT+ associations, LGBT+ people, and with academics

➢ For the first time in Italian official statistics indicators of gender identity (two-step

approach) - not taking the equivalence between biological sex and gender for granted. This

gives visibility to a plurality of gender representations beyond the binary one

➢ In the survey on trans and non-binary people the issue of gender has been addressed in its

complexity, trying to overcome mere legislative and medical definitions

➢ Providing a range of indicators: sex at birth, gender identity and also other aspects (e.g. gender

expression, intersexuality, actions to affirm gender) and other identity characteristics (e.g.

citizenship, religious belief)

16

Lessons learned and future prospects

➢ Finding the right trade-off between number of indicators and the questionnaire burden is

necessary, especially when shifting from the collection of data on a specific topic or population

(e.g. LGBT+ people) to general surveys on the whole population

➢ Further experimentation should be carried out using survey techniques other than the self-

administered web-based questionnaire and in surveys targeting the whole population (e.g.

the Istat Pilot Survey on Discrimination, 2023)

➢ The focus on representing different ways of acting, signifying and expressing gender in data

collection practices implies going beyond exclusively binary statistics. Non-binarism is a

challenge but also gives a critical perspective to the processes of reasoning, valuing,

measuring, and comparing through numbers

➢ The inclusion of self-indentification questions on multiple characteristics in general surveys

on population enables to investigate social position, direct and indirect discrimination,

multiple and intersectional discrimination and inequalities, and their development in time

➢ Comparable and harmonized data at an international level is a goal to be achieved

17

Lessons learned and future prospects

References

Report Istat-UNAR (2022), Survey on Labour Discrimination toward LGBT+ People (in Civil Union or formerly in union).

Year 2020-2021(https://www.istat.it/it/files//2022/05/REPORTDISCRIMINAZIONILGBT_2022_en.pdf)

De Rosa, E., Martino, V., Scambia, F., Nur, N. (2022), Perspectives on LGBT+ working lives: stakeholders, employers and

LGBT+ people, RIEDS - Rivista Italiana di Economia, Demografia e Statistica - The Italian Journal of Economic,

Demographic and Statistical Studies, SIEDS Societa' Italiana di Economia Demografia e Statistica, vol. 76(2), pages 4-12,

April-Jun.

De Rosa, E. (2022), Intersezionalità e discriminazioni LGBT+: paradigmi, concetti e indicatori, About Gender, numero

monografico “Fare intersezionalità: luoghi inesplorati”

De Rosa, E. e Inglese, F. (2020), Studi Lgbt+, mixed methods e intersezionalità: percorsi di ricerca sulle discriminazioni lavorative, About Gender, vol. 9 N° 17, pp. 142-17

De Rosa, E., De Vitiis, C., Inglese, F., Vitalini, A. (2020), Il web-respondent driven sampling per lo studio della popolazione

LGBT+, Rivista Italiana di Economia, Demografia e Statistica, vol. LXXIV, n.1, pp.73-84

Report Istat-UNAR (2020), Diversity Management for LGBT+ Diversities in Enterprises and Desirable Actions to Improve

Inclusiveness at Work. Year 2019 (https://www.istat.it/it/files//2021/01/LGBT-Report.pdf)

De Rosa, E. e Inglese, F. (2018), Disuguaglianze e discriminazioni nei confronti delle persone Lgbt: quale contributo della

statistica ufficiale, Rivista italiana di economia demografia e statistica Sieds, vol. LXXII, n. 4, pp. 77-88

Thank you! Eugenia De Rosa; [email protected]

Valeria de Martino; [email protected]

Francesca Scambia; [email protected]

CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH

WORKSHOP OF EXPERT ON

GENDER STATISTICS

Geneva, 10-12 May 2023

Istat | Directorate for studies and enhancement of social and demographic statistics by theme and for information requirements of the National Recovery and Resilience Plan

MARIA CLELIA ROMANO, CAROLINA FACIONI, ANNA VILLA

o The relevance of a gender-based approach

o Relationships between population and

environment: the Istat main sources

o Concerns for environmental issues

o Eco-friendly behaviours

o Some remarks

2

Outline

CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

The relevance of a gender-based approach

o The lack of data is a key challenge to overcome also in developed

countries

o Citizens and households are one of the major source of

environmental pressure in modern societies. Studying citizens'

sensitivity to environmental issues, the factors related to it and how

it is put into practice, is very important for environmental

sustainability policies

o Long commitment at Istat to measure the attitudes of citizens

towards the environment

3 CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

o Link between gender and environmental statistics: not yet fully explored

o There are some international initiatives aimed to develop gender-

disaggregation of environmental data especially in developing countries

where data availability is more limited

t

Population and environment: the Istat main sources

o Survey on Household energy consumption: a comprehensive picture of energy consumption and energy

characteristics of the residential sector. Household interview: results are not suitable for a gender analysis

4

Multipurpose survey Aspects of daily life

Carried out yearly since 1993

Important source to monitor the changes in everyday life, also with regard to environmental issues

Since 2012 several environmental issues are included in the questionnaires:

o individual level (14+): satisfaction with the environmental situation of the neighborhood (air, water,

noise…) and opinions on the landscape degradation; Concerns about environmental issues;

Transport/mobility habits; Eco-friendly behaviours

o household level: Electricity and gas (quality of supply services); Water (quality of supply services,

mineral water consumption…); Waste (separate collection/composting).

CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

Concerns for environmental issues

Climate change and air pollution at

the top of environmental concerns

(more than 50% of the population)

No significant gender

differences in the perception of

environmental problems

Youngsters up to 34: biodiversity

loss, forest destruction and

depletion of natural resources

Over 50: hydrogeological

instability and soil pollution

Gender differences wider among

youngest people: young women

are more worried

POPULATION AGED 14 AND OVER BY ENVIRONMENTAL CONCERN AND GENDER. YEAR 2021.

PERCENTAGES.

5

0

10

20

30

40

50

60

M F

CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

Concerns for environmental issues

Citizens express concern about

various environmental problems:

the majority of men and women

(59.5%, 60.4%) indicates 5

problems

Some gender differences by age:

o Women under 55 more often

than men in the same age class

indicate 5 issues of concerns

o Young women are again the

most worried group: 65.3% of

women aged 14-24 years

indicate 5 environmental

problems (compared to 59% of

peers)

POPULATION AGED 14 AND OVER WHO INDICATED 5 ENVIRONMENTAL CONCERNS BY GENDER AND

AGE CLASS. YEAR 2021. PERCENTAGES.

