Efficiency of the Slope Estimator in Repeated Measurements

In repeated measurement studies, one of the popular primary interests is to investigate the rate of changes in a response variable. Efficiency of the slope estimator is examined with different numbers of equally-spaced measurements imposed across a specified fixed study period. In this paper, we examine the implication of the frequency of repeated measurements in the efficiency of the slope estimator in studies of fixed duration. We present a closed-form formula for assessing the efficiency of the slope estimator in terms of the standard error of the slope when the true correlation is the autoregressive order I or compound symmetry in the generalized estimating equation model. Since missing values in repeated measurement studies are more often the norm than the exception, we also investigate the efficiency of the slope estimator in the presence of missing data.