Patterns of error - individual differences

The effects of errors or biases may not be uniform across individuals. For example, respondents may differ in the extent to which they are considered to be at risk of motivation, communication, memory or knowledge failures. Some of the cognitive question-answer processes described above assume that a respondent is sufficiently motivated to try to optimise their answers. However, Krosnick (1991) argues that when optimally answering a survey question would require substantial cognitive effort, some respondents might instead satisfice - i.e. provide what appears to be a satisfactory, rather than an optimal, answer. This could involve relying on some of the heuristics or biases listed in Table 2.1. Satisficing is particularly likely when the task is particularly difficult or for respondents with lower cognitive abilities, or those who might be tired, disinterested, impatient or distracted.

There is also evidence to suggest that a significant component of the variance in cross-sectional measures of self-reported survey data may be attributable to individual fixed effects - for example, individual differences in personality or generalised levels of affect, also known as “affectivity” (Diener, Oishi and Lucas, 2003; Diener, Suh, Lucas and Smith, 1999; Robinson et al., 2003; Suls and Martin, 2005). There also appear to be group differences in subjective well-being at the level of country (e.g. Vittersp, Biswas-Diener and Diener, 2005) and culture (e.g. Suh, Diener and Updegraff, 2008), which can operate over and above the differences that one might expect as a result of differences in current life circumstances. The extent to which these differences might be a result of differential susceptibility to error, or certain response styles, rather than substantive differences in underlying levels of subjective well-being, is hotly debated - and discussed in more detail in Section 5 of this chapter. Implications for data analysis are described in Chapter 4.

Summary - factors influencing patterns of error

Some of the factors thought to interact to influence the likelihood of error and/or biases in self-reported measures are summarised in Box 2.1. The essential question for this chapter is how potential sources of error and bias interact with - or can be caused by - aspects of survey methodology, when measuring subjective well-being.

Summary of issues investigated in this chapter

For ease of use, this chapter is organised around survey design and methodological considerations, ranging from the most specific (question construction) to the most general (the wider survey design and methods, and the role of response styles and cultural differences). Table 2.2 provides a guide to the key issues discussed and to some of the interactions or trade-offs between different elements of survey design that need to be considered.

< Prev   CONTENTS   Source   Next >