1.3 Study Design Features

Sound statistical results can be valid only if the study plan is well thought out and accompanied by appropriate data collection techniques. Even the most sophisticated statistical tests might not lead to valid inferences or appropriate characterizations of the population if the study itself is flawed. Therefore, it is imperative that statistical design considerations be addressed in clinical studies during protocol development.

There are many statistical design considerations that go into the planning stage of a new study. The probability distribution of the primary response variables will help predict how the measurements will vary. Because greater variability of the measurements requires a larger sample size, distributional assumptions enable the computation of sample-size requirements to distinguish a real trend from statistical variation. Determining the sample size is discussed in Chapter 2.