partner
found
strong
positive
correlation
height and weight
target
andthose
partner dogs
confine
presented
height
affect the outcomes. We used logistic regression to determinethe effect of body size and play bout duration on the probabilitythat rollovers would occur during a bout. In the analysis of theduration of the supine phase of the rollover (hereafter ‘rolloverduration’), where subjects contributed more than one datum toanalysis, we ran generalized linear mixed models (GLMM) underrestricted maximum likelihood estimation, with rollover durationas the dependent variable and subject identity as the randomeffect. We conducted all statistical tests using Stata 13 (StataCorp,2013) or JMP 10 (SAS Institute, 2012). Tests were two-tailed, withalpha = 0.05. Again, data from the two data sets were analyzed sep-arately so as to determine whether both converged on the samepatterns.