Validation and Findings Comparing VAS vs. Likert Scales for Psychosocial Measurements

Context: Psychosocial exposures commonly show large variation over time and are usually assessed using multiitem Likert indices. A construct requiring a five-item Likert index could possibly be replaced by a single visual analogue scale (VAS). Objective: To: a) evaluate validity and relative reliability of a single VAS compared to previously validated Likert based items and indices measuring the same construct, b) detect possible statistically significant differences in absolute levels between the single VAS and Likert items and indices respectively. Design: Cross-sectional study conducted in May 2004. Methods: 805 participants responded to a web-based questionnaire including both VAS and Likert based items. Intraclass correlations were utilized to assess agreement between VAS and Likert scales/indices and Marginal homogeneity tests were utilized to detect possible differences in absolute levels. Results: Moderate to strong correlations were found in responses between VAS and Likert based items and indices, and significant differences in absolute levels in seven out of eleven scales. Conclusion: Single VAS questions can, in some cases of uniform construct, replace a single Likert item and also be comparable, but not interchangeable, with multi-item Likert indices.