Follow the Fastest or the Slowest? Domain-Dependent Relations Between Response Time and Preference Prediction

Response time can identify useful judgments without a history of previous accuracy, but should the fastest respondent always be followed? The present study examines whether response speed identifies individual ratings that predict other people's average preferences and whether the useful direction of this cue is consistent across evaluative settings. In exploratory reanalyses of published music and visual-aesthetic datasets, six-person candidate groups were formed, responses were ranked from fastest to slowest for each item, and the rating at each rank was evaluated against the mean rating of people outside the candidate group. In the music data, mean absolute error decreased across response-time ranks: the fastest response had an error of 18.54, random individual selection 16.29, and the slowest response 14.62 on a 0–100 scale. The combined first-rating visual data showed a different profile: both endpoints improved on random selection, with a larger advantage at the fastest rank. The slowest advantage was not reproduced in both original visual experiments, and results varied across domains and rating occasions. In music, slow responses were less extreme, and their advantage nearly disappeared after descriptive adjustment for rating extremity. Selecting the slowest response also performed worse than always predicting the scale midpoint or averaging all six candidate ratings. Response time can therefore be a useful cue for selecting a single rating, but its predictive direction depends on the judgment setting and target. Differences in the temporal organisation of evaluated experiences may help explain this variation, although the present observational analyses do not identify that mechanism.