A Sampled Randomization test for analysing individual cells of behavioural transition matrices is described. The test is compared with two standard procedures by running all three on simulated data. The Randomization test proves roughly as efficient as the X2 test applied to a collapsed matrix; both are more powerful than a test proposed by SLATER & OLLASON (1972). The decisive advantage of the Randomization test lies in its ability to assess the interrelatedness of cell results within a matrix: expected values can be adjusted to show how any mechanism assumed to explain one cell result would affect the rest of the matrix.
Does the presence of absence of eggs in a male stickleback's nest affect the chance that a female will spawn with him? Female threespined sticklebacks (Gasterosteus aculeatus) were presented alternatively to males with or to males without eggs. Our results show that the chance that a female will follow a male to his nest is unaffected by whether he has eggs. Once a female has reached the nest, she can either enter it and spawn, or back out and refuse. There was a tendency for females to be more likely to spawn than to refuse if the nest contained eggs. In a sequential choice experiment females that had refused a male without eggs were then presented to a second male, either with eggs or, as a control, without eggs. Females were significantly more likely to spawn with the second male if he possessed eggs. The finding that females prefer to spawn with males with eggs suggests functional explanations for female refusal, male egg kidnapping, and male 'displacement fanning'.