Breaking Free of Sample Size Dogma

• Activity Number: 204
• Type: Invited
• Date/Time: Monday, August 4, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Consulting
• Author(s): Peter Bacchetti*+
• Companies: University of California, San Francisco
• Keywords: Ethics ; Sample Size ; Power ; Standards ; Myths

Abstract

Ethical condemnation of studies because of too small a sample size has been based mainly on the idea that they have insufficient projected scientific or practical value to justify the burdens accepted by study subjects. The talk will review this and other arguments for ethical condemnation based on sample size, emphasizing detailed analysis that undercuts the quantitative claim that too small a sample size produces an unfavorable ratio of projected value to burden. Other topics will include: characteristics that are associated with but not caused by small sample size (e.g., failure to disseminate results), participants' expectations and altruistic satisfaction, and the positive predictive value of P< 0.05. Reasons for the persistence of misconceptions about sample size may include status quo bias, longevity bias, and the appeal of simple conventions, particularly when compared to thorny realities.