Comparing Likert and visual analogue scales in ecological momentary assessment

Abstract

Abstract Measuring subjective experiences in ecological momentary assessment (EMA) studies has become pervasive in psychological science. A design choice that has to be made in all of these studies is which response scale to use. However, to date there is little guidance on this choice in the context of EMA. As a first step towards understanding the effects of different response scales, we experimentally vary the response scale and assess whether the resulting time series of subjective experiences are systematically different. We conducted a between-person experiment comparing a seven-point Likert scale ( $$n=63$$ n = 63 ) with a Visual Analogue Scale (VAS; $$n=56$$ n = 56 ) in an EMA study measuring affective states over 14 days. Using Bayesian multilevel models, we found that the VAS resulted in moderately higher within-person item means, lag-0 correlations, lag-1 autocorrelations, as well as lower within-person skewnesses and response frequencies of exact zeros. We found the largest difference in correlations with external criteria related to psychopathology, where correlations for the VAS were much higher. We did not observe reliable differences in within-person item variances, root mean squared successive differences, missing data, duration of measurements, and ratings about the experiences with the EMA survey. Apart from higher within-person means and higher correlations with external criteria in the VAS group, the differences were relatively small. While more research on response scales in EMA is needed, based on our results we conclude that the VAS should be preferred in studies aiming at capturing affective states relating to general psychopathology, as well as for items whose variation occurs close to scale limits. We conclude by discussing how our findings may contribute to a larger research agenda that addresses the fit of different response scales for different research aims.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 31 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.