2.3.2. Structural equation modeling
In order to test the proposed hypotheses displayed in Fig. 1, structural equation modeling (i.e., path analysis) was performed using M-Plus version 6.1 . Path analysis is an extension of multiple regression that allows researchers to examine multiple relationships between variables simultaneously. Variables may serve as both independent and dependent variables, permitting researchers to examine both direct and indirect (mediating) relationships . Given that our hypothesized models included several variables and predicted mediating relationships between them, path analysis provided a comprehensive statistical technique for evaluating this Cy5 hydrazide data.
Fig. 1. Handlers' salivary cortisol (CORT) (A) and testosterone (T) (B) levels and dogs' salivary cortisol (C) levels based on handlers' sex. Male handlers' values are depicted by gray triangles and female handlers' values are indicated by black squares. Error bars represent standard error of the mean. Asterisks denote statistically significant differences between groups shown by Bonferroni-adjusted post hoc tests following ANOVA. Male handlers' pre-competition T levels were significantly higher than females' levels. Cortisol levels of dogs' belonging to males were significantly higher in post-competition relative to baseline (not marked by an asterisk).Figure optionsDownload full-size imageDownload as PowerPoint slide