Power analysis


Power analysis is the probability of detecting an effect, given the effect is really there. It is the ability to statistically reject the null when it is false, and avoid a Type II error (accept the null).

There are certain conditions where it is easier to reject the null hypothesis.

1. If you use parametric statistics. Parametric statistics are the elite statistics.

2. If you use a directional hypothesis. One-tailed statistical tests (used with directional null hypotheses) are more powerful than two-tailed statistical tests (used with non-directional hypotheses).

3. If you use large sample sizes. These are statistically more powerful than small sample sizes.

4. If you use a more liberal probability level. If you choose .05 instead of .01, your chances of rejecting the null are statistically greater.

5. If you use superior measurement tools that are documented to have high reliability and validity, attributes of good instruments.

Why do we do a power analysis? A power analysis is often used to determine sample size and it is required by many doctoral advisors. Unfortunately, many students are uncertain what this means for sample size.

There are a few terms that students need to know when they conduct a power analysis for sample decisions.

*What is the alpha level for your dissertation research? .01 or .05 or .10? This is the percent of time your results will be due to chance alone.

*What is the effect size that you are estimating? Cohen D is commonly used with .2 (small), .5 (medium) and .8 (large).

*What power level are you looking for? .8 is the commonly used statistic. This means that 80% of time a difference will be detected if indeed it does exist.

*What is confidence level will you use? 90% (z=1.65), 95% (z=1.96) or 99% (z=2.58)?

*What is the accuracy or margin of error (MOE) you are comfortable with? This usually ranges between 1% and 4%.

There are many power analysis calculators online that you can access for free. But you need to know what you are doing and what type of research study you are conducting.

Return from power analysis to dissertation research questions and hypotheses.