PA 784

Homework 8
Fall 2002

    Due Dec. 5

 

Control Variables, Interval Level Program Data and Multiple Outcome Measures

 

 Open the file mentor2.sav.  Please attach your output to your answers.

 

These data are from a school in LA that received state money to operate a mentoring program. Students were nonrandomly selected into participant and comparison groups. A total of 5 pretest and posttest measures were taken--GPA, and scores on four STAR tests.  The dataset also includes two variables that might affect the outcome measures and so must be “controlled”—grade in school and language spoken in the home (coded 0 for a non-English language and 1 for English).

 

  1.  Regress the “after” GPA on the “before” GPA and the Program variable.

      bulletWhat kind of evaluation design is this?
      bulletWhat was the impact of the program on the “after” GPA, controlling for the “before” GPA?

 

 

 

  1. Now include the control variables—language and grade.  Did the measure of program impact change once these variables were controlled?  If so, how?

 

 

 

 

  1. Substitute the number of mentoring sessions attended by the participants for the program variable.  This will tell you the how much each mentoring session increases GPA, on average, controlling for language and grade. What do you find?

 

 

 

 

  1. Select at least two of the other outcome measures and test for the impact of the program, controlling for language and grade.   Write a brief paragraph to the school administrator summarizing the results of your evaluation, including any threats to validity that might need to be considered.  Would you recommend that the mentoring program be continued?