Specific non-parametric approaches to analyzing caries clinical trials.

Area of Science:

• Biostatistics
• Dental Research
• Clinical Trials

Background:

• Parametric statistical procedures are commonly used for caries clinical trials.
• Non-parametric methods are recommended for small sample sizes where parametric assumptions may not hold.

Purpose of the Study:

• To discuss the application and extension of non-parametric tests, specifically the Wilcoxon two-sample test and the Kruskal-Wallis test, in the context of caries clinical trials.
• To explore methods for handling stratified data and potential interaction effects within these non-parametric frameworks.

Main Methods:

• The Wilcoxon two-sample test is described in terms of rank sums and their expected values.
• The Kruskal-Wallis test is presented in an analysis-of-variance-like format, also utilizing rank sums.
• Extensions for stratified analyses are discussed, involving accumulation of rank sum differences and variances, and the need for variance-covariance matrices.

Main Results:

• Both Wilcoxon and Kruskal-Wallis tests can be formulated using rank sum differences.
• Stratified versions of these tests can be developed by accumulating differences and variances across strata.
• Current methods lack a specific test for interaction effects in stratified non-parametric analyses.

Conclusions:

• Non-parametric tests are valuable alternatives for analyzing caries clinical trials, especially with small sample sizes.
• Further research is needed to develop and validate methods for testing interaction effects in stratified non-parametric analyses, potentially using weighting strategies.