Measuring Agreement

CHAPTER 11 SAMPLE SIZE DETERMINATION

11.1 PREVIEW

This chapter presents a simulation-based methodology to determine sample sizes for planing method comparison studies involving two methods. The focus is on two designs that are most common in practice, namely, the paired measurements design and the repeated measurements designs for unlinked data. For a paired design, we determine the number of subjects that ensure a specified level of precision for estimate of a measure of inter-method agreement. For a repeated measurements design, we determine the number of subjects as well as the number of replications of a measurement that ensure a specified level of precision for estimates of both inter-and intra-method versions of an agreement measure. The methodology is illustrated through an example.

11.2 INTRODUCTION

So far in this book we have focussed on analysis of method comparison data. This chapter considers sample size determination for planning a method comparison study involving two measurement methods. Generally, by “sample size” we mean n—the number of independent subjects in the study. A study with too small a sample size may not be conclusive. On the other extreme, a study with too large a sample size is wasteful. Therefore, statistical considerations are needed to determine a sample size that may be considered adequate.

Two approaches for sample size determination are common. One is designed for the situation when testing hypotheses is the inference of primary interest. In ...