A General Framework for Research Design: Monte Carlo Simulation Methods for Sample Size Planning

• Activity Number: 345
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Social Statistics Section
• Author(s): Ken Kelley*+
• Companies: University of Notre Dame
• Keywords: Sample Size ; Research Design ; Power Analysis ; Accuracy in Parameter Estimation ; Effect Size ; Significance Testing / Confidence Intervals

Abstract

Data collection and analysis has grown considerably more complex in many areas. The research design literature has not kept up with the way in which research studies are implemented in many situations. In the design context, a principle concern is the sample size that should be used. Sample size considerations can be based on power, accuracy, or both simultaneously. To overcome the dearth of guidance on planning sample size in modern research designs, this presentation seeks to reconceptualize the way in which sample size is planned. Rather than an analytic based approach, many of which are approximate or does not exist, a general approach that would serve a multitude of approaches and goals would be beneficial. A unified framework is developed for sample size planning using a Monte Carlo study before the study is conducted, which solves issues of planning sample size in a most general way. The a priori Monte Carlo approach to sample size planning allows any design to be evaluated for whichever approach to sample size planning is of interest, indeed, multiple approaches can be evaluated simultaneously, something few existing methods are able to accomplish.