3.1. Where the response variable is not mortality during laboratory experiments involving adult workers

If a response variable to be measured (e.g. a phenotype of interest that may change with treatment) is quantitative or qualitative (i.e. diseased versus not diseased), then a generalised linear mixed model (GLMM) can be used to analyse data in which ‘cage’ is a ‘random effect’ parameter and treatment is a ‘fixed effect’ parameter (Crawley, 2005; Bolker et al., 2009). Several fixed and random effect parameters can be analysed in the same statistical model. If individuals in two or more experimental cages used in the same treatment group are drawn from the same colony, then a GLMM with ‘source colony’ as a random effect parameter should also be used to analyse data. This random effect accounts for the fact that, within the same treatment, variation between two cages of honey bees drawn from the same colony may not be the same as variation between two cages drawn from two separate colonies. This statistical approach accounts for the problem of pseudoreplication in experimental design. If the factor ‘cage’ and ‘source colony’ are non-significant, an experimenter may be tempted to treat individual honey bees from the same cage as independent samples (i.e. ignore ‘cage’). Logically, however, workers drawn from the same cage are not truly independent samples and therefore it would inflate the degrees of freedom to treat individual workers as individual replicates. This point requires further attention by statisticians. In lieu of an immediate solution to this statistical issue, an experimenter can consider using a nested experimental design in which ‘individual honey bee’ is nested within ‘cage’, as presented above.