Modeling body size information within weight labels using probability distributions

What images of bodies do we associate with thinness and fatness? Can our representations of weight-related words be described by simple probability distributions? To answer these questions, the present study examined participants' perceptions of a set of weight-related words using a pictural scale. 259 French women indicated the thinnest, fattest, and best-fitting figures for 13 words. We then used their responses to construct PERT probability distributions, simple skewed distributions allowing to visualize what body sizes were associated with each word. In particular, the variability of the distributions showed how different weight labels can have more or less precise meanings. We found some evidence that the lowest body mass index associated with a label shifted towards thinner figures as body dissatisfaction increased. Using the same method, we investigated the boundaries of what participants consider the ideal body, and showed that the inclusion of their own body in these boundaries predicted their levels of body dissatisfaction. We argue that PERT distributions can be a useful, easy-to-use tool in body image research for modeling the representations of weight labels in different populations.