Do Danes and Italians rate life satisfaction in the same way? Using vignettes to correct for individual-specific scale biases Viola Angelini 1 Danilo Cavapozzi 2 Luca Corazzini 2 Omar Paccagnella 2 1 University of Groningen and Netspar 2 University of Padua New Directions in Welfare 2011 OECD, Paris, 6-8 July 2011
Motivation In an ageing society, the well-being of the elderly is at the top of the European policy agenda
Motivation In an ageing society, the well-being of the elderly is at the top of the European policy agenda In economics measures of well-being have focused mainly on economic factors...
Motivation In an ageing society, the well-being of the elderly is at the top of the European policy agenda In economics measures of well-being have focused mainly on economic factors...... but money is not enough to make people happy
Motivation In an ageing society, the well-being of the elderly is at the top of the European policy agenda In economics measures of well-being have focused mainly on economic factors...... but money is not enough to make people happy Happiness (or life-satisfaction) seems to be a more accurate proxy of overall well-being than economic measures (Frey and Stutzer, 2002).
Motivation Are measures of happiness interpersonally comparable?
Motivation Are measures of happiness interpersonally comparable? Differential item functioning: inter-personal and inter-cultural variation in interpreting and using the response categories for the same question.
Motivation Are measures of happiness interpersonally comparable? Differential item functioning: inter-personal and inter-cultural variation in interpreting and using the response categories for the same question. The differences in reports of life satisfaction across seemingly similar countries [... ] appear implausibly large, and they raise additional doubts about the validity of global reports of subjective well-being, which may be susceptible to cultural differences in the norms that govern self-descriptions (Kahneman et al., 2004, AER).
This paper We use a vignettes methodology (King et al., 2004; Salomon et al., 2004; Bago d Uva et al., 2008; Kapteyn et al., 2007; van Soest et al., 2006; Kristensen and Johansson, 2008) to measure and correct the DIF bias in life satisfaction.
This paper We use a vignettes methodology (King et al., 2004; Salomon et al., 2004; Bago d Uva et al., 2008; Kapteyn et al., 2007; van Soest et al., 2006; Kristensen and Johansson, 2008) to measure and correct the DIF bias in life satisfaction. Individuals are presented with two categories of questions of life satisfaction:
This paper We use a vignettes methodology (King et al., 2004; Salomon et al., 2004; Bago d Uva et al., 2008; Kapteyn et al., 2007; van Soest et al., 2006; Kristensen and Johansson, 2008) to measure and correct the DIF bias in life satisfaction. Individuals are presented with two categories of questions of life satisfaction: Self-reported life satisfaction;
This paper We use a vignettes methodology (King et al., 2004; Salomon et al., 2004; Bago d Uva et al., 2008; Kapteyn et al., 2007; van Soest et al., 2006; Kristensen and Johansson, 2008) to measure and correct the DIF bias in life satisfaction. Individuals are presented with two categories of questions of life satisfaction: Self-reported life satisfaction; Evaluation of the life-satisfaction of hypothetical person(s) described in particular conditions and kept constant across respondents (anchoring vignettes).
This paper We use a vignettes methodology (King et al., 2004; Salomon et al., 2004; Bago d Uva et al., 2008; Kapteyn et al., 2007; van Soest et al., 2006; Kristensen and Johansson, 2008) to measure and correct the DIF bias in life satisfaction. Individuals are presented with two categories of questions of life satisfaction: Self-reported life satisfaction; Evaluation of the life-satisfaction of hypothetical person(s) described in particular conditions and kept constant across respondents (anchoring vignettes). By collecting individuals evaluation of the anchoring vignettes it is possible to filter out the level of self-reported life satisfaction from the DIF bias measured by vignettes and enhance the comparability of subjective assessments across individuals.
Data 2006 wave of the Survey of Health, Ageing and Retirement in Europe (SHARE).
Data 2006 wave of the Survey of Health, Ageing and Retirement in Europe (SHARE). SHARE is a multidisciplinary dataset that contains a large amount of information on both the economic and non economic conditions of individuals aged 50 and over.
Data 2006 wave of the Survey of Health, Ageing and Retirement in Europe (SHARE). SHARE is a multidisciplinary dataset that contains a large amount of information on both the economic and non economic conditions of individuals aged 50 and over. 5,606 individuals living in Sweden, Denmark, Germany, The Netherlands, Belgium, France, Spain, Italy, Poland and Czech Republic.
Data 2006 wave of the Survey of Health, Ageing and Retirement in Europe (SHARE). SHARE is a multidisciplinary dataset that contains a large amount of information on both the economic and non economic conditions of individuals aged 50 and over. 5,606 individuals living in Sweden, Denmark, Germany, The Netherlands, Belgium, France, Spain, Italy, Poland and Czech Republic. Question How satisfied are you with your life in general?
