Abstract: The variance component testing in the regression model is an important problem. In this paper, the authors study three cases in the variance component testing in semiparametric generalized linear model for longitudinal data, and the score statistic is obtained. In the first case, the authors just consider the nonconstant dispersion. Then in the second case, only the random effects are considered. In the last case, the authors consider both influences. Simulation studies are implemented to verify the efficiency of the method.

Key words: Longitudinal data, Semiparametric model, Score test, Penalized spline