Quasi-Experimental Designs

A program was designed to increase students’ understand of scientific concepts.  One class was selected to participate in the program while another did not. The latter was designated as the comparison group.  A posttest was administered at the end of the year. On this posttest, higher scores, representing a better understanding of scientific concepts, is desirable.  Open comp.sav

1. Regress the Posttest on the program variable. 

a)     What kind of design is this? (see Mohr, table 4.1). 

b)     What appears to be the effect of the program?

2.      Let’s now assume a pretest had been conducted before the program began.  Compare the means of the two groups on the pre-test.  Does this help explain the effect you describe in question 1b?  How so?

3. Regress the posttest on the program, while controlling for the pretest score. 

a)     What kind of design is this? (see Mohr,  table 4.1)

b)     What is the effect of the program?

c)     Would your interpretation be different if students had been randomly assigned to treatment and control groups?  How so?

4.   Are the slopes same for the treatment and control groups?   Run the necessary regressions

a)     What are the slopes for the two groups?

b)     Please draw a graph to show what the slopes of the two groups look like.

5. Compute an interaction term and include it in a model that pools the data from the two groups, along with the pretest and program variables independently.

a)     Please interpret the results.

b)     Are there any threats to validity that you should consider in drawing your conclusion about the effect of the program?