Comparing Likert and Slider Response Formats in Clinical Assessment: Evidence From Measuring Depression Symptoms Using CES-D 8

This study compared various response formats in fitting confirmatory factor analysis models. Participants responded to the eight-item center for epidemiologic studies depression scale across five different response formats in a within-subjects experimental design: the Likert-type scale, three types of slider response formats, and a number-entry response format. We compared the different response formats based on item-level scores, factor structure and psychometric properties of the scale, mean comparisons across groups, and individuals' sum scores. Similar results were observed across the response formats with respect to factor structure, measurement invariance, reliability, and validity of test scores. However, inconsistent results were found regarding group mean comparisons across groups. Individuals' item scores and sum scores also varied across different response formats, as did participants' subjective evaluations of response formats in terms of perceived accuracy, enjoyment, difficulty, and mental exhaustion. Based on study findings, we provide recommendations and discuss implications for researchers designing and conducting clinical assessments.