Skip to main content
Fig. 5 | BMC Medical Research Methodology

Fig. 5

From: Nothing wrong about change: the adequate choice of the dependent variable and design in prediction of cognitive training success

Fig. 5

Overview of the studentized bias of the regression coefficients of P-I or P-I x Group. The different regression models that were tested are displayed in the rows (Model 1 to 5) and the different dependent variables are displayed in the columns. In each subplot, the x-axis indicates the sample size and the y-axis the studentized bias for the predictor P-I or the P-I x Group interaction, depending on whether the respective model comprised the interaction term or not. For each sample size, the reliability is colour-coded. The dot indicates the mean of the bias distribution. The thick line covers the interval of the mean plus/minus one standard error and the thin line represents the 95% confidence interval. A bias of zero would indicate that the observed regression coefficient is identical to the true regression coefficient. Note: Red colour indicates a reliability of .60; blue colour indicates a reliability of .70; green colour indicates a reliability of .80; purple colour indicates a reliability of .90. Model 1: P-I + P-II; Model 2: P-I + P-II + Pre-test score; Model 3: P-I + P-II + Pre-test score + Group; Model 4: (P-I + P-II) x Group + Pre-test score; Model 5: (P-I + P-II) x Group

Back to article page