Using cognitive interviews to refine a questionnaire developed for Mexican Americans with diabetes

Cognitive interviewing is a qualitative research method recommended by the CDC's National Center for Health Statistics Questionnaire Design Research Laboratory to test questionnaires and improve their design. Cognitive interviews can be used to explore how people interpret and respond to survey items and response options and identify problems respondents have in comprehending survey items, retrieving their memories, choosing among response options, and deciding how to answer. This study used cognitive interviewing as a means of pre-testing the Diabetes Symptom Self-Care Inventory, a questionnaire designed to measure Mexican Americans' diabetes symptom prevalence, their perceptions of symptom importance, symptom frequency, and self-care strategies. The purposive sample of Mexican American adults with type 2 diabetes was recruited from Austin, Texas (n=6) and from Rio Grande City (n=10), a small city on Texas's border with Mexico. The participants were mostly Spanish-speaking (62%), women (81%), aged 54 years (range 29-69), with 12 years schooling or more (62%). The cognitive interviews were conducted in one-on-one sessions that lasted approximately one hour. English and Spanish versions of each item and responses were presented side-by-side so that bilingual respondents could listen to and read either or both versions. Participants responded to each item and then to a series of probing questions about the item and the thought processes they had used. The interviews were audio tape-recorded. Audio-recordings, field notes, and completed questionnaires were analyzed for common and noteworthy problems that emerged across the interviews. In general, respondents were eager to discuss their diabetes symptoms but were unaccustomed to providing their input to questionnaire development. Many respondents found the process of thinking aloud awkward and confusing and they responded more easily to verbal probing techniques. Respondents' comments formed the basis for refinements to the questionnaire including modified item wording, item ordering, and response formats, and a more culturally acceptable and conceptually valid Spanish translation. The revised questionnaire's psychometric properties were evaluated in a subsequent study. Recommendations will be shared for using cognitive interviewing techniques with respondents with low educational attainment and with Spanish-only speakers.