Statistical methods in evaluating endometriosis studies.

Abstract

There is no shortage of statistical methods available for use in presenting the results of endometriosis research. Those that are currently in widest use have underlying statistical assumptions that may not be met by our study populations. Those methods that are more robust and compensate more for defects in satisfying these underlying assumptions are computer-based and much less "friendly" to physician and patient alike in terms of comprehensibility. Statisticians, with little appreciation for clinical practice, cannot answer these questions alone. Research is needed by clinician scientists comparing the validity of these methods of analysis.