Well-Being Global Index: Assays of multivariate statistical - - PowerPoint PPT Presentation

Well-Being Global Index: Assays of multivariate statistical approaches

Maria do Carmo Botelho1 (maria.botelho@iscte-iul.pt) Rosário Mauritti1 (rosario.mauritti@iscte-iul.pt) Nuno Nunes1 (nuno.nunes@iscte-iul.pt) Daniela Craveiro2 (daniela.craveiro@iscte-iul.pt) Paulo Neto3 (neto@uevora.pt)

1ISCTE-University Institute of Lisbon, CIES-IUL 2ISCTE-University Institute of Lisbon, CIS-IUL, CIES-IUL 3Évora University, CICS.NOVA UEVORA, UMPP

, CIES-IUL

Introduction

This research is part of a larger project about well-being inequalities in Europe, named “Territories of Inequality and Well-being”, sponsored by Francisco Manuel dos Santos Foundation. This investigation is based in a multidimensional understanding of social inequality and well-being, explored within and across OECD European countries. To measure well-being, our orientation is the OECD Better Life Initiative and the quality of life results, available by Eurostat. The OECD’s How's Life (2011, 2013, 2015b, 2017a), allows the measure and the knowledge of the Europeans well-being, based on a multi-dimensional framework, following the “Beyond the GDP” agenda.

3

Project Goals

Understanding everyday life in Europe

Analysing territorial inequalities and their relationships with the well- being of Europeans Place-based monitoring of the impacts of European public policies and interventions Construction of a system of indicators to improve data availability, and analysis on the multidimensional relationships

Aims

The main goal of this presentation is to discuss the statistical construction of indicators of well-being, based on microdata bases. The specific objectives for this presentation are:

• 1. Present previous work about structural configurations of well-being inequalities

across individuals and countries, based on microdata from the European Social Survey (ESS), 2016;

• 2. Discuss data limitations and the undertaken statistical analyses;
• 3. Identify and select appropriate indicators for two dimensions of well-being:

environment and housing, using European Quality of Life Survey (EQLS) microdata, also from 2016;

• 4. Create two composite dimensions, reducing multidimensionality and construct

each dimension with multivariate statistical analysis.

1. Structural configurations of well-being inequalities across individuals and countries

The use of European Social Survey (ESS) microdata Based on OCDE well-being framework, we identify several key indicators for measuring nine dimensions of well-being. Only two were

• mitted: Housing and Education.

The used questions were mostly Likert type items. Well-being indicators were normalised, using de min-max method (OECD, 2016), resulting values from zero to 10 in all dimensions. For dimensions with two or more indicators, the arithmetic mean was calculated.

1. Structural configurations of well-being inequalities across individuals and countries

The use of European Social Survey (ESS) microdata Social inequalities multidimensionality

• Distributional inequalities: Income and Education
• Categorical inequalities: Gender and Social Class(1)

(1)ACM typology (Costa et al., 2002), Socio-occupation indicator constructed on the basis of a

cross matrix of class locations formed by the ISCO08 occupations * employment status Entrepreneurs and executives (EE), Professionals and managers (PM), Self-employed (SE), Routine employees (RE) and Industrial workers (IW).

8

1. Structural configurations of well-being inequalities across individuals and countries

Well-being Global Volume (WBGV) Well-being profiles among Europeans Well-being profiles among countries Cluster analysis Multiple Regression Analysis Association measures

9

Well-being profiles

among europeans

Elite (38.0%; N=7331) WBGV=7.1 Lowest well-being (17.9%; N= 3452) WBGV=4.9

10

Individualist well-being (30.5%; N= 5898) WBGV=5.9 Insecure well-being (13.6%; N=2637) WBGV=5.7

Well-being profiles

among europeans

11

Nordic high-rank (WBGV=6.9) IC, NO, SW Southern Europe medium-rank (WBGV=6.2) FR, PT, SP Central Europe medium-rank (WBGV=6.4) AT, BE, CH, DE, FI, GB, NL

Well-being profiles

among countries

12

Eastern Europe low-rank (WBGV=5.8) EE, HU, IE, PL, SI Social disengagement low-rank (WBGV=5.5) CZ, IT, LI

Well-being profiles

among countries

Few indicators in each well-being dimensions The min-max methodology is not the best for these indicators The use of the arithmetic mean for the global volume can be debatable Difficulty in measuring well-being related with housing and environment The exclusive use of subjective indicators

2. Discuss data limitations and the undertaken statistical analyses

For each dimension, a systematization of the indicators used by OECD and Eurostat was made, and also a comparison to ESS indicators, when possible. Analysis of different microdata bases, available in the UK Data Service, possible due to the support of the INGRIG-2 program, with the visit to the Cathie Marsh Institute, University of Manchester. The analyzed databases were British Social Attitudes, Community Life Survey, Continuous Household Survey, EU-SILC, Understanding Society, European Value Study and the European Quality of Life Survey (EQLS). Identification and selection of indicators for the two dimensions of well-being (individual level and country level).