6

59,0 59,2 59,9

64,5

55,8 59,5

65,3 63,4 62,7 64,7

52,1

60,4

0,0

10,0

20,0

30,0

40,0

50,0

60,0

70,0

14-24 25-34 35-54 55-64 65+ All

M F

CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

Concerns for environmental issues

Environmental concerns grow with

increasing educational level

o 1.5% of graduates do not

express any concern against

4.5% of those with at most a

lower secondary level of

education

o 74% of graduates indicate five

problems against 50.6% of less

educated people

Territorial differences emerge

between people living in the cities

and in small municipalities, and

across the North and South of Italy

POPULATION AGED 14 AND OVER BY NUMBER OF ENVIRONMENTAL CONCERNS AND EDUCATIONAL

LEVEL. YEAR 2021. PERCENTAGES.

7

1,5 4,8 4,3

7,1 8,3

74

2,2

7,6 6,4

9,1 9,9

64,8

4,5

12,4 9,5

12,5 10,6

50,6

0

10

20

30

40

50

60

70

80

No problem One Two Three Four Five

University degree and beyond Upper-secondary school diploma

Up to lower secondary school diploma

CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

Environmental concerns in brief

o In short, women and men share environmental concerns

without significant differences. Even dynamics of concerns over

time (2012-2021) are similar for men and women

o However young and adult women indicate more concerns than

men

o Girls aged between 14 and 24 appear to be particularly sensitive

to environmental issues

o The widespread of the concerns changes according to territory

and, mainly, educational level

o But only climate change and air pollution worry the majority of

the population, the other problems are indicated by a minority of

citizens. We still need to build environmental awareness

8 CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

tre

Eco-friendly behaviours

In 2021 most citizens were careful

not to waste energy (67.6%) or

water (65.9%). 49.6% never adopts

noisy driving behaviour, 37.1%

reads the labels of food products

and 24.4% buys zero km products

Women show more often and a

higher number of eco-sustainable

behaviours

PERSONS AGED 14 AND OVER BY ENVIRONMENT-FRIENDLY BEHAVIOURS USUALLY ADOPTED AND

GENDER. YEAR 2021. PERCENTAGE

9 CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

F 18.4

M 13.7

5 F 12.6

M 16.6

0

4,6 5,3

-2,5

12,3

3,1

3,3 4,9

0

10

20

30

40

50

60

70

80

Not to waste energy

Not to waste water

Watching out for noisy behavior

Reading food labels before

buying

Buying products at

zero km

Choosing alternative transport

Buying organic foods/products

All F M

Eco-friendly behaviours

Gender differences in all age

groups

Young girls (14-24) are more

respectful of natural resources like

water (+7.4 points) and energy

(+9.5 points)

Greater sensitivity of girls also in

spending behaviours, such as

reading product labels before

buying them (+7.5 points) or

buying organic food and products

(+5.8 points)

Eco-friendly behaviours are more

widespread among those aged 35

and over (except for the use of

alternative means of transport)

PERSONS AGED 14-24 BY ENVIRONMENT-FRIENDLY BEHAVIOURS USUALLY ADOPTED AND GENDER.

YEAR 2021. PERCENTAGE

10

46,7 43,0

37,3

23,0 22,8

14,9

9,6

54,1 52,5

42,2

30,5 25,5

16,5 15,4

0,0

10,0

20,0

30,0

40,0

50,0

60,0

Being careful not to waste

water

Being careful not to waste

energy

Watching out for noisy behavior

Reading food labels before buying them

Choosing alternative means of

transport to private ones

Buying products at

zero kilometer

Buying organic foods and products

M F

CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

Reference person/partner

0

10

20

30

40

50

60

70

80

Being careful not to waste

energy

Being careful not to waste

water

Watching out for noisy behavior

Reading food labels before buying them

Buying products at

zero kilometer

Buying organic foods and products

Choosing alternative means of

transport to private ones

M F

Son/Daughter

0

10

20

30

40

50

60

70

80

Being careful not to waste

energy

Being careful not to waste

water

Watching out for noisy behavior

Reading food labels before buying them

Buying products at

zero kilometer

Buying organic

foods and products

Choosing alternative means of

transport to private ones

M F

Eco-friendly behaviours

CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

PERSONS AGED 14 AND OVER BY ENVIRONMENT-FRIENDLY BEHAVIOURS USUALLY ADOPTED, GENDER AND HOUSEHOLD POSITION. YEAR 2021. PERCENTAGE

11

Eco-friendly behaviours: sustainable mobility

In 2021 about 30 millions of

people moved every day to study

or work.

Employed women go to work on

foot more often than employed

men (27.6% against 17.4%) and

use public transport more,

regardless of the distance (6.3%

against 4.9%)

But 84.6% of employed people

use private means, with not

significant gender differences

More sustainable mobility also

among female students

EMPLOYED PEOPLE AGED 18 AND OVER WHO USE PUBLIC MEANS OF TRANSPORT BY DISTANCE

FROM THE WORKPLACE AND GENDER. YEAR 2021. PERCENTAGES.

12

4,4 3,6

8,9

4,9

6,5

4,3

13,4

6,3

5,2

3,8

10,0

5,4

0

2

4

6

8

10

12

14

16

Same municipality where he/she lives

Same province where he/she lives

Greater distance All distances

M F All

CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

Eco-friendly behaviours: sustainable mobility

Sustainable and shared mobility:

carpooling is more used by men

(9.1% against 6%), especially if

employed (8.9% compared to

5.2% of women)

In 2021, 1.3% of the population

used bike sharing: 1.6% of men,

2.2% of employed men

1.2% of the adult population used

car sharing services (1.6% among

men)

PEOPLE AGED 18 AND OVER WHO REACH WORK/SCHOOL WITH COLLEAGUES/SCHOOLMATES BY

CONDITION AND GENDER. YEAR 2021. PERCENTAGES.

