Media Exposure, Current and Future Body Ideals, and Disordered Eating Among Preadolescent Girls: A Longitudinal Panel Study

As part of the informed consent procedure, parents were told that their children would be participating in a study about “mass media exposure and self-perceptions, including eating attitudes and behaviors,” whereas children were told that the study was about “the things you like to watch on TV and the way you feel about yourself and your body.” Thus, parents concerned about their children's weight and children concerned about their own weight may have been less likely to give their consent. Unfortunately, we were unable to test weight differences between participating and nonparticipating children, but the body sizes of participating children were typical for their age range (see footnote 4), which suggests that any children who selected out of the study on the basis of weight probably fell at the more extreme ends of the normal range.

As an alternate measure of body size we asked children to report their height and weight so we could calculate their body mass index (BMI). Participating school districts declined to release school height/weight records, so self-report measures were the only remaining option. Only 112 girls at wave 1 and 130 at wave 2 were willing or able to supply both height and weight data. Of these, 96 at wave 1 and 122 at wave 2 had plausible BMI scores (which we defined as falling between 10 and 40, acknowledging that 10 would be extremely underweight and 40 extremely overweight). On the basis of these scores, the average BMI was 18.49 (SD = 4.75) at wave 1 and 19.02 (SD = 4.65) at wave 2. These values are typical for preadolescent girls: the 50th percentile for 7-year-old girls is approximately 15.5 and rises to about 18.8 by age 13 (Cooperberg and Faith, 2004). Given the unreliability of self-report height and weight data and the fact that only a fraction of our sample furnished such data, we opted to use perceived body shape in lieu of BMI as a covariate in analyses. For our purposes, perceived body shape is actually a more appropriate covariate than BMI because perceived body shape was measured on the same type of figure-drawing scale as 2 of the predictor variables, current and future body ideals. BMI, in contrast, cannot be easily linked to figure drawings because muscular children may weigh more than their peers but still be fairly slim. Furthermore, perceived body shape appears to be a good proxy for BMI because, among those participants for whom BMI could be calculated, perceived body shape and BMI were positively correlated, r = 0.46, p < 0.001.