ABSTRACT
Since an initial focus on “fold change” (cf., Reference 1), researchers have recognized the need to quantify statistical significance of estimated differences in genetic expression between two or more treatment groups using microarrays [2]. In light of this, replication has been recognized as a critical element of the design of microarray studies [3]. Replication may imply spotting a single gene multiple times on one array [4] or multiple tissue samples that each have their own array (e.g., [5,6]). We consider the latter and use the term “sample size” to refer to number of arrays in a study. For another discussion of levels of replication see Simon and Dobbin [7].
