A Guide to Sample Size and Power for Non-Parametric Analysis

Sample Size and Power for Non-Parametric Analysis
 A guide to non-parametric analysis methods, tools & sample size considerations

Parametric tests, such as t-tests and ANOVA, are commonly used in clinical trials in order to establish the efficacy of candidate treatments. However, parametric tests require a number of assumptions to be met including the distribution of the data.

In cases where these assumptions may not be met, researchers may consider semi-parametric tests and non-parametric tests which are robust in a wider variety of scenarios. 

Semi-parametric methods such as Cox Regression and non-parametric tests such as Mann-Whitney U tests are therefore an essential tool for statistical analysis. Given this, there is a wide range of sample size methods available for finding the appropriate size for a trial using non-parametric analysis.

You will learn about:

• Parametric, Semi-Parametric and Non-Parametric Analysis

• Statistical Methods for Non-Parametric Analysis

• Sample Size for Non-Parametric Tests

Sample Size and Power for Non-Parametric Analysis

Parametric tests are commonly used in clinical trials in order to examine candidate treatments. However, these tests require that the underlying data satisfies certain assumptions.

For example, common methods such as t-tests and ANOVA are optimal when the data is continuous and normally distributed. In the cases where distributional or other assumptions are not met then semi-parametric and non-parametric tests may be useful.

Semi-parametric methods such as Cox Regression for survival data or proportional odds models and non-parametric tests such as Mann-Whitney U tests and quantile regression are therefore an essential tool for statistical analysis.

Common scenarios where non-parametric tests are useful are for ordinal (e.g. pain severity scales) and interval (e.g. Likert scales) data, where the outcome data has an order but is not continuous, and for complex continuous data such as when comparing data from fat-tailed distributions. 

Given the wide range of scenarios considered for non-parametric tests, there is also a wide range of sample size methods available for finding the appropriate size for a trial. These methods can allow researchers to better explore the implications of their trial outcome on study power and feasibility.

In this tutorial we have covered non-parametric tests, non-parametric equivalents to popular parametric tests and practical examples of sample size determination for a selection of non-parametric tests including the Wilcoxon signed-rank test and the Wilcoxon-Mann-Whitney U tests for both continuous and ordinal outcomes.

nQuery helps make your clinical trials faster, less costly and more successful with tools for sample size calculation - refining Frequentist, Bayesian & Adaptive designs. 

Start your unlimited 14-day trial to help design efficient, informative and ethical adaptive trials.