Home Computer Science OECD guidelines on measuring subjective well-being.
Key messages on unipolar and bipolar scales
The literature on scale polarity is relatively sparse and largely limited to affect measures, rather than evaluative or eudaimonic measures of well-being. Although not widely studied, there is some evidence to support the view that scale polarity matters, in particular for affective measures of well-being. As one goal of affect measurement may be to test the determinants of different affective states, unipolar measurement scales are often preferred. Given that the evidence implies a general default tendency to interpret affect measures as bipolar, steps may be required to convey this unipolarity more strongly. Options include asking respondents to make simple yes/no judgments about a range of affective experiences, although the risks to scale sensitivity and acquiescence need to be considered.
Many existing and widely-used scales for both life evaluations and eudaimonia are bipolar. Given the apparent tendency of respondents to interpret these types of measures in a bipolar manner, regardless of the actual question wording, adopting bipolar scale anchors may be the least ambiguous in terms of all respondents interpreting the scale in the same way. This suggests that existing bipolar scale structures are probably adequate for these classes of measure, although Davern and Cummins’ work does suggest that bipolar response formats may cause problems when explicitly attempting to measure negative constructs (i.e. dissatisfaction). There are some grounds to think that adopting negative and positive numerical labels for scale intervals (e.g. -5 to +5) could further reinforce the bipolarity of measures, but as the additional advantage of this approach is likely to be small, there seems less rationale to depart from current practice - which would in itself reduce comparability with previous work.
|< Prev||CONTENTS||Next >|