Impact of actual and self-perceived body type on visual perception of distances
Date
2015
Authors
Branan, Matthew, author
Turk, Phil, advisor
Witt, Jessica, committee member
Hess, Ann, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
We investigate several questions regarding the proposition that physical body size and one's image of their own body type affect the ability to make accurate judgements of distances. Data collected include subjects' guesses of distances of four cones set 10, 15, 20, and 25 meters away and the weight, BMI, and self-perception of body image for each of 67 subjects. Interest lies in determining the covariates that are most important in explaining one's ability to accurately judge distances and whether weight or BMI is the better explainer among the physical body size predictors. We utilize linear mixed models to account for correlation among each subjects' own distance guesses and to allow for flexible modeling of subject-specific effects. Flexibility is further promoted through use of model averaging techniques to account for model selection uncertainty inherent in typical approaches in which an analyst selects only one model from which inferences are made. A generalization of the coefficient of determination from ordinary linear models is made to the linear mixed model setting (R²LMM) in order to provide an additional goodness measure for fixed effects and for individual fixed effects themselves. Baseline differences among subjects' ability to accurately judge distances are so vast that extracting the importance of the fixed effects becomes difficult. It is found that body size is a significant predictor of subjects' ability to accurately judge distances but body image is not at the 0.05 significance level. We recommend choosing weight over BMI as a predictor of guessing behavior based on information criteria, model averaging, and the generalized R²LMM. Specifically, heavier individuals tend to guess more accurately.
Description
Rights Access
Subject
linear mixed model
robust linear mixed model
generalized R squared
self-perception
model averaging