Notes

  • 1. See the section of Chapter 4 (Output and analysis of subjective well-being measures) relating to cost-benefit analysis for an example of this distinction when comparing stated preference approaches to estimating non-market values as opposed to using subjective well-being measures.
  • 2. The distinction between cardinal and ordinal measures is important to measuring subjective well-being. With ordinal measures the responses are assumed to show the rank order of different states, but not the magnitude. For example, with ordinal data a 5 is considered higher than a 4 and an 8 is considered higher than a 7. However, nothing can be said about the relative size of the differences implied by different responses. For cardinal data it is assumed that the absolute magnitude of the response is meaningful, and that each scale step represents the same amount. Thus, a person with a life satisfaction of 5 would be more satisfied than someone reporting a 4 by the same amount as someone reporting an 8 compared to a 7. Most subjective well-being measures are technically ordinal, but the evidence suggests that treating them as cardinal does not generally bias the results obtained (Ferrer-i-Carbonell and Frijters, 2004).
  • 3. In many cases it may be possible to collect one income measure (say, gross household income) and impute net household income on the basis of household size and composition and the relevant tax rates and transfer eligibility rules.
  • 4. One example of these sorts of measure is provided by the material deprivation questions contained in the EU-SILC.
  • 5. See for example the Eurostat social inclusion and living conditions database, http:// epp.eurostat.ec.europa.eu/portal/page/portal/income_social_inclusion_living_conditions/data/database.
  • 6. These other factors potentially include both the confounding effects of shared method variance if the quality-of-life measures in question are subjective, or more substantive factors such as events earlier in the life course that may impact both income and quality of life.
  • 7. The Five Factor Model is a psychological framework for analysing personality type. It identifies five main factors that relate to personality: Neuroticism; Extraversion; Openness to experience; Agreeableness; and Conscientiousness. The scale is widely used in psychological research and is well suited to inclusion in survey questionnaires.
  • 8. “Frame of reference” refers to the situation or group on which respondent’s base comparisons when formulating a judgement about their lives or feelings. The respondent’s knowledge of how others live and their own prior experiences can influence the basis on which judgements are reached about the respondent’s current status.
  • 9. This is not, in fact, beyond the realm of possibility. Many government agencies may have an interest in collecting measures of client satisfaction. However, the case for collecting general measures of subjective well-being as a standard part of interactions with government service delivery agencies is beyond the scope of this paper.
  • 10. The need for a relatively large sample size is one reason to prefer a simple measure of subjective well-being with a low respondent burden in place of a technically more reliable multi-item measure with a higher respondent burden. The quality gains from a more detailed measure need to be assessed carefully against the quality losses associated with any reduction in sample size associated with a longer measure.
  • 11. Internet surveys are, from this perspective, a way of implementing CASI.
  • 12. In this case the precise transition question used was: “Now thinking about your personal life, are you satisfied with your personal life today”, and the subjective well-being measure that followed was the Cantril self-anchoring ladder of life measure. It does not follow that the same transition question will work in other contexts, and transition questions should be tested empirically before being relied on.
  • 13. Some versions of the satisfaction with life question use different response scales, such as a 5-point labelled Likert scale or a 1-10 scale. Based on the conclusions from Chapter 2, the core module uses a 0-10 end-labelled scale.
  • 14. Technically the Circumplex model implies that positive and negative affect are ends of a single dimension rather than a way of grouping several independent types of feeling. Here the Circumplex model is used as an organising framework to help impose some structure on the range of different affective states, without assuming continuity on the positive/negative axis.
  • 15. Information on affect yesterday from general household surveys, while of interest in its own right, does not allow analysis of how different activities, locations and the people with whom the respondent is with impact on subjective well-being.
 
Source
< Prev   CONTENTS   Source   Next >