Skip to main content
Figure 1 | BMC Medical Research Methodology

Figure 1

From: Evaluation of the Propensity score methods for estimating marginal odds ratios in case of small sample size

Figure 1

Evolutions of the type I error (Panel A) and the bias and the mean square error (MSE) in the estimated coefficients (Panel B) when decreasing the sample size according to the method used, analysis of PS matched or inversely probability weighted (IPW) data sets. These results were obtained using a non parsimonious PS model that included the four baseline covariates. The strength of the association between the baseline covariates, the treatment and the outcome was uniformly set as strong (odds ratio of 2.5) with a marginal prevalence of the treatment of 0.5. In the upper panel (Panel A), the type I error rate was obtained under the null hypothesis. In the lower panel (Panel B), the bias and the mean square error were computed using a treatment effect set at 2.5.

Back to article page