Continuing with the same problem concerning cholesterol and diet. In the study describe above the power was very low. Suppose now that the researcher completed the study described above and that his result was not statistically significant. (Remember, in truth the alternative hypothesis is true.) Another researcher wants to redo the experiment modified in such a way as to increase the power. This second researcher plans on measuring subject's cholesterol level before going on the diet and then measuring it again after each subject was on the diet for 3 months. The measurement of interest is the change in cholesterol level calculated as "Before Diet cholesterol level - After Diet cholesterol level".

b) The researcher wants to keep ?=.05. What can the researcher to do maximize the power? Select all that apply.

Use as large a sample size as possible (given constraints of time and money)

Put each person on a different diet

Choose all the subjects to be close together in initial cholesterol level, engage in the same level of exercise, have approximately the same BMI (body mass index - a standardized measure of weight) and to belong to the same socio-economic group and ethnic group to reduce the standard deviation as much as possible.

Monitor subjects for diet compliance.