13

8,9

11,8

9,1

5,2

12,8

6,0

7,3

12,3

7,3

0

2

4

6

8

10

12

14

Employed Students All

M F All

CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

Eco-friendly behaviours and sustainable mobility in brief

o As other studies claim, women engage more in pro-environmental behavior

o Gender differences are clear in all age classes, especially among the youngsters

o There are still groups of population where environmental sensitivity must be built also through awareness-

raising policies

14 CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

o One third of population usually does not pay

attention to waste energy or water

o Young people less frequently adopt pro-environment

behaviours. So it is very important to invest in order

to increase their awareness of environmental

emergencies

o Too many people use private cars to reach the

workplace, still too few people use public transport.

Shared mobility is not yet widespread

o The data confirm the importance to study the interaction between gender and environmental sustainability

o Women can contribute to raising awareness of the new generations within the household and train them to

respect the environment

o The next steps include: i) multivariate analysis for a more complete investigation of the factors that shape

eco-friendly behaviours and ii) intra-household analysis for studying the sensitivity to environmental issues

among family members and the relationships between parents’ and children’s behaviours

o Male and female different commuting habits induce further analysis to shape mobility policies more

adequate and careful to citizens needs

o Istat aims to invest more in exploiting the available data and producing new ones according to a gender-

based approach. More in general, statistical offices can play an important role to support policymakers in

identifying key actors and factors that can foster the creation of an environmental culture

Some remarks

15 CITIZENS’ ATTITUDES AND BEHAVIOURS IN ENVIRONMENTAL MATTERS: A GENDER-BASED APPROACH | M.C. Romano

Thank you [email protected]

*Prepared by Alessandra Capobianchi, Maria Giuseppina Muratore, Claudia Villante.

NOTE: The designations employed in this document do not imply the expression of any opinion whatsoever on the part

of the Secretariat of the United Nations concerning the legal status of any country, territory, city or area or of its

authorities, or concerning the delimitation of its frontiers or boundaries.

Economic Commission for Europe

Conference of European Statisticians

Group of Experts on Gender Statistics Geneva, Switzerland, 10-12 May 2023

Item E of the provisional agenda

Measuring violence against women

Using big data to study violence against women and girls and its challenges online

Note by ISTAT (Italian National Institute of Statistics) *

Abstract

Social networks are the "modern agoras” and, during the pandemic, it has become a

social space where violence against women increased by moving from real to virtual.

Social media are the new source of data for understanding, along with other data

sources, the changing nature of gender-based violence. Finding methodologies and

techniques able to use these new sources of Big Data is essential to understanding

and monitoring that phenomenon. Observing what is happening on social media, and

measuring the sentiment of online conversations, however, is not the only goal.

Indeed, we know that it is violence against women itself that has become virtual

(online violence) and finds its channel of dissemination in social media. ISTAT has

piloted an Index related to gender-based violence and a sentiment and emotion

analysis based on social media content (Twitter, Facebook, Instagram, and web

newspaper post) aimed at measuring how the users of social react to the violence

against women and/or generates discussions around it. In addition, the adopted

method helps to understand which are the messages that unleash the discussion, by

providing a map of the topics (such as body-shaming, femicide, and rape) where the

sentiment is most solicited (both positive and negative). The results of the

experimentation are furthermore compared to another source of data, coming from

the national helpline against violence and stalking against women (1522) that ISTAT

has recently adopted (from 2020) as a new and timely administrative source of data

in order to gather information during the pandemic period. The piloted method can

help researchers in monitoring the future challenges of violence against women and

girls online and opening a new way aimed at integrating different sources of data, as

our study has attempted to provide.

Working paper 19

Distr.: General

01 May 2023

English

Working paper 19

2

I. On line Gender Based Violence, cyber-violence: different names and new tools of analysis

1. Online gender based violence is not a new phenomenon. In its 2018’s report, Mapping study

on cyber -violence, Cybercrime Convention Committee (T-CY)1, even acknowledging the

challenge of the definition due to its ongoing modification, reached consensus on using

“cyber-violence” as the most concise term to be used consistently throughout the study,

defining it as follows: Cyber-violence is the use of computer systems to cause, facilitate, or

threaten violence against individuals that results in, or is likely to result in, physical, sexual,

psychological or economic harm or suffering and may include the exploitation of the

individual’s circumstances, characteristics or vulnerabilities (…) Considering “cyber” as an

adaptation of the “cyber” context of the definition of violence against women of Article 3 of

the Istanbul Convention. ‘… the landscape of gender-based violence has been transformed

… [but] rather than there being a dramatic reduction in violence against women, … the

challenges have become more complex, the resistance to change deeper, the backlash against

the empowerment of women more blatant and the methods used to uphold the status quo

more sophisticated and insidious’2.

2. An extensive literature, over the last decades on this topic has been animated the discussion

on how to define violence against women and girls (VAWG) and, consequentially, how to

study and measure it. Based on 2022’s EIGE report3 there is not commonly accepted

definition of online violence against women’ (European Commission, Advisory Committee

on Equal Opportunities for Women and Men, 2020), but it is possible to identify some

characteristics which helps to define its boundaries and mapping the forms. We can

furthermore considers on line or cyber VAWG “as a burgeoning phenomenon on a global

scale: an emerging new dimension of gender-based violence that is likely to result in physi-

cal, sexual or psychological harm or suffering to women and girls”4. On line VAWG is

often referred to as a new form of violence, grounded in the increased use of new digital

technologies and maximised by the constant connectivity of Web 2.0. Undoubtedly this form

of new violence against women and men which is grounded in the increased use of new

digital technologies, exploded during the pandemic. The Covid-19 pandemic deepened

economic and social stress, enforcing restricted movement and social isolation measures,

resulting in increased risks of gender-based violence, particularly in the domestic context.

