Table 2. Results of Factor analysis.
Principal components analysis (PCA) with oblique rotation; for ratings of photographs and video clips: Factor loadings, percentages of explained variance and internal contingencies for Factor I (aesthetics) and Factor II (personality).
| Traits Cluster | Factor s Photo | Factor s Video | |||
|---|---|---|---|---|---|
| I | II | I | II | ||
| Aesthetics assoc. traits Cluster | Beautiful (Item 1) | .90 | -.01 | .89 | .04 |
| Attractive (Item 3) | .90 | .02 | .91 | .05 | |
| Intelligent (Item 4) | .80 | .08 | .86 | .03 | |
| Pleasant (Item 2) | .79 | .27 | .82 | .27 | |
| Confident (Item 6) | .74 | -.28 | .73 | -.18 | |
| Personality Cluster | Gentle (Item 7) | .08 | .91 | .13 | .90 |
| Flexible (Item 8) | -.17 | .89 | -.22 | .89 | |
| Good natured (Item 5) | .14 | .88 | .17 | .87 | |
| % of variance accounted for | 45.18 | 30.51 | 49.05 | 28.86 | |
| Cronbach´s Alpha | .88 | .88 | .90 | .87 | |
Items had loadings > = .74 (photos) and > = .73 (videos) on the Aesthetics Traits Cluster Factor, and loadings < = | .28 | (photos) and loadings < = | .27 | (videos) on the respective other Factor II. Loadings on the Aesthetics Traits Cluster Factor and Personality Traits Cluster Factor on videos and photos are shown in Table 2.
Factor s appeared to be uncorrelated (r = .11) for the photo series as well as for the video clips series (r = .17). Internal consistency of the photo series data was excellent for Factor I (Cronbach´s alpha = .88) and Factor II (alpha = .88) as well as for the video clips series Factor I (alpha = .90) and Factor II (alpha = .87), respectively (Table 2).
Factor analyses separately performed for males and females replicated the two-Factor structure (73.40% accounted variance for photos and 77.19% accounted variance for videos in women, 76.89% accounted variance for photos and 78.29% accounted variance for videos in men, respectively).
Gender-related comparisons yielded significant rating differences in Factor I as well as in Factor II for both the photo series and the video clip series. Generally, women gave better ratings than men, but the effect sizes were small (Table 3). The final analysis was carried out using the General Linear Model (GLM) for repeated measures, including gender as a covariate. The GLM for repeated measures with skeletal class as grouping variable and modus of presentation as the repeated variable revealed for Factor I a highly significant result for class (F2, 590 = 135.405, p<.001, partial eta2 = .315), but an insignificant result for mode of presentation (F1, 590 = 2.562, p = .110) as well as an insignificant class*mode interaction (F2, 590 = 0.303, p = .739). Gender appeared to be significant (F1, 590 = 34.913, p<.001, partial eta2 = .056)