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

Fig. 5

From: The relationship between statistical power and predictor distribution in multilevel logistic regression: a simulation-based approach

Fig. 5

Power curves for the variance component of the random slope are presented. Benchmark conditions (normally-distributed Level 2 and balanced Level 1 predictors or ‘Benchmark’ in the figure legend) as well as moderate (skewness of \( \sqrt{8/5} \) and 70/30 imbalance) and extreme (skewness of \( \sqrt{8} \) and 90/10 imbalance) conditions are shown. Variance for the random slope is \( {\sigma}_1^2=0.3 \) with an ICC of 0.3. Power of 80% is marked with a horizontal line. Horizontal axis denotes Level 1 sample size and vertical axis shows power. Level 2 sample sizes are shown on top of each panel in grey. LV1 stands for Level 1 and LV2 stands for Level 2

Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us