Wald one-sided test using generalized estimating equations approach

We discuss in this paper the application of a Wald type-test to assess one-sided hypotheses in generalized linear models with correlated responses. The generalized estimating equations (GEE) approach is used to obtain the unrestricted estimates while the restricted estimates are obtained by the least-squares method, The proposed Wald statistical test has for large sample a mixture of chi-square distributions. This statistic generalizes recent test procedures for assessing ordered group effects using GEE. Simulation studies on the small-sample behavior of the proposed statistic are performed, Finally, two examples with real data are given for illustration. (C) 2001 Elsevier Science B.V. All rights reserved.