We used three models of analysis of variance with the food
photographs as repeated measurements. The first one (2 2
ANOVA) was run on the eating desire scale value for the first series
of photographs (food products in a non-social context), with
participants’ gender as a between-subject factor, and food
categories (liked vs. disliked) as a within-subject factor. The
following two analyses (2 2 2 ANOVAs) were carried out on
the data with the same factors (participant’s gender, food
category), and an additional context within-subject factor: nonsocial
vs. social, i.e., food presented without an eater and food
presented with an eater. This was done for each of the three
emotions expressed by the eaters in order to test the effect of facial
expressions under each emotional condition. When significant
interactions were found, we used other ANOVA models. For
example, we performed an ANOVA with context as the with in subject
factor for each type of product (liked vs. disliked).
Moreover, for each food category, we carried out an ANOVA to
determine the effect of the participant’s gender. The purpose of the
last ANOVA (3 2 2 2) was to compare the three emotional
facial expressions, the two food categories and the impact of eater
and participant’s gender on food eating desire. When overall
significant effects were found, Student’s t-tests were run. A
Bonferroni correction was used for t-tests in the case of the
comparison of the three expressions.
The eater’s gender effect was not significant, so it was not
pursued further.