Interval estimation and optimal design for the within-subject coefficient of variation for continuous and binary...

Area of Science:

• Biostatistics
• Statistical Methods
• Reliability Theory

Background:

• Proposes the within-subject coefficient of variation (WSCV) as a reliability index for continuous variables.
• Addresses the need for robust statistical methods in reliability studies.
• Highlights the importance of sample size determination in study design.

Purpose of the Study:

• To develop and validate statistical methods for using WSCV in reliability assessment.
• To derive a variance-stabilizing transformation for WSCV in continuous variables.
• To establish sample size estimation procedures for WSCV for both continuous and binary variables.

Main Methods:

• Employs maximum likelihood estimation for WSCV.
• Utilizes a one-way random effects model for confidence interval construction.
• Conducts Monte Carlo simulations to assess the validity of confidence intervals and investigates optimal sample resource allocation, considering subject number and repeated measurements.

Main Results:

• The variance-stabilizing transformation improves coverage probabilities for WSCV in continuous variables.
• Novel contributions include maximum likelihood and sample size estimation for binary variables based on confidence interval width.
• Demonstrates efficiency of estimation and cost considerations for optimal sample allocation.

Conclusions:

• Provides practical sample size formulas to aid clinical epidemiologists and statisticians in designing efficient reliability studies.
• The proposed methods are applicable to both continuous and binary variables.
• Aims to enhance the design and execution of reliability studies in various research fields.