Submission 228
How to Measure Beauty? Comparing Ratings and Pairwise Comparisons of Different Stimuli
SymposiumTalk-01
Presented by: Barbara Mühlbauer
In most empirical aesthetic research, data on the aesthetic appeal of stimuli is obtained by asking participants to rate them on a Likert scale, requiring them to indicate, for example, their perceived beauty on a scale from 1 (not at all beautiful) to 7 (very beautiful). This rating method has many advantages, as it is easy to implement and allows multiple dimensions of an aesthetic impression to be evaluated simultaneously. However, it is uncertain whether data collected with a different method such as pairwise comparison yield similar results. Furthermore, it is unclear how stable individual aesthetic judgments are. Estimating the variance of aesthetic judgments is particularly important because rating studies typically only ask for one judgement.
We investigated these questions in two studies. In the first study, we compared judgments of the beauty of chair images, which were collected using either ratings or pair comparisons. To check the judgements' stability over time, each measurement was repeated ten times. The beauty assessments obtained using the two methods were very similar. These results demonstrate that both methods effectively capture aesthetic judgements. Furthermore, strong positive correlations were obtained between measurement repetitions within participants, suggesting that their opinion of beauty is stable and does not change substantially over time. In the second study, we used abstract artworks to test whether these results could be generalized to other stimuli. In summary, this systematic comparison of the two evaluation methods provides insights into their respective advantages and disadvantages, which will be discussed in the talk.