Periodic logistic regression

We show how to use logistic regression in situations where the phenomenon under consideration is periodic, i.e., follows a cyclic pattern. The method is illustrated by using data for the Snail Potamopyrgus antipodarum, which exhibits a 24-hr periodic pattern of foraging. Snails are found on the top or bottom of rocks, depending on time of day. We show how periodic logistic regression models can be fitted simultaneously to several groups of data, taking into account that the groups may exhibit similar (but not identical) cyclic patterns. The proposed procedure is applicable to ecological studies that analyze the variability in a dichotomous response variable exhibiting a cyclic pattern over time.