Effect of questionnaire response scale on items response characteristics in primary care patients survey: Comparing unipolar versus bipolar response formats

Abstract

Quantitative surveys on patient reported experiences of primary health care (PHC) are often conducted with questionnaires designed with scales or categorical responses. This cross-sectional analytic study involving 322 regular adult users of primary health centre in Nigeria compared the acceptability and score characteristics of 5-point bipolar and 5-point unipolar response formats in a multi-dimensional questionnaire. Outcome measured were the item score characteristics, effect sizes on item scores, response rate, difficulty rating. Significantly higher item scores were associated with the bipolar response format with an overall effect size of 0.12 (95%CI: 0.08 - 0.17). The Odds of responding to bipolar compared to the unipolar response format was 0.65 (95% CI: 0.53 – 0.79, p = 0.000). The response formats in scaled survey questionnaires could influence findings on patients’ self-report. This have implications for the conduct of meta-analysis with data pooled from individual studies with different response formats.