Data 2006 wave of the Survey of Health, Ageing and Retirement in Europe (SHARE). SHARE is a multidisciplinary dataset that contains a large amount of information on both the economic and non economic conditions of individuals aged 50 and over. 5,606 individuals living in Sweden, Denmark, Germany, The Netherlands, Belgium, France, Spain, Italy, Poland and Czech Republic. Question How satisfied are you with your life in general? 5-item scale: 1. Very Satisfied, 2. Satisfied, 3. Neither satisfied nor dissatisfied. 4. Dissatisfied. 5. Very dissatisfied.
Cross-country differences in life satisfaction
Vignettes on life-satisfaction - John Vignette 1 John is 63 years old. His wife died 2 years ago and he still spends a lot of time thinking about her. He has 4 children and 10 grandchildren who visit him regularly. John can make ends meet but has no money for extras such as expensive gifts to his grandchildren. He has had to stop working recently due to heart problems. He gets tired easily. Otherwise, he has no serious health conditions. How satisfied with his life do you think John is?
Vignettes on life-satisfaction - Carry Vignette 2 Carry is 72 years old and a widow. Her total after tax income is about e 1,100 per month. She owns the house she lives in and has a large circle of friends. She plays bridge twice a week and goes on vacation regularly with some friends. Lately she has been suffering from arthritis, which makes working in the house and garden painful. How satisfied with her life do you think Carry is?
Vignette ratings
The HOPIT model: the self-assessment component The life satisfaction perceived by individual i is: Y i = X i β + ε i ; (1) ε i X i N(0, 1)
The HOPIT model: the self-assessment component The life satisfaction perceived by individual i is: Y i = X i β + ε i ; (1) ε i X i N(0, 1) Individuals report their life satisfaction on a 5-point scale: Y i = j if τ j 1 i < Y i τ j i, j = 1,..., 5. (2)
The HOPIT model: the self-assessment component The life satisfaction perceived by individual i is: Y i = X i β + ε i ; (1) ε i X i N(0, 1) Individuals report their life satisfaction on a 5-point scale: Y i = j if τ j 1 i < Y i τ j i, j = 1,..., 5. (2) The thresholds are individual specific and are given by: τ 0 i = ; τ 5 i = ; τ 1 i = X i γ 1 ; (3) τ j i = τ j 1 i + exp(x i γ j ), j = 2, 3, 4. (4)
The HOPIT model: the self-assessment component The fact that different individuals have different thresholds is what is referred to as DIF.
The HOPIT model: the self-assessment component The fact that different individuals have different thresholds is what is referred to as DIF. Note also that if γ = 0, the model is a standard ordered probit model.
The HOPIT model: the self-assessment component The fact that different individuals have different thresholds is what is referred to as DIF. Note also that if γ = 0, the model is a standard ordered probit model. Using self-reports on life satisfaction only, the parameters β and γ 1 cannot be separately identified, while the other γs are dubiously identified only from the nonlinearities in the thresholds.
The HOPIT model: the vignette component The evaluation of the vignettes are modelled using similar ordered response equations: Zil = θ l + ν il ; (5) ν il N(0, σv 2 )
The HOPIT model: the vignette component The evaluation of the vignettes are modelled using similar ordered response equations: Zil = θ l + ν il ; (5) ν il N(0, σv 2 ) where ν il is assumed to be independent of ɛ i. Z il = j if τ j 1 i < Zil τ j, j = 1,..., 5 (6) i
The HOPIT model: the vignette component The evaluation of the vignettes are modelled using similar ordered response equations: Zil = θ l + ν il ; (5) ν il N(0, σv 2 ) where ν il is assumed to be independent of ɛ i. Z il = j if τ j 1 i < Zil τ j, j = 1,..., 5 (6) We can now identify the θ and γ parameters from the vignette equation. From the self-reports, β can be identified in addition. i
The HOPIT model: the vignette component The evaluation of the vignettes are modelled using similar ordered response equations: Zil = θ l + ν il ; (5) ν il N(0, σv 2 ) where ν il is assumed to be independent of ɛ i. Z il = j if τ j 1 i < Zil τ j, j = 1,..., 5 (6) We can now identify the θ and γ parameters from the vignette equation. From the self-reports, β can be identified in addition. Estimation is carried out via maximum likelihood. i
The HOPIT model: the identifying assumptions 1 Vignette equivalence: there are no systematic differences in how respondents perceive the same vignette (θ l does not vary across individuals).
The HOPIT model: the identifying assumptions 1 Vignette equivalence: there are no systematic differences in how respondents perceive the same vignette (θ l does not vary across individuals). 2 Response consistency: the same reporting styles are used for the self-assessment and the vignette component (same set of thresholds).