ISTAT has monitored this period using data of national helpline against violence against

women and stalking (1522)5. During Covid-19, violence against women manifested in

different, extended forms, including domestic violence, and online and ICT-facilitated

violence. The growing reach of the internet, the rapid spread of mobile information and

communications technologies (ICTs), and the wide diffusion of social media have presented

1 Cybercrime Convention Committee (T-CY) Working Group on cyberbullying and other forms of online

violence, especially against women and children, Mapping study on cyberviolence with recommendations

adopted by the T-CY on 9 July 2018. Strasbourg, France 2 Cyber Violence Against Women nd Girls: A World-Wide Wake-Up Call. Report By The UN Broadband

Commission For Digital Development Working Group On Broadband And Gender. Available online at:

https://en.unesco.org/sites/default/files/genderreport2015final.pdf 3 EIGE, 2022 ibidem 4 EIGE, 2022 Ibidem p.36. 5 ISTAT, Gender-based violence in the time of COVID-19: calls to the 1522 helpline, 30 June 2020,

https://www.istat.it/en/archivio/245001

Working paper 19

3

new opportunities and enabled various efforts to address online VAWG. The ICT is only a

new tool to perpetuate gender-based violence against women and girls which takes place in a

context of widespread systemic gender-based discrimination. In fact, internet is not just a

communication tool. It is actually an environment that forces society to reorganize itself in

relation to ICT technologies. We are constantly connected and most of our daily actions

happen through the internet. The distinction between reality and virtual reality is obsolete,

from a relationship point of view and from a violence point of view.

3. As the UN Women paper suggests6 to prevent and respond to online and technology

facilitated VAWG, new methods and tools of analysis has to be set up in order to catch this

“extension” of the gender-based violence from offline to online, such us using Big Data.

4. ISTAT, is now pursuing the goal of producing data on the phenomenon by pursuing two

parallel paths:

i. Introducing new questions specifically dedicated to cybercrime and online violence

in the Citizen’s Security Survey (currently underway).

ii. By analysing Big Data, with reference to interactions on social media. In this paper

only this second statistic is briefly reported.

II. Using social media to detect online violence against women and girls

5. Social networks are the "modern agoras" to be adequately studied in order to draw from

them valuable information that can provide insights to understand and predict the evolution

of social phenomena; even the topic of gender violence must be studied from this

perspective, observing what the users themselves think, say and share, with the aim of being

able to monitor, constantly, what are the evolutions of the new forms of online violence and,

more generally, of the sentiments and opinions spread on the web. It is precisely in this

second area of research that Sentiment Analysis goes to position itself, that is, among those

tools that make it possible to observe how people's opinions react to violence against

women. As social media advances, discussions of gender-based violence provide a central

source of information on the topic, making it possible to observe over time whether the ways

in which gender-based violence is discussed have changed: have levels of aggression in

conversations increased? What is the "climate" of these conversations? Is it positive?

Negative? Angry? Sad?

6. Since 2021, Istat has launched an experimental study on social messages for the study of the

phenomenon of gender-based violence, with the aim of providing, in real-time, information

on the social image of gender-based violence and gender stereotypes, adding useful data to

the VAWG information framework for monitoring and the mode of communication

conveyed on social channels on this issue. Among the various sources of Big Data available,

social media (Van den Brakel J., Söhler E., Daas P., Buelens B., 2017; Daas P., Puts M.,

2014) is a particularly useful information resource for analyzing and monitoring the

phenomenon of gender-based violence and the presence of stereotypes. Sentiment analysis

6 UN WOMEN, Accelerating efforts to tackle online and technology-facilitated violence against women and

girls, 2022 https://www.unwomen.org/en/digital-library/publications/2022/10/accelerating-efforts-to-tackle-

online-and-technology-facilitated-violence-against-women-and-girls

Working paper 19

4

and emotion detection constitute a methodology of analysis that returns information on the

themes and ways in which social media represent the phenomenon and, at the same time,

allows its understanding to be extended to its digital dimensions (such as cyber-bullying and

cyber-violence). The digitization of this phenomenon, like other social issues, requires the

development of new methodologies and tools that can capture it. Through a process of

capturing content conveyed by social, based on the presence of at least one word belonging

to a set of filter words prepared by domain experts, it is possible to collect and observe the

opinions and content of "posts," tweets and messages on which sentiment and emotion

analysis can be applied. In this way, it is possible to measure how the phenomenon is

represented on social media and whether they are used to counter, condemn and isolate the

culture of gender stereotypes and gender-based violence or whether, on the contrary, by

exalting the negative side of its use, they contribute to its persistence and spread. The

following graphs show how much and how the phenomenon of gender-based violence is

represented on social content (Twitter, Facebook, Instagram), through time graphs related to

1 year of observation. The period considered is from November 1, 2021 to November 30,

2022

7. In addition, a Topic Analysis on particular issues discussed on social media shows that

social media amplifies typical effects that we can observe on VAWG, such as, for example,

secondary victimization. Through machine learning techniques, it is possible to perform

thematic analysis on content from news, facts, and events in the virtual context in which they

are commented on and reprocessed within digital communities.

A. First Results

8. Data from November 1, 2021 to November 30, 2022 covered a total of 1,231,385 messages

on gender-based violence: predominantly 1,012,110 Twitters, followed by 103,442 Web

messages, 55,358 Instagram posts, and 35,205 Facebook comments. Tweets absorb 82.2

percent of the messages.

Working paper 19

5

Chart 1 - Daily trend on social media of journalistic (info) and non-journalistic (no-info) posts (Period

November 1, 2021 - November 30, 2022) Absolute values

Source, Istat 2023

9. The social message does not always contain personal opinions or experiences. Sometimes

the content of a message represents objective information (references to news, promotional

tweets, informative tweets, etc.). The algorithms and the methodology adopted makes it

possible to isolate these messages from those that are not strictly related to the dissemination

of the news alone. A binary classifier was then used to identify this type of content in order

to select purely journalistic content (info) from non-journalistic content (no-info). In terms

of analysis, this selection allows the analysis to focus on comments rather than content.

Information content shows peaks of higher circulation on the days around November 25 in

2021 and 2022, reflecting the "pull" function that these moments of commemoration have on

public opinion.

10. Below is a chart of calls received by the 1522 toll-free number against gender-based

violence and stalking. As can also be seen from this graph, the celebratory event of

November 25 also represents a peak moment for those who contact the helpline for support

(both for themselves and others), despite the fact that the pandemic effect has generated an

overall increase in call traffic. In general, therefore, awareness-raising campaigns, on the one

hand and calls from institutions and civil society, on the other, provide a central opportunity

to stimulate discussion and raise social awareness about the phenomenon of gender-based

violence in our country.

