Ideal Body Image, 19
= 2.6), t (125) = 2.33, p < .05. Similarly, among men the mean for the ideal body shape
was greater in the plus-models condition (M = 3.5) than in the thin-models condition (M
= 3.1), t (114) = 3.06, p < .01. Essentially, these findings support H1, which predicted a
main effect for women and H3, which predicted a main effect for men.
Next, the Ideal Body Image x BMI and Ideal Body Image x Self-Esteem
interactions were examined by recoding BMI and Self-Esteem into a categorical variable
with three levels: low, medium, and high. The highest and lowest quartiles were coded as
high and low, and the middle two quartiles were coded as medium.
For the three levels of self-esteem, there were no statistically significant
differences in two of the three body image dependent variables, namely self-ideal and
projected ideal for other women. But when projecting the ideal for other men, there was
significant difference between the high (M = 2.8) and low (M = 2.4) levels of self-
esteem, which was significant at p < .07. Because this interaction was not central to the
key research questions in this paper, we did not pursue this any further.
The Ideal Body Image x BMI interaction was addressed by examining differences
in body image evaluations in the three levels of BMI. For projections of ideal body image
to other women and other men, no significant differences were observed. For the self-
ideal, those with high BMI (> 26) projected a heavier body ideal (M = 3.6) compared to
those with a low BMI (< 21), who chose a thinner ideal (M = 2.6), which was significant
at p < .001. The interaction with the covariates partially supports H2. BMI was the only
variable that interacted meaningfully with ideal body shape perceptions.
In H4 we had predicted that the thinness of the experimental treatment would
have a greater effect on women than it would on men. This interaction was not
Convention