Practical Power Analysis
May 5, 2019, 3:00 - 6 pm (Eastern)
Presenter: Eric Hedberg
Power Analysis is increasingly required by funding agencies in evaluation proposals. The reason for this is important: under powered studies are at risk of producing inconclusive results and ceding publication attention to possibly unreliable findings. However, power analysis is typically not a major topic area in many graduate programs, which leads to a disconnect between graduate training and professional practice.
This workshop seeks to introduce participants to the broad strokes of power analysis, focusing on what most evaluations include: a contrast between two treatment groups. Rather than tour through a particular software program, this workshop covers the major conceptual foundations that support all the available tools:
- Finding power using the null and alternative distributions
- Noncentrally parameters
- Formulas for power, sample size, and minimum detectable effects
- How covariates influence power
- Multilevel/multifactor ANOVA models as extensions to two-group contrasts
The presentation materials (available for preview at
hedbergec.com) include widgets (apps) that allow participants to calculate power interactively to reinforce the conceptual presentation.