3. Environment and housing

The use of European Quality of Life Survey (EQLS) microdata

Environment

Study Dimension Indicators Measures Source

Eurostat Quality

• f Life
• 8. Natural

and living environment 8.1 Pollution (including noise)

Urban population exposure to air pollution by particulate matter (PM10) Quant Var: Micrograms per cubic meter European Environment Agency (EEA), yearly Perception of pollution, grime or other environmental problems EU-SILC, yearly; in the future, this variable would be collected within the 3-year rolling module on housing Noise from neighbours or from the street

8.2 Access to green and recreational spaces

Satisfaction with recreational and green areas EU-SILC 2013 ad hoc module on well-being

8.3 Landscape and built environment

Satisfaction with the living environment EU-SILC 2013 ad hoc module on well-being; next: EU-SILC 3-year rolling module on housing

OECD Environmental quality Air pollution

Population weighted average

• f annual concentrations of

particulate matters less than 2.5 microns in diameter (PM2.5) in the air Quant Var: Micrograms per cubic meter OECD calculations based on data from the Global Burden of Disease assessment (Brauer, M. et al. (2016) "Ambient Air Pollution Exposure Estimation for the Global Burden of Disease 2013." Environmental Science & Technology 50 (1), Pages 79-88).

Satisfaction with water quality

In the city or area where you live, are you satisfied

• r

dissatisfied with the quality

• f water?

Gallup World Poll

ESS Environmental concern Climate change

D24-How worried are you about climate change? 1-Not at all worried, 5- Extremely worried D21- D25

• ther

questions about climate change*

Measures and sources of indicators used by OECD, Eurostat Comparison with the chosen ESS indicators

Environment

Indicator Measure Values/categories Source Noise Neighbourhood problems: noise 1 - Major problems 2 - Moderate problems 3 - No problems EQLS individuals Air quality Neighbourhood problems: air quality Litter or rubbish Neighbourhood problems: litter or rubbish on the street Traffic Neighbourhood problems: heavy traffic Green areas How easy or difficult is your access to access to recreational

• r green areas

1 - Very difficult 2 - Rather difficult 3 - Rather easy 4 - Very easy Living environment* Satisfaction with the living environment High, medium, low 2013 EU-SILC module

• n subjective WB

Eurostat Satisfaction with water quality* In the city or area where you live, are you satisfied or dissatisfied with the quality of water? Gallup World Poll Pollution** Urban population exposure to air pollution by particulate matter (PM10) Quant Var: Micrograms per cubic meter European Environment Agency (EEA): Eurostat *Value by country **Objective measure; Value by country

Selected indicators

• f the

European Quality of Life Survey

Housing

Study Dimension Indicators Measures Source

Eurostat Quality of Life

• 1. Material

living conditions 1.3.2 Housing conditions

Structural problems of the dwelling EU-SILC Space in the dwelling (overcrowding/under-occupation) Satisfaction with accommodation EU-SILC, 2013 ad hoc module on well- being; in the future: EU-SILC 3-year rolling module on

OECD Housing Number of rooms per person

Rate (number of rooms divided by the number of people living in the dwelling) European Union Statistics on Income and Living Conditions (EU-SILC)

Dwellings without basic facilities

Percentage of the population living in a dwelling without indoor flushing toilet for the sole use of the household

Housing expenditure

Percentage of the household gross adjusted disposable income * OECD calculations based

• n

OECD National Accounts Database

ESS ____________ ____________

____________ ____________ *expenditure of households in housing and maintenance of the house, as defined in the SNA (P31CP040: Housing, water, electricity, gas and other fuels; P31CP050: Furnishings, households’ equipment and routine maintenance of the house). It includes actual and imputed rentals for housing, expenditure in maintenance and repair of the dwelling (including miscellaneous services), in water supply, electricity, gas and other fuels, as well as the expenditure in furniture and furnishings and households equipment, and goods and services for routine maintenance of the house as a percentage of the household gross adjusted disposable income. Data refer to the sum of households and non-profit institutions serving households (S14_S15)

