Usage Note 22569: Computing power or sample size for a chi-square test on a table
Agresti (1990, 2002, 2013) shows how to compute approximate power for a test of two independent proportions and more generally for the Pearson and likelihood ratio statistics. This information can be used before data collection to select a sample size that yields sufficiently low Type I and Type II error rates, or after data collection and analysis to tell you what the power was for the test. Two sample programs are available:
- PROC POWER or the POWER2X2 macro computes the power of a test that compares proportions from two independent samples for a range of total sample sizes.
- The POWERRXC macro computes the power of the Pearson and likelihood ratio chi-square tests (provided in PROC FREQ) for a range of total sample sizes. A short article that discusses this topic and the programs is available.
Operating System and Release Information
*
For software releases that are not yet generally available, the Fixed
Release is the software release in which the problem is planned to be
fixed.
| Type: | Usage Note |
| Priority: | low |
| Topic: | SAS Reference ==> Procedures ==> FREQ Analytics ==> Power and Sample Size Analytics ==> Categorical Data Analysis SAS Reference ==> Procedures ==> POWER
|
| Date Modified: | 2019-05-02 16:41:56 |
| Date Created: | 2002-12-16 10:56:39 |