Improved Test Procedure and Sample Size Calculation for Assessing Similarity in Two-Group Comparative Studies with Heterogeneous Variances

Show Li Jan, Gwowen Shieh*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

The two one-sided tests (TOST) method for mean equivalence or average equivalence has been extended to assessing similarity or switchability for individual equivalence in clinical trials. Tolerance interval procedures are available to establish similarity with respect to the proportion of the response differences covered by a prespecified threshold range. However, the extended TOST procedures based on tolerance intervals are potentially susceptible to the control of Type I errors. This article aims to present an exact approach with the specified Type I error probability for appraising similarity between two treatments in comparative studies with heterogeneous variances. Analytic examination and numerical comparison are conducted to clarify the utility of the suggested similarity test and the drawback of the current TOST procedures. To enhance the usefulness of the described exact method, the related power and sample size issues are also considered. Computer algorithms are provided to implement the proposed test procedure, power calculation, and sample size determination in similarity studies.

Original languageEnglish
Pages (from-to)724-742
Number of pages19
JournalSankhya B
Volume86
Issue number2
DOIs
StatePublished - Nov 2024

Keywords

  • C12
  • C18
  • Equivalence trials
  • I10
  • method comparison
  • percentile
  • similarity test
  • tolerance interval

Fingerprint

Dive into the research topics of 'Improved Test Procedure and Sample Size Calculation for Assessing Similarity in Two-Group Comparative Studies with Heterogeneous Variances'. Together they form a unique fingerprint.

Cite this