0

5000

10000

15000

20000

25000

30000

35000

NOINFO INFO

Working paper 19

6

Chart 2 - Daily trend of total calls to 1522 (Period January 1, 2018 - September 25, 2022) Absolute

values

Source, ISTAT, Violence against women, 2023 https://www.istat.it/en/violence-against-women

1. Sentiment and emotional analysis

11. Chart 3, which reports sentiment analysis, shows that the peak on November 25 stimulates

positive and neutral feelings and reactions. In contrast, the peak related to interactions

generated by a news event related to rape (August 2022) is characterized by negative

comments and, as better highlighted in the following graphs, emotionally related to anger,

fear and sadness.

Chart 3- Daily trend on social media of sentiment expressed in posts with positive, negative, neutral

comments (Period November 1, 2021 - November 30, 2022). Absolute Values

0

5000

10000

15000

20000

25000

POS NEU NEG

0

100

200

300

400

500

600

700

0 1 JA N 2 0 1 8

2 7 JA N 2 0 1 8

2 2 -f e b -1 8

2 0 -m

ar -1 8

1 5 -a p r- 1 8

1 1 M A Y2 0 1 8

0 6 JU N 2 0 1 8

0 2 JU L2 0 1 8

2 8 JU L2 0 1 8

2 3 A U G 2 0 1 8

1 8 SE P 2 0 1 8

1 4 O C T2 0 1 8

0 9 -n o v- 1 8

0 5 D EC

2 0 1 8

3 1 D EC

2 0 1 8

2 6 JA N 2 0 1 9

2 1 -f e b -1 9

1 9 -m

ar -1 9

1 4 -a p r- 1 9

1 0 M A Y2 0 1 9

0 5 JU N 2 0 1 9

0 1 JU L2 0 1 9

2 7 JU L2 0 1 9

2 2 A U G 2 0 1 9

1 7 SE P 2 0 1 9

1 3 O C T2 0 1 9

0 8 -n o v- 1 9

0 4 D EC

2 0 1 9

3 0 D EC

2 0 1 9

2 5 JA N 2 0 2 0

2 0 -f e b -2 0

1 7 -m

ar -2 0

1 2 -a p r- 2 0

0 8 M A Y2 0 2 0

0 3 JU N 2 0 2 0

2 9 JU N 2 0 2 0

2 5 JU L2 0 2 0

2 0 A U G 2 0 2 0

1 5 SE P 2 0 2 0

1 1 O C T2 0 2 0

0 6 -n o v- 2 0

0 2 D EC

2 0 2 0

2 8 D EC

2 0 2 0

2 3 JA N 2 0 2 1

1 8 -f e b -2 1

1 6 -m

ar -2 1

1 1 -a p r- 2 1

0 7 M A Y2 0 2 1

0 2 JU N 2 0 2 1

2 8 JU N 2 0 2 1

2 4 JU L2 0 2 1

1 9 A U G 2 0 2 1

1 4 SE P 2 0 2 1

1 0 O C T2 0 2 1

0 5 -n o v- 2 1

0 1 D EC

2 0 2 1

2 7 D EC

2 0 2 1

2 2 JA N 2 0 2 2

1 7 -f e b -2 2

1 5 -m

ar -2 2

1 0 -a p r- 2 2

0 6 M A Y2 0 2 2

0 1 JU N 2 0 2 2

2 7 JU N 2 0 2 2

2 3 JU L2 0 2 2

1 8 A U G 2 0 2 2

1 3 SE P 2 0 2 2

Working paper 19

7

Source: ISTAT, 2023

12. Comments with positive sentiment corresponding to the November peaks report neutral

emotions, surprise and joy; those in August (related to an episode of rape) anger and again

neutral emotions. Social users who comment positively on both the news event and the

November 25 celebratory event thus seem to refrain from providing additional emotional

connotations. The annual anniversary that invites public opinion to reflect on gender-based

violence connotes social interactions with a strong component of surprise.

Chart 4 - Monthly trend of emotional reactions on social media containing posts with negative comments (Period

November 1, 2021 - November 30, 2022) Absolute Values

Source: ISTAT, 2023

13. In this chart it is anger and fear characterize the interactions. In the case of comments on the

August act of violence (rape), a component of fear also emerges, which has never been

recorded so strongly in previous months.

III. Conclusion and next steps

14. The analysis based on the evidences of messages (buzzes) collected, underlines the

importance of social media contents to better understand and monitor the opinion regarding

some topics and to detect some emerging issues. More specifically,

i. Focusing relevant themes linked to GBV such as femicide, body shaming, rape and

sexual violence;

ii. Following up awareness campaign linked to helpline 1522;

iii. Profiling type of users (institutions and politicians, non-governmental associations,

influencers, common users).

15. Moreover, in addition to the analyses of some specific “peaks” that drive the socials, a new

category of analysis has been recently introduced: hate speech and outrage. Specifically,

social content has been reclassified into two polarities: one group of words and expressions

overall related to hate speech against VAWG and a second composed of a set of words and

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

LOVE JOY NEUTRAL FEAR ANGER SURPRISE SADNESS

Working paper 19

8

expressions related to the feeling of "outrage." This polarity allows observing the volume of

hate speech and aggression about VAWG events and those instead of awareness raising

(expressed in indignation, compassion, rebellion) to VAWG events.

15. Finally, a Topic Analysis is underway to understand the extent to which specific episodes of

rape convictions and legal sentences are linked to a social media “secondary victimization”

reactions, when social media reacts with comments accusing victims of being partly

responsible for the acts of violence they have suffered. This analysis is closely related to the

level of acceptability of VAWG. In this sense, it will be very interesting and challenging, at

the same time, to analyse online VAWG and the analysis of gender stereotypes and the

social image of VAWG (a survey this one that ISTAT conducts every 3 years). Therefore,

the next methodological step of this experimental statistics through Big Data will be to find

the integration between different data sources, also to give linearity and robustness to the

Big Data treatment process.

*Подготовлена Алессандрой Капобьянки, Марией Джузеппиной Мураторе и Клаудией Вилланте.

