RETR_PWR: an SAS macro for retrospective statistical power analysis

Area of Science:

• Statistics
• Biostatistics
• Quantitative Research Methods

Background:

• Prospective power analysis is conducted before a study to determine sample size.
• Retrospective power analysis, or post-hoc power analysis, is performed after data collection.
• Estimating statistical power retrospectively can inform interpretation of non-significant findings.

Purpose of the Study:

• To present methods for retrospective power analysis.
• To review existing research on the bias and sampling error of power estimates.
• To describe a practical tool (SAS macro) for calculating retrospective power.

Main Methods:

• Delineation of three approaches for point estimates of statistical power.
• Description of an interval estimation algorithm for power.
• Review of prior research concerning the bias and sampling error of retrospective power estimates.

Main Results:

• The article details methods for calculating point and interval estimates of statistical power.
• An SAS macro is presented for implementing these retrospective power estimation techniques.
• The macro is initially designed for F tests but adaptable to other statistics like chi-square and t tests.

Conclusions:

• Retrospective power analysis offers valuable insights after study completion.
• The described methods and SAS macro provide tools for estimating statistical power post-hoc.
• The adaptability of the macro enhances its utility across diverse statistical analyses.