Measures and sources of indicators used by OECD, Eurostat ESS doesn’t have similar questions for this dimension

Housing

*objective measure; Value by country Indicator Measure Values or categories Source Housing Conditions Number of rooms per person Rate (nº rooms /nº of people living in the dwelling) EQLS individuals Satisfaction with accommodation 1 very dissatisfied, 10 very satisfied Dwellings without basic facilities Shortage of space 1-Yes, 2-No Rot in windows, doors or floors Damp or leaks in walls or roof Lack of indoor flushing toilet Lack of bath or shower Lack of facilities (heating or cooling) Housing expenditure* Percentage of the household gross adjusted disposable income OECD calculations based

• n OECD National

Accounts Database

Selected indicators of the European Quality of Life Survey

19

Create the two composite dimensions, according to the theoretical framework. Aggregate the items, reduce multidimensionality and construct each dimension with multivariate statistical analysis In conformity with the objectives pursued and using a different methodology, a Categorical Principal Components Analysis (CatPCA) was applied. This allows a structural definition of the latent variables (dimensions) according to the inter-relations between the observed variables.

Environment and Housing

20 Indicators n % Missing

Y16_Q54a Neighbourhood problems: noise 1 Major problems 1824 5.9 2 Moderate problems 7900 25.7 3 No problems 21053 68.4 Total 30777 100.0 32 Y16_Q54b Neighbourhood problems: air quality 1 Major problems 1656 5.4 2 Moderate problems 6344 20.7 3 No problems 22707 73.9 Total 30706 100.0 103 Y16_Q54c Neighbourhood problems: litter or rubbish on the street 1 Major problems 1822 5.9 2 Moderate problems 6810 22.1 3 No problems 22131 71.9 Total 30763 100.0 46 Y16_Q54d Neighbourhood problems: heavy traffic 1 Major problems 2916 9.5 2 Moderate problems 7985 26.0 3 No problems 19861 64.6 Total 30763 100.0 46 Y16_Q56d Access to recreational or green areas (break in trend: not applicable not shown from 4th EQLS) 1 Very difficult 1018 3.4 2 Rather difficult 2642 8.7 3 Rather easy 11713 38.6 4 Very easy 14951 49.3 Total 30324 100.0 485

Opinions about neighbourhood problems and access to green areas

Environment

21 The access to green areas has a small loading is this dimension

Indicators Environment Neighbourhood problems: noise 0.807 Neighbourhood problems: heavy traffic 0.800 Neighbourhood problems: air quality 0.800 Neighbourhood problems: litter or rubbish on the street 0.735 Access to recreational or green areas 0.240 Cronbach's Alpha 0.756

Environment

(5 variables)

Environment

(4 variables)

Indicators Environment Neighbourhood problems: noise 0.811 Neighbourhood problems: heavy traffic 0.807 Neighbourhood problems: air quality 0.800 Neighbourhood problems: litter or rubbish on the street 0.732 Cronbach's Alpha 0.796

Excluding the access to green areas, the Cronbach’s Alpha and loadings are slightly higher.

22

Environment

(5 variables)

Environment

(4 variables)

23

n Min Max Mean SDev Number of rooms per person 30619 0.09 12 1.63 1.04 Satisfaction with accommodation (1 very dissatisfied, 10 very satisfied) 30777 1 10 7.69 1.93

Opinions about housing conditions Opinions about housing conditions and satisfaction with accommodation

Housing

Indicators n % Missing Y16_Q25a Shortage of space 1 Yes 5229 17.0 2 No 25543 83.0 Total 30772 100.0 37 Y16_Q25b Rot in windows, doors or floors 1 Yes 2421 7.9 2 No 28349 92.1 Total 30770 100.0 39 Y16_Q25c Damp or leaks in walls or roof 1 Yes 3931 12.8 2 No 26833 87.2 Total 30764 100.0 45 Y16_Q25d Lack of indoor flushing toilet 1 Yes 710 2.3 2 No 30075 97.7 Total 30785 100.0 24 Y16_Q25e Lack of bath or shower 1 Yes 767 2.5 2 No 30015 97.5 Total 30782 100.0 27 Y16_Q25f Lack of facilities (heating or cooling) to keep a comfortable temperature at home 1 Yes 1672 5.4 2 No 29100 94.6 Total 30771 100.0 38

24 The number of rooms per person has a small value of loading.

Housing

(8 variables)

Indicators Housing Lack of bath or shower 0.709 Lack of indoor flushing toilet 0.690 Lack of facilities (heating or cooling) 0.622 Rot in windows, doors or floors 0.580 Damp or leaks in walls or roof 0.554 Satisfaction with accommodation 0.482 Shortage of space 0.440 Number of rooms per person 0.239 Cronbach's Alpha 0.684

Housing

(7 variables)

Excluding the number

• f

rooms per person, the Cronbach’s Alpha and loadings are slightly higher.

