Count data as response variable.

Abstract

This chapter is devoted specifically to count data for three reasons: (i) they are common in ecological studies (e.g. clutch sizes, numbers of fledglings from a nest, numbers of seeds per pod...); (ii) they are simple to collect and are therefore often the data collected by students (e.g. numbers of beetles in a pitfall trap, number of pollinator visits to flowers...); and (iii) they pose numerous issues that linear models with their normal error structure cannot deal with. Two studies will be examined with the response variable being counts, starting with one that nearly fits the ideals of a Poisson distribution well, the other less so. Example 1 deals with fledgling numbers in relation to clutch initiation date. The data are on the northern cardinal bird, Cardinalis cardinalis, and were collected to test the hypothesis that birds that start their clutches later may suffer higher pre-fledging offspring mortality. Example 2 focuses on pollinator flower visits in Passiflora speciosa in relation to flower size.