This is about whether an item in a multi-item self-report questionnaire is cued negatively, e.g. “I have felt terribly alone and isolated” or positively, e.g. “I have felt O.K. about myself” to take two items, #1 and #4, from the CORE-OM.
Details #
Does this matter? Many questionnaires are cued entirely negatively or entirely positively and some have mixed cueing. Psychometricians can get quite heated about these choices. Some argue that using mixed cueing increases “careless” responding where someone doesn’t notice the reversed cueing. There is pretty ubiquitous finding that when people complete positively and negative cued measures, whether whole measures (i.e. one entirely made up of positively cued items and another made up entirely negatively cued items) or one or more measures each with mixed cueing the correlation between a negative score and a positive score is clearly lower than minus one (-1). This in turn has led to arguments that any specificity of scores to positive or negative cueing is “just a method effect” or is a substantive difference in how people process and respond to questions about internally positive and negative states.
Another thread, one that impressed me, is that there is qualitative evidence that people find entirely negatively cued measures capable of inducing negative feelings. That came across clearly from both client, ex-client and professional respondents in an old paper I helped: Blount, C., Evans, C., Birch, S., Warren, F., & Norton, K. (2002). The properties of self-report research measures: Beyond psychometrics. Psychology and Psychotherapy: Therapy, Research and Practice, 75, 151–164. https://doi.org/10.1348/147608302169616 with one participant saying “This questionnaire made me feel depressed even though I answered ‘no’ to most things. It feels very negative.” Having said that, I haven’t seen any replication of that work (the start of interest in what I call “qualitative psychometrics“) and it was a smallish sample and set of measures. It’s also arguable that as measures have become shorter over the last 20 (and more) years these dangers of long, entirely negatively cued measures may be less important than they were.
Try also #
- GULP (Generalisability, Usability, Linguistics (and cultural Location) and Psychometrics)
- Psychometrics
- Qualitative psychometrics
Chapters #
Touched on in Chapter 3 in the OMbook.
Online resources #
The CORE measures are all available from https://www.coresystemtrust.org.uk.
Dates #
First created 13.x.25.