Economic and non-economic factors we control for... Demographics: age and gender
Economic and non-economic factors we control for... Demographics: age and gender Socio-economic variables: employment, income, wealth and education..
Economic and non-economic factors we control for... Demographics: age and gender Socio-economic variables: employment, income, wealth and education.. Health: number of chronic diseases, arthritis, limitations with mobility, symptoms, ADL, IADL, obesity and having been diagnosed with depression.
Economic and non-economic factors we control for... Demographics: age and gender Socio-economic variables: employment, income, wealth and education.. Health: number of chronic diseases, arthritis, limitations with mobility, symptoms, ADL, IADL, obesity and having been diagnosed with depression. Social relationships: marital status, family bonds (contacts with children, grandchildren and parents) and extra-familiar activities (voluntary or charity work, activities of a religious, political or community-related organization, attendance of training courses and sport or social activities).
Economic and non-economic factors we control for... Demographics: age and gender Socio-economic variables: employment, income, wealth and education.. Health: number of chronic diseases, arthritis, limitations with mobility, symptoms, ADL, IADL, obesity and having been diagnosed with depression. Social relationships: marital status, family bonds (contacts with children, grandchildren and parents) and extra-familiar activities (voluntary or charity work, activities of a religious, political or community-related organization, attendance of training courses and sport or social activities). (Religious background, left or right in politics).
Country effects - no correction for DIF Baseline.6.4.2 0.2.4.6.8 IT CZ FR PL ES BE NL SE DK
Country effects - HOPIT model Hopit model.6.4.2 0.2.4.6.8 CZ IT PL BE DK ES FR SE NL
Thresholds Medians of individual specific thresholds SE DK DE NL BE FR ES IT PL CZ 4 2 0 2 Threshold #1 Threshold #2 Threshold #3 Threshold #4
Ordered probit vs Hopit A formal likelihood-ratio test strongly rejects the model not allowing for response scale variation against the more general model that does allow for correction of the DIF bias.
Ordered probit vs Hopit A formal likelihood-ratio test strongly rejects the model not allowing for response scale variation against the more general model that does allow for correction of the DIF bias. We also test the joint significance of all the coefficients but the constant in the threshold equations separately for each threshold and the null hypothesis is always rejected.
What happiness is associated with... Demographics: age and gender
What happiness is associated with... Demographics: age and gender Socio-economic variables: employment, income, wealth and education..
What happiness is associated with... Demographics: age and gender Socio-economic variables: employment, income, wealth and education.. Health: number of chronic diseases, arthritis, limitations with mobility, symptoms, ADL, IADL, obesity and having been diagnosed with depression.
What happiness is associated with... Demographics: age and gender Socio-economic variables: employment, income, wealth and education.. Health: number of chronic diseases, arthritis, limitations with mobility, symptoms, ADL, IADL, obesity and having been diagnosed with depression. Social relationships: marital status, family bonds (contacts with children, grandchildren and parents) and extra-familiar activities (voluntary or charity work, activities of a religious, political or community-related organization, attendance of training courses and sport or social activities).
What happiness is associated with... Demographics: age and gender Socio-economic variables: employment, income, wealth and education.. Health: number of chronic diseases, arthritis, limitations with mobility, symptoms, ADL, IADL, obesity and having been diagnosed with depression. Social relationships: marital status, family bonds (contacts with children, grandchildren and parents) and extra-familiar activities (voluntary or charity work, activities of a religious, political or community-related organization, attendance of training courses and sport or social activities). (Religious background, left or right in politics).
Counterfactual - DK and IT thresholds
To sum up Self-reported life-satisfaction is highly heterogeneous across European countries.
To sum up Self-reported life-satisfaction is highly heterogeneous across European countries. We find that variations in response scales explain a large part of the differences found in raw data.
To sum up Self-reported life-satisfaction is highly heterogeneous across European countries. We find that variations in response scales explain a large part of the differences found in raw data. When response scales are assumed to be constant, Danes and Italians are the most and the least satisfied respectively.
To sum up Self-reported life-satisfaction is highly heterogeneous across European countries. We find that variations in response scales explain a large part of the differences found in raw data. When response scales are assumed to be constant, Danes and Italians are the most and the least satisfied respectively....when we correct for scale biases, the difference between Danes and Italians disappears.
To sum up Self-reported life-satisfaction is highly heterogeneous across European countries. We find that variations in response scales explain a large part of the differences found in raw data. When response scales are assumed to be constant, Danes and Italians are the most and the least satisfied respectively....when we correct for scale biases, the difference between Danes and Italians disappears. Using the Danish scale, more than 95% of respondents in all countries would rate themselves as satisfied with their life.