“To calibrate is to use empirical data and prior knowledge for determining how to predict unknown quantitative information Y from available measurements X via some mathematical transfer function.” (Martens and Naes 2, page 2) This is a standard definition out of engineering statistics. This formalization enables systematically creating virtual faces corresponding to any shape regression of interest that can then be rated for any selected property. The variable here is body fat percentage (usually highly correlated with body mass index, the more familiar but less physiologically accurate quantification 1 of body fat). This paper proposes an approach for calibrating facial stimuli that allows facial image input to be formalized by constraining variation solely to the biological or appearance variable of interest. The method is also highly relevant for other studies on how biological facial variation affects ratings. This has implications for theories of social perception, specifically, the relevance of individual rater scale anchoring. The patterns of dependence of ratings on the BFP calibration differ for the different ratings, but not substantially across the six groups of raters. Each subject rated all five morphs for maturity, dominance, masculinity, attractiveness, and health. 274 raters of both sexes in three age groups (adolescent, young adult, senior) rated five morphs of the same averaged facial image warped to the positions of 72 landmarks and semilandmarks predicted by linear regression on BFP at five different levels (the average, ☒ SD, ±5 SD). We prototype such a rigorous analysis for a set of five social ratings of faces varying by body fat percentage (BFP). Studies of human social perception become more persuasive when the behavior of raters can be separated from the variability of the stimuli they are rating.
0 Comments
Leave a Reply. |