ПРИМЕЧАНИЕ: Обозначения, используемые в настоящем документе, не подразумевают выражения какого-

либо мнения со стороны Секретариата Организации Объединенных Наций относительно правового статуса той

или иной страны, территории, города или района или их властей, или относительно делимитации их границ или

рубежей.

Европейская экономическая комиссия

Конференция европейских статистиков

Группа экспертов по гендерной статистике Женева, Швейцария, 10–12 мая 2023 года

Пункт E предварительной повестки дня

Измерение показателей насилия в отношении женщин

Использование больших данных для изучения онлайн-насилия в отношении женщин и девочек и связанных с ним проблем

Записка ISTAT (Итальянского национального института статистики) *

Резюме

Социальные сети — это «современные агоры», которые во время пандемии

стали общественным пространством, где насилие в отношении женщин

усилилось за счет перехода от реальной формы к виртуальной. Социальные

сети представляют собой новый источник данных, с помощь которого, наряду с

другими источниками данных, можно понять изменения в характере

гендерного насилия. Поиск методологий и методов, позволяющих

использовать эти новые источники больших данных, имеет важное значение

для понимания и мониторинга этого явления. Однако наблюдение за тем, что

происходит в социальных сетях, и измерение настроений в разговорах в сети не

является единственной целью. Нам несомненно известно, что именно само

насилие в отношении женщин как таковое стало виртуальным (онлайн-

насилие) и нашло канал распространения в социальных сетях. ISTAT в

экспериментальном порядке разработал индекс гендерного насилия и провел

анализ настроений и эмоций на основе контента социальных сетей (Twitter,

Facebook, Instagram и публикации в интернет-изданиях) для количественной

оценки того, как пользователи социальных сетей реагируют на насилие в

отношении женщин и/или какие дискуссии развертываются на эту тему. Кроме

того, используемый метод помогает понять, какие сообщения дают толчок

дискуссии, посредством картирования тем (таких как бодишейминг, фемицид и

изнасилование), вызывающих наиболее активное проявление эмоций (как

позитивных, так и негативных). Кроме того, результаты эксперимента

сравниваются с другим источником данных, который обеспечивает

Рабочий документ 19

Distr.: General

08 мая 2023

English

Рабочий документ 19

2

национальная «горячая линия» по борьбе с преследованием и насилием в

отношении женщин (1522), недавно (с 2020 года) созданная ISTAT в качестве

нового и оперативного административного источника данных с целью сбора

информации в период пандемии. Апробированный метод может помочь

исследователям в отслеживании будущих проблем, связанных с онлайн-

насилием в отношении женщин и девочек и открыть новые возможности для

интеграции различных источников данных, что и было предпринято в рамках

нашего исследования.

I. Гендерное онлайн-насилие, кибернасилие: разные названия и новые инструменты анализа

1. Гендерное онлайн-насилие — явление не новое. В своем датированном 2018 годом

докладе «Картографическое исследование кибернасилия» Комитет Конвенции о

киберпреступности (T-CY)1, признавая сложность определения этого понятия по

причине его непрерывного изменения, достиг консенсуса в отношении применения

слова «кибернасилие» в качестве наиболее краткого термина, который будет

последовательно использоваться в рамках всего исследования, определив его

следующим образом: Кибернасилие — это использование компьютерных систем для

причинения насилия, содействия совершению насилия и угрозы совершения насилия в

отношении людей, которое приводит или может привести к физическому,

сексуальному, психологическому или экономическому ущербу или страданиям и

может включать в себя использование обстоятельств, индивидуальных особенностей

или уязвимых сторон человека (…) Рассматривая слово «кибер» как адаптацию

определения насилия в отношении женщин, содержащегося в Статье 3 Стамбульской

конвенции, к кибер-контексту, «… ситуация в области гендерного насилия

изменилась,… [но] вместо резкого сокращения насилия в отношении женщин…

проблемы усложнились, сопротивление переменам усилилось, противодействие

расширению прав и возможностей женщин стало еще более вопиющим, а методы,

используемые для поддержания статус-кво, более изощренными и коварными»2.

2. В обширной литературе по этой теме в течение последних десятилетий велась

дискуссия о том, какое определение сформулировать для насилия в отношении

женщин и девочек (НОЖД) и, следовательно, как его изучать и измерять. Согласно

отчету ЕИГР за 2022 год3 общепринятого определения онлайн-насилия в отношении

женщин не существует (Европейская комиссия, Консультативный комитет по

вопросам равенства возможностей женщин и мужчин, 2020 год), но можно выделить

некоторые характеристики, которые помогают определить его границы и картировать

1 Cybercrime Convention Committee (T-CY) Working Group on cyberbullying and other forms of online

violence, especially against women and children, Mapping study on cyberviolence with recommendations

adopted by the T-CY on 9 July 2018. Strasbourg, France 2 Cyber Violence Against Women nd Girls: A World-Wide Wake-Up Call. Report By The UN Broadband

Commission For Digital Development Working Group On Broadband And Gender. Available online at:

https://en.unesco.org/sites/default/files/genderreport2015final.pdf 3 EIGE, 2022 ibidem

Рабочий документ 19

3

его формы. Кроме того, мы можем рассматривать онлайн- или кибер-НОЖД «как

активно развивающееся явление глобального масштаба: формирующееся новое

измерение гендерного насилия, которое может причинить физический, сексуальный

или психологический вред или страдания женщинам и девочкам»4. Онлайн-НОЖД

часто называют новой формой насилия, основанной на более широком использовании

новых цифровых технологий и усиленной до максимума за счет постоянного

подключения к Веб 2.0. Вне всякого сомнения, эта новая форма насилия в отношении

женщин и мужчин, основанная на более широком использовании новых цифровых

технологий, начала активно развиваться во время пандемии. Пандемия COVID-19

усилила экономическую и социальную напряженность в результате введения

ограничений на перемещение и мер социальной изоляции, что привело к повышению

риска гендерного насилия, особенно в бытовом контексте. В этот период ISTAT

осуществлял мониторинг, используя данные национальной горячей линии по

вопросам насилия в отношении женщин и преследований (1522)5. Во время пандемии

COVID-19 насилие в отношении женщин проявлялось в различных и расширенных

формах, включая домашнее насилие, а также насилие в интернете и с использованием

ИКТ. Растущий охват интернета, быстрое распространение мобильных

информационно- коммуникационных технологий (ИКТ) и широкое распространение

социальных сетей открыли новые возможности и способствовали различным усилиям

по борьбе с НОЖД в интернете. ИКТ — это всего лишь новый инструмент,

используемый для закрепления гендерного насилия в отношении женщин и девочек,

которое имеет место в контексте широко распространенной системной гендерной

дискриминации. На самом деле интернет — это не просто средство общения. В

действительности он представляет собой среду, которая вынуждает общество

реорганизовываться во взаимосвязи с ИКТ-технологиями. Мы постоянно находимся в

сети, и большая часть наших повседневных действий совершается через интернет.

