Key messages on cultural response styles and differences in scale use
Although there do appear to be some cultural or country differences in the patterns of responses observed across subjective well-being questions, very little is known about the extent to which this represents error in the data (rather than genuine differences in how people feel, or how they assess their lives). Perhaps the surest method for dealing with both individual and cultural variation in response styles is to adopt a repeated-measures panel survey design, which enables fixed effects to be controlled in subsequent analyses of the data. Analyses can then focus on examining the determinants of change in subjective well-being over time - which both side-steps the response style issue and offers the considerable advantage that causal relationships can begin to be explored.
Panel data do not, however, solve the problem of response styles potentially influencing the average levels of subjective well-being that might be reported by data providers such as national statistical agencies. Given the concerns around response styles and cultural bias, one option may be to focus international comparisons not on the level of responding, but (as in the analysis of panel data) on any changes in the pattern of responses over time (Cummins and Lau, 2010) - including on any differences in the rate of change between different population sub-groups over time. Internationally, then, the comparator of interest would be something like the percentage change in subjective well-being in different countries within a defined time period. There is already a precedent for this sort of approach in the reporting of GDP, where much of the headline reporting (such as the OECD’s regular press releases) focuses on GDP growth per quarter, rather than on absolute levels of GDP between countries.
However, much as in the case of GDP, there will remain a strong desire to be able to compare average levels of subjective well-being between countries. Because of this, some authors have proposed methods to detect and adjust for national differences in scale use post hoc. These are discussed in more detail in Chapter 4.