Frequency vs. intensity: Framing effects on patients' use of verbal rating scale anchors

The present study examined the hypothesis that patients' ratings would be less congruent and stable if they were asked to rate imprecise terms on frequency and intensity that were embedded in a behavioral or perceptual framework. Based on data acquired from the patients' ratings the presented terms were analyzed concerning their interindividual congruency, intraindividual stability across time and distinguishability of adjacent terms. Afterward, the results were compared to the results regarding the same analysis of unframed terms from an earlier investigation [16].

In a longitudinal design, 44 patients (age M=39.1, SD=15.2, 68.2% female) with a depressive disorder filled out two established questionnaires (BDI or SCL-90) and questionnaires containing frequency and intensity terms framed in sentences concerning the subjective experience of sadness. Patients should rate the terms with regard to the percentage of time or intensity that is reflected by each term at two different measuring times within one week. Data analysis contained t-tests for paired samples and effect sizes d according to Cohen.

The congruency of framed terms was influenced by an additional factor (vocabulary skills) in comparison to unframed terms. However, congruencies for both sets of terms were rather low. In contrast to unframed items, framed terms showed no intraindividual instability for frequency and intensity terms at all, but were influenced by all of the analyzed factors (age, gender, vocabulary skills, depression, and overall mental symptom burden). Patients could distinguish more adjacent framed terms than unframed terms.

The results give no clear suggestion if unframed or framed terms should be preferred as verbal anchors in self-report instruments. Unframed terms seem to have a slight advantage over framed terms as they are less influenced by the patient's background. However, patients are able to distinguish more adjacent terms if presented framed in a behavioral or perceptual context they are familiar with. Frequency terms showed a higher intraindividual stability of mental representations while both groups of terms exhibited low interindividual congruency. No more than four different verbal anchors could be used safely together in rating scales, as patients with a depressive disorder would not be able to reasonably differentiate more than these.