Quite often the popular press carries stories that compare happiness or life satisfaction across nations (for example see last October’s story: Denmark is Happiest Country). Regular readers of this blog will recognize these reports as summaries of research on subjective well-being (SWB) and would be somewhat skeptical of SWB comparisons across populations with very different characteristics and cultures. Why? Because these characteristics and cultures affect both how we interpret constructs like happiness or satisfaction, and how we report them.
This inconvenient fact has long been recognized by both psychologists and economist. Uchida, Norasakkunkit, and Kitayama review the literature on cultural variation in the predictors of happiness and satisfaction and conclude that what constitutes the good life varies substantially across cultures, and this variation translates into differing predictors of happiness. For example in a sample of Japanese adults, emotional support from others is a strong predictor of happiness but in the US emotional support has almost no bearing on happiness. Instead high levels of self-esteem strongly predict individual happiness in the US but less so in Japan.
Not only are there differences in what makes us happy or satisfied, there are clear differences in how we describe our levels of SWB. van Herk, Portinga and Verhallen (2004) find systematic differences in extreme reporting across different European populations. For example Greek, Spanish, and Italian respondents are significantly more prone to extreme responses (such as “Absolutely, I’m very satisfied”) than are French, German, and Brits (who are much more likely to assent with a “Yes, I’m somewhat satisfied”). Minkov also explores differences in extreme reporting and concludes that Middle Eastern societies show the highest degrees of polar judgments (expressions of either extreme satisfaction or dissatisfaction) while East and Southeast Asian societies use the fewest polar terms.
So what can we do if we wish to meaningfully contrast subjective well-being across countries, or even distinct sub-groups within countries? One thing we can do is draw from the nascent literature that attempts to standardize survey responses by anchoring them to vignettes (this approach is introduced in a 2004 paper by King and others). Vignette anchoring is precisely what Angelini and co-authors attempt in an investigation of life satisfaction across 10 European countries in their recent working paper.
In Angelini et al.’s unadjusted cross-country data, Danes are the most likely among all ten nationalities to report extreme life satisfaction. In contrast, Italians are the most likely to report extreme dissatisfaction. But do these national differences reflect real differences in life satisfaction or rather inter-cultural differences in interpreting and using the response categories for the SWB question?
The authors exploit the fact that standardized vignettes on life stories were included in each of the ten national surveys. These vignettes involve short descriptions of a fictional life such as “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… He gets tired easily. Otherwise he has no serious health conditions. How satisfied with life do you think John is?”
It turns out that Danes report both a high level of (own) life-satisfaction and a high level of satisfaction for vignette subjects, suggesting that at least some portion of their reported high level of satisfaction is due to a culturally specific style of reporting. This observation lies behind the idea of vignette anchoring, which uses the vignette responses to standardize the response categories of the SWB questions in order to enhance the inter-subjective comparability of life satisfaction.
By incorporating the vignette responses in the analysis (through econometrically rescaling self-responses according to how the respondent rates the standardized vignettes), national disparities in life-satisfaction are reduced and the ranking across countries changes. After controlling for differences in response styles, Swedes now appear to have the highest level of life satisfaction and Czechs the lowest.
The anchoring vignettes method promises great improvement in welfare comparisons but also requires certain assumptions. One main assumption is termed vignette equivalence, which requires the situation described in the vignette to be perceived and interpreted in the same way by all respondents (if this holds then the variability in vignette responses can only be due to different reporting styles).
Vignette equivalence may be a strong assumption depending on the populations considered and the questions asked, For example a vignette with the following statement: “John is 63 and has 4 adult children but lives independently” can be interpreted as a positive situation in cultures that stress individual autonomy but negatively by cultures that have a norm of adult care for elderly parents.
One way to test vignette equivalence is to compare how different populations order a series of vignettes. In Angelini et al., the 10 national populations show a high degree of congruence when ranking two life vignettes – in each country at least 90% of the population ranks the vignettes in the same order as the mean European ranking. So this is some evidence that vignette equivalence holds, although I would consider ranking tests to be a rather weak test since it involves congruence in only relative orderings.
Vignette anchoring can indeed improve the inter-comparability of different samples so where researchers have the opportunity to add meaningful vignettes to a planned survey then they should do so. But the assumption of vignette equivalence is not guaranteed, especially when comparing dramatically different populations in terms of culture and custom. If we have any doubt about vignette equivalence there may be no alternative to more focused mixed-methods research into the interpretation of SWB concepts specific to the populations studied.