NTTS SATELLITE EVENT ON MULTIINDICATOR SYSTEMS AND PARTIALLY

NTTS SATELLITE EVENT ON MULTI-INDICATOR SYSTEMS AND PARTIALLY ORDERED SETS MULTIDIMENSIONAL WELL-BEING AND DEPRIVATION HOW TO AGGREGATE AND HOW TO SYNTHESIZE COMPLEX SOCIOECONOMIC PHENOMENA? A COMPARISON OF CFA, AF AND POSET APPROACHES CHRISTIAN SUTER, KATIA IGLESIAS, TUGCE BEYCAN UNIVERSITY OF NEUCH TEL, SWITZERLAND Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

FRAMEWORK • Develop new synthetic measures of wellbeing (WB) – For the Swiss Social Report – To monitor long-term social changes and WB Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

FRAMEWORK • Swiss Social Report – Started in 2000’s – Edited every four years – Broad coverage of the various life domains – Based on 5 domains: 1) Production and consumption of social goods, 2) Cultural diversity, 3) Social integration and social cohesion, 4) Political participation and integration and 5) Environment and ecological sustainability Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

FRAMEWORK • Domain 1: Production and consumption of social goods – A multidimensional concept of WB: 4 dimensions • • Christian Suter Educational and occupational positioning Employment situation Financial situation Work-life balance Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

DATA • Swiss Household Panel (SHP) – Longitudinal survey of households and people living in Switzerland. – Conducted on a yearly basis since 1999 – Sub-sample: • Active household members • Aged between 18 and 65 years • Year 2013 Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

DATA • Swiss Household Panel (SHP) – Dim 1: Conciliation professional/private life • Interference work/private activities • Exhausted after work • Difficult to disconnect from work – Dim 2: Financial situation of household (HH) • • Christian Suter Satisfaction with financial situation Financial situation manageable Assessment of income and expenses Arrears of payments Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

DATA • Swiss Household Panel (SHP) – Dim 3: Employment situation • Job security • Risk of unemployment • Satisfaction with working conditions – Dim 4: Educational and occupational position • Highest level of education achieved • Swiss socio-professional category of main job (CSP) Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

METHODOLOGIES • Ordinal data • In a top-down approach • Compare methods to synthesize information – Confirmatory factor analysis (CFA) – Alkire & Foster counting approach (AF) – Posetic approach (POSET) Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

METHODOLOGIES • CFA: Test the adequacy data-theoretical framework Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

METHODOLOGIES • AF: Compute the number of deprived people in society from multidimensional perspective – Dual cut-off • Determine ‘deprived or not-deprived’ within each dimension • Identify ‘deprived’ across dimensions Difficult to disconnect from work • 2 scores possible: – Deprived – Not deprived 0, 1 -4, 6 -10 Assessment of income and expenses save money, use what they earn, eat in its assess and savings or get into debts Job security secure, almost secure, quite insecure, insecure Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

METHODOLOGIES • POSET: based on the dual cutoff of AF • No aggregation • Each single profile = combination of WB scores – Difficult to disconnect from work 3: 0 2: 1 -4 1: 6 -10 133 123 132 333 223 323 232 213 332 313 222 312 331 231 321 – Assessment of income and expenses 3: save money 2: use what they earn 1: eat in its assess and savings or get into debts – Job security 3: secure or almost secure 2: quite insecure 1: insecure Christian Suter 113 122 112 131 212 121 221 311 211 111 Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

METHODOLOGIES • POSET: based on the dual cutoff of AF • No aggregation • Each single profile = combination of WB scores • Relationship between two objects 133 – Can be ordered 123 132 333/233/123… – Incomparable objects 113 123 and 332 – Score of degree of deprivation ranges between 0 and 1 333 223 323 232 213 122 112 332 313 222 131 322 312 212 121 221 331 231 321 311 211 111 Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

MODELS TESTED • AF: same first cutoff – 12 items: • 4/12 (model 0, 4/12) • 6/12 (model 0, 6/12) – 4 items: 2/4 • Model 1: 1) Interference work/private activities, 2) Job security, 3) Assessment of income and expenses, 4) Educational level • Model 2: 1) Difficult to disconnect from work, 2) Risk of unemployment, 3) Satisfaction with financial situation, and 4) Occupational level Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

MODELS TESTED • POSET: same first cutoff – 4 items: 2/4 • Model 1: 1) Interference work/private activities, 2) Job security, 3) Assessment of income and expenses, and 4) Educational level • Model 2: 1) Difficult to disconnect from work, 2) Risk of unemployment, 3) Satisfaction with financial situation, and 4) Occupational level Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

RESULTS: CFA Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

RESULTS: AF 12 items % of deprived people 4 items 6/12 4/12 Model 1 Model 2 4, 5% 16, 6% 8, 8% 13, 0% • % of deprived persons depends on the cutoff (6/12 or 4/12) and on the number of variables (4 or 12) included in the model Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

RESULTS: AF • Same theoretical framework, same pool of variables, same first cutoff, but: – Different models do not identify the same people as deprived • 25% deprived in model 0 6/12 model and not deprived in the model 1 • 61% deprived in model 1 are not identified as deprived by the model 0, 6/12 • 53% deprived in model 1 are not deprived in model 2 • 68% deprived in model 2 are not identified as deprived in model 1 Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

RESULTS: POSET 1 96. 0% less than 0. 125; 2 with 90. 4% less than 0. 125 • as noted for the AF models, they do not identify the same people as deprived Iglesias et al. , University of Neuchâtel WB: How to aggregate? 12. 07. 2016

RESULTS: CFA-AF (MODEL 1) Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

RESULTS: CFA-POSET • As expected, we found a high association between the different CFA and POSET deprivation degree – Correlation between |0, 33| and |0, 58| • The size of the correlation depends on – Correlation between the variables of the model and the latent factor Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

RESULTS: AF-POSET • POSET identifies the deprived units based on the dual cutoff of AF – Expect AF and POSET to identify the same deprived persons (score=1) and that all non-deprived persons with POSET (score=0) will also be non-deprived with AF – Confirmed except for one POSET profile (2/1/2/2) identified as deprived, despite the fact that it does not include multidimensional deprivation • Comparing model 1 to model 2, as within AF and within POSET – Do not identify the same people as deprived Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

DISCUSSION • Based on the same theoretical framework and same pool of variables: – The different synthetic indicators correlate within and between methodologies – But the individuals identified as multidimensionally deprived differ within and between methodologies • It is not only a matter of difference du to methodology used, but more a question of – Implicit weights: Item selection – And number of item used Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

CONCLUSION • CFA: – To test the consistency of the data and theoretical framework – Not a single synthetic measure • AF – Simple and immediately understandable – High communicative power regarding dissemination – Not adapted for prevention • POSET – Based on the AF dual cutoff, but not easily communicable – Useful for prevention: identify the social groups which are at risk of multidimensional deprivation, i. e. , those who are not (yet) deprived but may eventually become deprived Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

SUGGESTION: COMBINE THE APPROACHES 3 steps: 1) CFA to test the consistency of the data and the 2) AF by dimension: count the number of item by dimensions are deprived Depending on the user of the synthetic measure: 3) theoretical framework o Non academic public: AF on the number of deprivation by dimension o Prevention: POSET on the number of deprivation by dimension Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017

THANK YOU FOR YOUR ATTENTION Christian Suter Multidimensional Well-Being and Deprivation: How to aggregate and how to synthesize? 17. 03. 2017