A common experimental situation is to compare the effects of two or more treatments or the differences among two or more groups. For example, consider assessing the effect of a drug used to reduce high blood pressure. If the treatment is given to a group of patients, the effect of the drug could be determined in one of two principal ways.
A paired experiment consists of measuring the effect before and after administering the drug. This experiment could be designed by randomly selecting 25 patients and measuring their blood pressures before the drug is given. Blood pressure is then measured again on each patient at a specified time after administering the drug. This type of experiment was discussed in a previous module.
A second experiment consists of selecting a group of 25 patients to receive the drug and a second group of patients to receive a placebo.The mean blood pressures can then be compared after a specified time following the treatments. This is called a two-group or two-sample design.
Example #1 constructs a two-sample confidence interval to compare the mean forward and backward change in balance between elderly and young test subjects. The assumptions underlying the confidence interval are checked.
Example #2 (Requires Lisp-Stat)
Exercise #1 constructs a two-sample confidence interval to compare the mean weights of members on the Cambridge and Oxford crew teams.
Exercise #2 constructs a two-sample confidence interval on the mean specific power difference of land and carrier-based aircraft.