Fig. 4From: Why item response theory should be used for longitudinal questionnaire data analysis in medical researchPooled variance estimates. Selection of the pooled variance estimates for different distributions, sample sizes (N = 100, 500, 1000), and number of items (K = 5, 10, 20). The upper left plot for example represents 13 different skewness conditions (from negative to positive) with 100 participants measured on six time points with a five-item questionnaire. The points represent the IRT and CTT-based estimates for the level-1 (repeated measurement) and level-2 (between person) variance. The horizontal lines represent the true values for the variance parametersBack to article page