Различие между реальностью и виртуальной реальностью устарело с точки зрения

отношений и с точки зрения насилия.

3. Как предлагается в документе Структуры «ООН-женщины»6, для предотвращения и

реагирования на НОЖД, совершаемого в интернете и с использованием современных

технологий, необходимо внедрить новые методы и инструменты анализа, чтобы

уловить это «расширение» гендерного насилия из офлайна в онлайн, например, с

использованием больших данных.

4. ISTAT в настоящее время преследует цель получения данных об этом явлении,

работая параллельно по двум направлениям:

i. Включение новых вопросов, непосредственно касающихся кибер-

преступности и онлайн-насилия, в Обследование безопасности граждан

(проводимое в настоящее время).

4 EIGE, 2022 Ibidem p.36. 5 ISTAT, Gender-based violence in the time of COVID-19: calls to the 1522 helpline, 30 June 2020,

https://www.istat.it/en/archivio/245001 6 UN WOMEN, Accelerating efforts to tackle online and technology-facilitated violence against women and

girls, 2022 https://www.unwomen.org/en/digital-library/publications/2022/10/accelerating-efforts-to-tackle-

online-and-technology-facilitated-violence-against-women-and-girls

Рабочий документ 19

4

ii. Анализ больших данных в контексте взаимодействия в социальных сетях. В

настоящем документе кратко представлена статистика только по второму

направлению.

II. Использование социальных сетей для выявления онлайн- насилия в отношении женщин и девочек

5. Социальные сети — это «современные агоры», которые необходимо надлежащим

образом изучить, чтобы извлечь ценную информацию, которая может обеспечить

понимание и возможность прогнозирования эволюции социальных явлений; даже

тему гендерного насилия необходимо изучать с этой точки зрения, наблюдая за тем,

что сами пользователи думают, говорят и каким контентом делятся с другими

пользователями, чтобы иметь возможность постоянно отслеживать эволюцию новых

форм онлайн-насилия и, в более общем плане, настроений и мнений,

распространяемых в сети. Анализ настроений позиционирует себя именно в этой

второй области исследований, то есть среди тех инструментов, которые позволяют

наблюдать, как общественное мнение реагируют на насилие в отношении женщин. По

мере развития социальных сетей дискуссии по вопросам гендерного насилия

становятся основным источником информации по этой теме, позволяя наблюдать за

происходящими с течением времени изменениями в характере обсуждения гендерного

насилия: повысился ли уровень агрессии в разговорах? Каковы «умонастроения» в

этих разговорах? Позитивные? Негативные? Гневные? Печальные?

6. С 2021 года ISTAT приступил к проведению экспериментального исследования

сообщений в социальных сетях для изучения феномена гендерного насилия с целью

получения в режиме реального времени информации о социальном образе гендерного

насилия и гендерных стереотипах, что позволяет дополнить информационную базу по

НОЖД полезными данными для мониторинга, и о модальности коммуникации в

социальных сетях по этому вопросу. Среди различных доступных источников

больших данных социальные сети (Van den Brakel J., Söhler E., Daas P., Buelens B.,

2017; Daas P., Puts M., 2014) представляют собой особо полезный информационный

ресурс для анализа и мониторинга феномена гендерного насилия и наличия

стереотипов. Анализ настроений и выявление эмоций составляют методологию

анализа, которая позволяет получить информацию о темах и способах представления

этого явления в социальных сетях и одновременно позволяет расширить его

понимание на цифровые аспекты (таких как кибербуллинг и кибернасилие).

Цифровизация этого явления, как и других социальных проблем, обуславливает

необходимость разработки новых методологий и инструментов, способных его

выявлять. Посредством процесса захвата контента, передаваемого через социальные

сети, на основе наличия хотя бы одного слова, принадлежащего к набору фильтров,

составленному экспертами в предметной области, можно собирать и отслеживать

мнения и содержание постов, твитов и сообщений, к которым можно применить

анализ настроений и эмоций. Таким образом, можно определить, как это явление

представлено в социальных сетях и используются ли они для противодействия,

осуждения и изоляции культуры гендерных стереотипов и гендерного насилия или,

наоборот, превознося негативные аспекты его использования, они способствуют его

сохранению и распространению. На приведенных ниже графиках показано, насколько

и каким образом феномен гендерного насилия представлен в социальном контенте

Рабочий документ 19

5

(Twitter, Facebook, Instagram), на основе временных графиков, охватывающих 1 год

наблюдений. Рассматриваемый период – с 1 ноября 2021 года по 30 ноября 2022 года.

7. Кроме того, тематический анализ конкретных вопросов, обсуждаемых в социальных

сетях, показывает, что социальные сети усиливают типичные эффекты, которые мы

можем наблюдать в отношении НОЖД, такие как, например, вторичная

виктимизация. С помощью методов машинного обучения можно выполнять

тематический анализ контента из новостей, фактов и событий в виртуальном

контексте, в котором они комментируются и повторно обрабатываются в цифровых

сообществах.

A. Первые результаты

8. Данные за период с 1 ноября 2021 года по 30 ноября 2022 года охватывают в общей

сложности 1 231 385 сообщений о гендерном насилии: основную долю – 1 012 110 –

составляют сообщения в Twitter, за которыми следуют 103 442 сообщения в

интернете, 55 358 сообщений в Instagram и 35 205 комментариев в Facebook. На долю

твитов приходится 82,2% сообщений.

