The effects of species, treatment and colony on the survival
of individual ants in Experiments 1 and 3 were analysed
using a Cox proportional-hazards regression model. Pairwise
comparisons were conducted using the Breslow
statistic in a Kaplan–Meier analysis. Where ants were kept
in groups in Experiment 2, survival was instead analysed
using a repeated measures General Linear Model to identify
significant survival differences over time, between species
and treatments. The numbers of ants self-grooming over the
course of the 20-s observation periods in Experiment 1 was
analysed using a Generalized Estimating Equation (GEE)
with binomial distribution and logit link function to test for
differences in grooming frequency between species and
treatments, with the effect of colony nested within species.
A GEE was also used with the data from Experiment 2 to
analyse the numbers of ant groups from each colony in
which self-grooming or allogrooming took place in each
20-s observation period, with a Poisson distribution and log
link function. All analyses were conducted in SPSS 19.0
with non-significant terms removed in a stepwise manner to
arrive at the minimum adequate models.