A Completely Different Approach to Sample Size Planning

• Activity Number: 223
• Type: Invited
• Date/Time: Tuesday, August 8, 2006 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Consulting
• Author(s): Peter Bacchetti*+ and Charles E. McCulloch and Mark R. Segal
• Companies: University of California, San Francisco and University of California, San Francisco and University of California, San Francisco
• Keywords: sample size ; power ; cost efficiency ; innovation ; study design ; research funding

Abstract

We propose a new approach for choosing sample size based on cost efficiency, the ratio of a study's scientific and/or practical value to its total cost. We show that choosing sample size to minimize cost per subject (which does not require projecting the study's value) never falls short of the most cost-efficient sample size and can give smaller sample sizes than traditional methods. An alternative, minimizing total cost divided by the square root of sample size, never falls short for innovative studies, often performs well, and is justifiable in other cases. Notably, for study value assumed to be proportional to power at a given alternative and total cost a linear function of sample size, this choice is more cost-efficient than the sample size producing 90% power. In many situations, these methods are easier, more reliable, and better justified than current conventional approaches.