On confidence intervals of a common intraclass correlation coefficient

Area of Science:

• Statistics
• Biostatistics
• Quantitative Psychology

Background:

• Intraclass correlation coefficient (ICC) is crucial for assessing reliability and agreement.
• Estimating ICC with multiple samples and unequal family sizes presents statistical challenges.
• Existing methods may lack precision under these complex conditions.

Purpose of the Study:

• To develop a novel, robust method for confidence interval estimation of a common ICC.
• To address the complexities of unequal family sizes across multiple samples.
• To evaluate the performance of the new method against established procedures.

Main Methods:

• Utilized the concept of generalized pivots for interval estimation.
• Applied the method to derive confidence intervals for a common ICC.
• Conducted comparative analyses using coverage probabilities.

Main Results:

• The proposed generalized pivot method demonstrates reliable confidence interval estimation.
• The novel approach shows favorable performance, particularly under unequal family sizes.
• Coverage probabilities indicate the accuracy and validity of the new estimation technique.

Conclusions:

• The generalized pivot approach provides a valuable tool for ICC confidence interval estimation.
• This method enhances statistical accuracy in studies with heterogeneous sample structures.
• The findings support the adoption of this novel technique in relevant research fields.