Power and Sample Size

Part of new and extended survival analysis commands (the other two include treatment-effects and multilevel models) the new command power replaces the old command stpower to allow the ability to obtain custom tables and automatic graphs of power curves. The new command computes power for tests of odds ratios through power cmh to compute power, sample size or the minimum detectable odds ratios for the Cochran-Mantel-Haenszel test of association in stratified 2x2 tables and for 1:M matched case-control studies. The command power trend produces power and sample for the Cochran-Armitage (CA) trend test.

The existing commands stpower cox, stpower exponential and stpower logrank have been converted to power cox, power exponential and power logrank. All of the new methods and features are featured in 151 new pages in the Stata Power and Sample Size Reference Manual.

A selected list of available methods and features in Stata 14 for Power analysis:

Available Methods	Available Features
Stratified Cochran-Mantel-Hazenszel test (stratified 2x2 tables)	Solve for power, sample size, minimum detectable effect, and effect size
Matched case-control studies	Compare: means, variances, odds ratios, linear trends and hazards
Cochran-Armitage trend test (linear trend in Jx2 tables)	Produce custom and automated tables and graphs
Two-sample log-rank test
Two-sample exponential test (comparison of exponential survival rates)
Cox proportional hazards model

Return to Stata 14 New Features