Review Question - QID 103894

The power of a study is proportional to the sample size. As such, all else being equal, the larger a study the greater the power.

The power of a study is its ability to detect an association when one truly exists. In the proposed pair of studies, the investigators hope to avoid both Type I errors/alpha (claiming an association when one doesn't exist) and Type II errors/beta (failing to detect a true association). Power is the probability of not making a Type II error. Historically, the most commonly used values for alpha and beta are 0.05 and 0.20, respectively. The power of a study is usually calculated before the actual study is performed, either to determine the number of patients needed to be enrolled to achieve a certain power, or if the sample size is predetermined, what the power is expected to be.

Rosner describes the factors which determine the power of a t-test. 1) The smaller the significance level (alpha) the lower the power. 2) The larger the effect size, the greater the power. 3) The lower the variance of the effect size, the greater the power. and 4) The larger the sample size the greater the power. As the alpha value is almost always set at 0.05, and both the effect size and variance are expected to be similar in studies A and B, only the sample size will affect the power of each study differently.

Incorrect Answers:
Answer 1: Study A may have a superior surgeon (although being world renowned is no guarantee of it), but his surgical abilities will not affect the power if they are not reflected in the effect size.
Answer 2: Being a multi-center trial can arguably increase the generalizability of the study, but unless they collectively recruit more subjects than the single center the power will be less.
Answer 4: Double-blinding increases the internal validity of the study but does not affect the power in and of itself.
Answer 5: All else being equal, the study with the larger sample size will have the greater power.