Strengthening Interactions between Statisticians and Collaborators: Objectives and Sample Sizes

Effective determinations of sample size require interaction between statisticians and their research collaborators who wish to initiate studies. The majority of grant applications, clinical trial protocols, and IRB submissions will not be approved unless there is some statistical justification for the planned sample size that is appropriate for the study’s primary objective [1]. However, obtaining a suitable sample size is non-trivial and may necessitate dynamic conversations regarding aims, objectives, study design, and future directions. Such conversations may address the types of data to be collected, analytic plans, and whether the specified endpoints and objectives will answer the goals of the study post-completion. Beyond meeting the challenge of satisfying reviewers, a well-informed sample size justification will ensure reasonable precision in estimation and adequate statistical power for hypothesis tests. Such a justification is also made in the context of proposing a budget and assessing feasibility of the study design. The sample size must be sufficient to test the primary objective and large enough to obtain preliminary data for secondary objectives and correlative studies; but an excessive sample size may be seen as wasteful and perhaps even unethical, if unnecessarily many subjects are randomized to an ineffective treatment. Thi