How To Adjust the Power for a Small Sample Population

The power of a statistical test is its ability to observe an effect. For example, if you were investigating whether a new drug cured a cold, the greater the power of the test, the more likely that you would discover if the drug worked. Sample size, effect size, and the level of significance all influence power. In cases where a small sample decreases power, you can adjust other variables to help increase power.

Instructions

    • 1

      Increase the level of significance, known as the alpha level, which is usually set to .05. This number refers to the chance that the effect you observed in the study was due to random chance. With an alpha of .05, there is a 5 percent chance that the effect you saw in the study resulted by chance rather than from your treatment. By raising alpha, to .1, for example, you increase the power of the test by decreasing your chances of finding a significant effect. Essentially, you are making the test harder.

    • 2

      Increase the effect size. The effect size is the degree to which the treatment worked. The greater the effect size, the greater the effects of the treatment. Therefore, if you raise the effect size, you increase the power by improving your chances of finding a significant treatment effect. The problem is that the effect size may not always be under your control.

    • 3

      Use the finest and most reliable measurement techniques. This will help ensure you are able to measure even a small difference between the experimental groups.

    • 4

      Choose a sample from the extremes of the population. If the sample contains elements, such as people, that are very similar, it will be harder to find a treatment effect. Therefore, increase the power of the test by making the control and experimental groups as different from each other as possible. Be aware, however, that this may reduce your ability to generalize, that is, apply the results to others in the population.

    • 5

      Repeat measurements and/or repeat the study. In situations where it is possible, measure the subjects multiple times and average the measurements. This will help ensure you have an accurate measurement to use in the statistical test.

Learnify Hub © www.0685.com All Rights Reserved