Figure 1

Four forest plots of the posterior mean differences ${\bar{\beta }}^b-{\bar{\beta }}^c$ . Each panel is a forest plot of ${\bar{\beta }}^b-{\bar{\beta }}^c$ for the first 30 datasets, 95% confidence intervals calculated using the Monte Carlo standard error. Panels (a) and (b)  correspond to binary variable Y imputed under the logistic regression model, with β = 1 and β = 3 respectively. Panels (c) and (d) correspond to Y imputed under the linear discriminant model, with β = 1 and β = 3 respectively.