График 1. Динамика количества журналистских (информационных) и нежурналистских

(неинформационных) постов в социальных сетях (за период с 1 ноября 2021 года по 30 ноября

2022 года) Абсолютные значения

Источник, ISTAT, 2023 год

9. Сообщение в социальной сети не всегда отражает личное мнение или опыт. Иногда

содержание сообщения представляет собой объективную информацию (ссылки на

новости, рекламные твиты, информационные твиты и т. д.). Используемые алгоритмы

и методология позволяют отделить эти сообщения от тех, которые в строгом смысле

не связаны с распространением только новостей. Затем был использован бинарный

классификатор для идентификации этого типа контента, чтобы отделить чисто

0

5000

10000

15000

20000

25000

30000

35000

NOINFO INFO

Рабочий документ 19

6

журналистский контент (информационный) от нежурналистского контента

(неинформационного). С точки зрения анализа этот отбор позволяет сосредоточить

анализ на комментариях, а не на контенте. Информационный контент демонстрирует

пиковые значения распространения в дни, близкие к 25 ноября, в 2021 и 2022 годах,

что отражает «притягивающую» функцию этих памятных дат для общественного

мнения.

10. Ниже приведен график звонков, поступивших на бесплатный номер 1522 горячей

линии по вопросам борьбы с гендерным насилием и преследованием. Как видно из

этого графика, отмечаемая 25 ноября дата также является пиковым моментом для тех,

кто обращается на горячую линию за поддержкой (как для себя, так и для других),

несмотря на то, что последствия пандемии привели к общему увеличению числа

обращений. Таким образом, в целом кампании по повышению осведомленности, с

одной стороны, и призывы со стороны учреждений и гражданского общества, с

другой, обеспечивают основную возможность для стимулирования дискуссии и

повышения осведомленности общества о феномене гендерного насилия в нашей

стране.

График 2. Ежедневная динамика общего количества звонков на номер 1522 (за период с 1

января 2018 года по 25 сентября 2022 года) Абсолютные значения

Источник, ISTAT, Насилие в отношении женщин, 2023 год. https://www.istat.it/en/violence-

against-women

1. Анализ настроений и эмоций

11. График 3, на котором представлен анализ настроений, показывает, что пик,

наблюдаемый 25 ноября, стимулирует позитивные и нейтральные чувства и реакции.

Напротив, пик, который относится к информационному обмену, вызванному

новостным событием, связанным с изнасилованием (август 2022 года),

характеризуется негативными комментариями и, как лучше показано на следующих

графиках, эмоционально связан с гневом, страхом и печалью.

0

100

200

300

400

500

600

700

0 1 JA N 2 0 1 8

2 7 JA N 2 0 1 8

2 2 -f e b -1 8

2 0 -m

ar -1 8

1 5 -a p r- 1 8

1 1 M A Y2 0 1 8

0 6 JU N 2 0 1 8

0 2 JU L2 0 1 8

2 8 JU L2 0 1 8

2 3 A U G 2 0 1 8

1 8 SE P 2 0 1 8

1 4 O C T2 0 1 8

0 9 -n o v- 1 8

0 5 D EC

2 0 1 8

3 1 D EC

2 0 1 8

2 6 JA N 2 0 1 9

2 1 -f e b -1 9

1 9 -m

ar -1 9

1 4 -a p r- 1 9

1 0 M A Y2 0 1 9

0 5 JU N 2 0 1 9

0 1 JU L2 0 1 9

2 7 JU L2 0 1 9

2 2 A U G 2 0 1 9

1 7 SE P 2 0 1 9

1 3 O C T2 0 1 9

0 8 -n o v- 1 9

0 4 D EC

2 0 1 9

3 0 D EC

2 0 1 9

2 5 JA N 2 0 2 0

2 0 -f e b -2 0

1 7 -m

ar -2 0

1 2 -a p r- 2 0

0 8 M A Y2 0 2 0

0 3 JU N 2 0 2 0

2 9 JU N 2 0 2 0

2 5 JU L2 0 2 0

2 0 A U G 2 0 2 0

1 5 SE P 2 0 2 0

1 1 O C T2 0 2 0

0 6 -n o v- 2 0

0 2 D EC

2 0 2 0

2 8 D EC

2 0 2 0

2 3 JA N 2 0 2 1

1 8 -f e b -2 1

1 6 -m

ar -2 1

1 1 -a p r- 2 1

0 7 M A Y2 0 2 1

0 2 JU N 2 0 2 1

2 8 JU N 2 0 2 1

2 4 JU L2 0 2 1

1 9 A U G 2 0 2 1

1 4 SE P 2 0 2 1

1 0 O C T2 0 2 1

0 5 -n o v- 2 1

0 1 D EC

2 0 2 1

2 7 D EC

2 0 2 1

2 2 JA N 2 0 2 2

1 7 -f e b -2 2

1 5 -m

ar -2 2

1 0 -a p r- 2 2

0 6 M A Y2 0 2 2

0 1 JU N 2 0 2 2

2 7 JU N 2 0 2 2

2 3 JU L2 0 2 2

1 8 A U G 2 0 2 2

1 3 SE P 2 0 2 2

Рабочий документ 19

7

График 3. Ежедневная динамика настроений в социальных сетях, выраженная в постах с

позитивными, негативными и нейтральными комментариями (за период с 1 ноября 2021 года по

30 ноября 2022 года). Абсолютные значения

Источник, ISTAT, 2023 год

12. Комментарии с положительным настроем, соответствующие ноябрьским пикам,

выражают нейтральные эмоции, удивление и радость; августовские же (связанные со

случаем изнасилования) выражают гнев и снова нейтральные эмоции. Таким образом,

пользователи социальных сетей, которые положительно комментируют как новость,

так и событие, отмечаемое 25 ноября, воздерживаются от дополнительных

эмоциональных коннотаций. Ежегодно отмечаемая дата, которая побуждает

общественное мнение задуматься о гендерном насилии, ассоциируется с социальным

взаимодействием, в котором присутствует сильный компонент удивления.

График 4. Ежемесячная динамика эмоциональных реакций в социальных сетях, содержащих посты с

негативными комментариями (за период с 1 ноября 2021 года по 30 ноября 2022 года) Абсолютные

значения

0

5000

10000

15000

20000

25000

POS NEU NEG