Read the following scenario and explain what power issues may arise. What factors influence statistical power?
A researcher is interested in investigating how teaching style (structured vs. unstructured) affects learning of vocabulary in children who are developmentally delayed, typically developing and gifted. The researcher organizes an after-school program and all participants are given a pretest. The program consists of 30 minute sessions, two times a week for four weeks. A total of 15 participants are randomly assigned to the 'high structure' class" or the 'low structure' class. All classes focus on word knowledge and vocabulary. At the end of the program, the researcher assesses 'word knowledge' for developmentally delayed, typical and gifted students in the high structure group, as well as developmentally delayed, typical and gifted students in the low structure group. The researcher is wondering whether the level of structure (high vs. low) might affect the participant groups differently.
The issue here is that the sample size is far too small. Power is directly related to sample size; if we want to increase power, the easiest way to do it is to increase sample size. In this case, we essentially have 6 groups: high structure/developmentally delayed, high structure/typical, high structure/gifted, low structure/developmentally delayed, low ...
Explains how low sample size influences power.