Indicators Housing Lack of bath or shower 0.723 Lack of indoor flushing toilet 0.704 Lack of facilities (heating or cooling) 0.628 Rot in windows, doors or floors 0.585 Damp or leaks in walls or roof 0.555 Satisfaction with accommodation 0.472 Shortage of space 0.416 Cronbach's Alpha 0.692

25

Housing

(8 variables)

Housing

(7 variables)

26

Concluding remarks

The measurement of well-being is an important issue in the assessment

• f citizens' life progress.

The use of 'more detailed' data, such as microdata, can make it possible to measure the real well-being that is actually felt by citizens. A first approach was presented, with the use of European Social Survey microdata, following the methodology applied for the construction of the Better Life Index. Different limitations were detected, related to the nature of the data that were used. There was also difficulty in measuring the environment and housing dimensions, with the use of the microdata base.

27

Concluding remarks

A systematisation of well-being and quality of life indicators, has been presented, used the OECD and Eurostat framework. Some differences were found in the two dimensions analysed. Analysis of the environment and housing based on EQLS microdata base was proposed. For the construction of the two composite dimensions a Categorical Principal Components Analysis was used. It proved to be an adequate procedure for the data, although it was necessary to exclude some variables to improve the internal consistency

• f the construct.

The substantive results

• f

the well-being dimensions cause slight differences in the comparative analysis of countries.

28

Ongoing research

Two dimensions have been worked on in this presentation, but the eleven dimensions proposed by the OECD are being studied. The indicators and their data sources used by the OECD and Eurostat have already been compared and analysed. After this assay with indicators at individual level only, further analyses will be carried out including indicators at country level. In the next step of the analysis, a multilevel confirmatory factorial analysis will be carried

• ut,

after defining the variables that best contribute to the construction of the dimensions. In parallel, some additional inequality variables are also being defined, such as labour and regions.

Thank you

Maria do Carmo Botelho

maria.botelho@iscte-iul.pt

30

n % n % n % <= 5040 1006 9.6% 1137 10.5% 2144 10.1% 5041 - 8250 1058 10.1% 1152 10.7% 2209 10.4% 8251 - 9375 952 9.1% 1093 10.1% 2045 9.6% 9376 - 11187 1064 10.1% 1298 12.0% 2362 11.1% 11188 - 12388 1120 10.7% 1117 10.3% 2237 10.5% 12389 - 13428 971 9.3% 979 9.1% 1950 9.2% 13429 - 15416 1020 9.7% 955 8.8% 1975 9.3% 15417 - 17954 1225 11.7% 1132 10.5% 2357 11.1% 17955 - 26996 952 9.1% 944 10.5% 1896 8.9% >=26997 1119 10.7% 999 10.5% 2118 9.9% Total 10487 100.0% 10807 10.5% 21294 100.0% Basic 3246 25.9% 3182 10.5% 6428 25.2% Upper secondary 5307 42.4% 5277 10.5% 10585 41.5% Higher 3973 31.7% 4545 10.5% 8518 33.4% Total 12526 100.0% 13005 10.5% 25530 100.0% Entrepreneurs and executives (EE) 2295 19.0% 1370 10.5% 3665 15.1% Professionals and managers (PM) 3246 26.9% 4171 10.5% 7416 30.6% Self-employed (SE) 760 6.3% 592 10.5% 1352 5.6% Routine employees (RE) 2027 16.8% 4888 10.5% 6915 28.6% Industrial workers (IW) 3720 30.9% 1137 10.5% 4857 20.1% Total 12047 100.0% 12158 10.5% 24206 100.0% Total Gender 12526 49.1% 13005 10.5% 25530 100.0% Social Class (5 categories) Characterization Total Female Male Gender Equivalent income deciles EU € Level of education

Table A1. Sample description

n Minimum Maximum Mean

• Std. Dev.

Years of full-time education completed 25233 48 13.65 3.922 Equivalent income 1000€ 21294 2.57 82.60 13.97 8.394 Valid N (listwise) 21151

Table A2. Education and income sample description