Home Computer Science OECD guidelines on measuring subjective well-being.
Key messages on the order and presentation of response categories
While there is evidence that scale order can impact results when verbally-labelled scales are used, the reliance on numerical rather than fully-labelled scales for many measures of life evaluation and affect implies that scale ordering is unlikely to be a significant problem for these measures. There is, however, merit in presenting even numerical response order consistently, such that scales run from 0-10 (the classic presentation), rather than 10-0.
Complex questions are sometimes broken down into two separate steps for the purposes of telephone presentation. There is some evidence to suggest that this can alter the overall pattern of responses. As the vast majority of established subjective well-being measures adopt a normal 1-step phrasing structure, it seems preferable to maintain this wherever possible for the sake of comparability (and to reduce survey complexity, both for respondents and interviewers). If compelling evidence emerges to suggest that the 2-step procedure is preferable in some circumstances - and this is perhaps most likely in the case of short-term affective subjective well-being measures - then their use could be further reconsidered. If response categories are limited to simple numerical scales (as recommended for life evaluations as a minimum), these should be less challenging to deliver in telephone interviews, as respondents are required to remember fewer verbal labels. Thus breaking these questions down into two parts may be unnecessary.
The impact of asking respondents to perform mental switching between the underlying “good” and “bad” normative direction of response formats needs further research. The relative merits of frequency (never... all the time), intensity (not at all... completely), and binary (yes... no) response formats for affect items also needs to be investigated systematically, with and without verbal and numerical labels for the scale points, and with reference to the impact on scale sensitivity. Other issues with regard to question ordering, including whether to ask positive or negative items first in a battery of questions, are discussed in the section on question order that follows.
|< Prev||CONTENTS||Next >|