Abstract: This paper focuses on the bias of the group sequential estimate of treatment effect for correlated data using the generalized estimating equation (GEE) method and the Lan and DeMets alpha-spending function. Linear and logistic regr essions are used to examine (a) the magnitude of the bias of a sequential estimate with correlated data; (b) the influence of the true correlation structure on bias. A bias-corrected s equential estimate is proposed using a Brownian motion approximation and numerical simul ation. Logistic regression is used to illustrate and to assess the performance of the proposed method.
Key words and phrases: Alpha-spending function, Brownian motion, correlation structure, GEE, interim analysis, linear regression, logistic regression, sequential boundaries, simulation, working correlation.