Suppose that a researcher is doing a study of the effects of Artistic Training on 4th grade students in Virginia. Ten schools volunteer and she gets agreements to have the Art teachers provide the training that she has prepared. A test of Art Knowledge is also developed and will be administered at the beginning and at the end of the art training. She also knows that she should use a control group, but the Virginia schools will not let her randomly select schools. So she selects control schools that are as similar to the 10 experimental schools as possible, using a set of 3 variables (average SES of each school, State-test average for each school, and percentage of non-native English speaking students in each school). She will test all the 4th grade students in those "control/comparable" schools at the same time that she tests the students from the experimental schools.

(a) What are the risks with this approach to finding a comparison group?

(b) What does she gain from getting a comparison group in this way, compared to not having a comparison group of students at all?

(c) What statistical analysis should she use for the research design described? Defend your choice of analysis and discuss the advantages and disadvantages of the analysis you recommend.

(d) Discuss the advantages and disadvantages of the research design described compared to using randomly selected schools as the experimental and control schools (i.e., full random selection).