Novel Assessment of Own and Ideal Body Perception Among Women: Validation of the Computer-Generated Figure Rating Scale

The “emaciation/undernourishment” category refers to bodies with a BMI < 17.00 kg/m² (“moderate/severe underweight”) and the “underweight” category refers to bodies with a BMI between 17.00 and 18.49 kg/m² (“mild underweight”).

Three participants did not complete the second session because of personal unexpected circumstances. They could thus be considered as experimental mortality.

The classes of bodies differed significantly in plausibility ratings (Friedman’s nonparametric analysis of variance [ANOVA]: χ²(4) = 315.41, p < 0.001, Kendall’s W = 0.698). Post hoc tests were performed with a conventional significance level adjusted for multiple comparisons (Bonferroni correction: p = .05/10 = 0.005). The results revealed that, except for the comparisons between the “mild underweight” and “overweight” BMI classes (z = 2.40, p = 0.017) and “normal range” and “overweight” classes (z = 2.69, p = 0.007), which did not differ (i.e., p-values superior to the adjusted level of significance), all other differences were significant (all zs ≥ 6.57, all ps < 0.001, all abs(r) ≥ 0.618).

All the participants who incorrectly classified Body #9 did not choose the closest category (see Fig. 3b): Body #9 has an estimated BMI of 17.28 kg/m² (i.e., “underweight” classification category), but 58.41% of the participants classified it in the “normal weight” category (from a BMI of 18.50 kg/m²), while the closest classification category corresponded to “emaciation/undernourishment” (from a BMI of 16.99 kg/m²). This result is discussed later.

The classes of bodies differed significantly in classification percentages (Friedman’s ANOVA: χ²(4) = 180.16, p < 0.001, Kendall’s W = 0.399). More specifically, except for the comparisons (post hoc tests adjusted for multiple comparisons with the Bonferroni correction: p = .05/10 = 0.005) between the “severe/moderate underweight” and “overweight” BMI classes (z = 1.34, p = 0.181), the “severe/moderate underweight” and “obesity” classes (z = 0.508, p = 0.612), and the “overweight” and “obesity” classes (z = 2.24, p = 0.025), which did not differ (i.e., p values superior to the adjusted level of significance), all other differences were significant (all zs ≥ 4.91, all ps < 0.001, all abs(r) ≥ 0.462).

Given that all ICCs smaller than 1.00 denote a proportion of a measurement’s variability (Shrout 1998; Weir 2005), paired samples t tests were performed on the perceived and ideal BMIs to better understand this variability. Neither the average perceived BMI (mean diff. = 0.34, t(109) = 1.44, p = 0.152) nor the average ideal BMI (mean diff. = −0.17, t(109) = −1.18, p = 0.242) differed between the two sessions, meaning that although some participants did not select the same body, the differences were not systematic.

AN-R (N = 10) and EDNOS (N = 10) patients were compared by performing Mann–Whitney U tests. As it could be expected, AN-R patients presented a significantly lower BMI (M = 15.32, SD = 1.28) than EDNOS patients did (M = 18.97, SD = 1.26; z = 3.59, p < 0.001, abs(r) = 0.803) and, consequently, chose a lower perceived BMI (M = 18.38, SD = 4.51) than EDNOS patients did (M = 24.06, SD = 3.95; z = 2.57, p = 0.010, abs(r) = 0.575). Nevertheless, these patient groups differed neither on the perception/estimation bias score; the ideal BMI chosen; the BD assessed with the BSQ, the EDI, or the CGFRS; or the DT score (all zs ≤ 1.32, all ps ≥ 0.05). Moreover, they differed neither on their general classification performance, nor on their classification performances for the different classes of bodies (all zs ≤ 1.02, all ps ≥ 0.05). Concerning the plausibility ratings, AN-R patients (M = 6.72, SD = 1.35) evaluated the general set of bodies as being less plausible than EDNOS patients did (M = 8.26, SD = 0.77; z = 2.68, p = 0.007, abs(r) = 0.600), but these patient groups did not differ on their ratings for the different classes of bodies (all zs ≤ 2.49, all ps ≥ 0.01 corresponding to the adjusted significance level).