Measures of Interobserver Agreement and Reliability, Second Edition covers important issues related to the design and analysis of reliability and agreement studies. It examines factors affecting the degree of measurement errors in reliability generalization studies and characteristics influencing the process of diagnosing each subject in a reliability study. The book also illustrates the importance of blinding and random selection of subjects.
New to the Second Edition
- New chapter that describes various models for methods comparison studies
- New chapter on the analysis of reproducibility using the within-subjects coefficient of variation
- Emphasis on the definition of the subjects’ and raters’ population as well as sample size determination
This edition continues to offer guidance on how to run sound reliability and agreement studies in clinical settings and other types of investigations. The author explores two ways of producing one pooled estimate of agreement from several centers: a fixed-effect approach and a random sample of centers using a simple meta-analytic approach. The text includes end-of-chapter exercises as well as a CD-ROM of data sets and SAS code.
Table of Contents
Probability and Its Application in Medical Research
Reliability for Continuous Scale Measurements
Models for Reliability Studies
Testing the Equality of Two Independent ICCs
Testing the Equality of Two Dependent ICCs
Large Sample Confidence Interval on ρ2
Raters Agreement for Continuous Scale Measurements
Estimating Agreement When Gold Standard Is Present
Several Faulty Methods and a Gold Standard
Sample Size Requirements for the Design of Reliability Study under One-Way ANOVA
Sample Size for Case 2
Estimation of ICC from Two-Way Random Effects Model with Replicated Measurements
Bland and Altman’s Approach
Regression Methods for Methods-Comparison
Correlated Measurement Errors
Assessing Agreement in Methods-Comparison Studies with Replicate Measurements
Population Coefficient of Variation as a Measure of Precision and Reproducibility
Inference from Single Normal Sample
Coefficient of Variation from the Gamma Distribution
Tests for Equality of Coefficients of Variation
Statistical Inference on Coefficient of Variation under the One-Way Random Effects Model
Maximum Likelihood Estimation
Variance Stabilizing Transformation
Estimating the WSCV When the Error Variance Is Not Common
Sample Size Estimation
Analysis of WSCF from Two Dependent Samples
Measures of Agreement for Dichotomous Outcomes
Indices of Adjusted Agreement
Cohen’s Kappa: Chance Corrected Measure of Agreement
The 2 × 2 Kappa in the Context of Association
Sample Size Requirements
Dependent Dichotomous Assessments
Adjusting for Covariates
Simultaneous Assessment of Two Binary Traits by Two Raters
Coefficients of Agreement for Multiple Rates and Multiple Categories
Multiple Categories and Two Raters
Agreement for Multiple Raters and Dichotomous Classification
Multiple Raters and Multiple Categories
Testing the Homogeneity of Kappa Statistic from Independent Studies
Exercises appear at the end of each chapter.
Mohamed M. Shoukri is a principal scientist and the chairman of the Department of Biostatistics, Epidemiology, and Scientific Computing at King Faisal Specialist Hospital & Research Centre. He is also an adjunct professor of biostatistics in the Department of Epidemiology and Biostatistics, Schulich School of Medicine and Dentistry at the University of Western Ontario. Dr. Shoukri is a fellow of the Royal Statistical Society of London and an elected member of the International Statistical Institute.
"The new edition of this book includes popular and important new topics on assessing agreement and reliability … this book is very readable and it contains detailed examples and programming codes that can be easily followed and carried out by practicing statisticians. It is a valuable reference for someone who has limited background on assessing agreement and reliability."
—Huiman X. Barnhart, Journal of the American Statistical Association, September 2014, Vol. 109
"There are fairly substantial changes from the first edition. These include more focus on defining the subjects’ and raters’ populations, the loss of efficiency consequent on categorizing continuous scores, sample size determination, and pooling multicentre studies. Two new chapters are added, one on models for methods comparison studies, and one on the analysis of reproducibility using the within-subjects coefficient of variation. … Overall the volume provides an accessible introduction to the area, at a sufficient depth for most practitioners."
—David J. Hand, Mathematics Department, Imperial College London, UK, in International Statistical Review
Praise for the First Edition:
… an important strength of this text is that it brings together technical results on agreement statistics from a variety of sources, and it often provides information about the precision of the estimates and interval estimates … this text provides a handy compilation of statistical approaches to interobserver agreement … biostatisticians …are likely to find the summary and results to be helpful, and they will want to add this book to their collection.
—Patrick E. Shrout, New York University, Statistics in Medicine, Vol. 25, 2006
Mohamed Shoukri is well placed to write this book, as he is one of the foremost current researchers in this field. I recommend it.
—Journal of the Royal Statistical Society
The well-written specialized book by Shoukri illustrates the meaning and importance of the measures of interobserver agreement with fun examples. The author is known for clear presentation.
—Journal of Statistical Computation and Simulation, Vol. 75, No. 4, April 2005
This book can be recommended…