Regression-based tests for overdispersion in the Poisson model☆

摘要: Abstract A property of the Poisson regression model is mean-variance equality, conditional on explanatory variables. ‘Regression-based’ tests for this are proposed in a very general setting. Unlike classical statistical tests, these require specification only relationship under alternative, rather than complete distribution whose choice usually arbitrary. The optimal regression-based test easily computed as t-test from an auxiliary regression. If alternative hypothesis fact specified and Katz system distributions or Cox's local approximation to Poisson, score equivalent test.