Stat 379 Homework 7Due: Wednesday April 9, 2008

A new method for teaching math is being tested on children in the fifth grade. Half of the children are randomly assigned to learn math in a "student-focus" method, where the teacher focuses the lecture on the needs of the students, constantly encourages them to ask questions, and links the study of math to their day to day experiences (e.g., games). The other half of the students are assigned to a "math-focus" method, where the teacher focuses the lecture on math as a subject, doesn’t allow for many questions and overall pays less attention to the students' needs. The outcome measure is a score on a math test at the end of the term.

The variables in your data set HW7s08.sav are:

Stud_id - student number

Method - teaching method (0 = math focus, 1 = student focus)

Mood - mood of the student after class (averaged across time)

SES – student's SES

SE – student's self reported Self Esteem after class (averaged across time)

Gender – student's gender

Math – scores on math test at the end of the term

1. Which variables in this data set can be potential moderators of the effect of teaching method on math? Identify the potential moderator(s) and explain why you might expect the variable(s) to moderate the effect. (You don't need to test the moderation / interaction for this assignment, but since you know how (center, form product terms, test sequentially), you may want to do so just as an exercise.)

2. Which variables in this data set can be potential mediators for the effect of teaching method on math? Identify the potential mediator(s) and explain why you might expect the variable(s) to mediate the effect.

3. Conduct a mediation analysis on the potential mediators you identified in (2).

a.Run the mediation test on your first (in order in the data set) potential mediator using multiple regression, employing Baron & Kenny's steps. Identify the IV(s) and DV and report the results for each step (b, , t, and p values). You should identify each b-weight as c, a, b, and c', as appropriate.

b.Test for the indirect effect using Preacher’s calculator. Report the Sobel Z and p value.

c.Run the mediation test using bootstrapping: download Preacher and Hayes's syntax and run it; then run the following line of syntax in a new syntax window, substituting your actual variable names for 'DV', 'IV', and 'Med' (and including the period!):

SOBEL y = DV / x = IV / m = Med / boot = 3000.

Report the estimate (called "mean") of a*b and the 95% confidence interval's lower limit and upper limit.

[Note: the output includes Baron and Kenny unstandardized (b not ) estimates of paths c, a, b, and c' (in that order), so you should be able to compare these to the b estimates you obtained in (a) to verify you've done it correctly. The value of a*b is, well, the product of a and b; the Z value of the Sobel test is the one corresponding to the Aroian version of the test in Preacher's calculator output. Also note, this command took nearly four minutes to execute on my Mac G4, so don't be confused by a long delay before output appears.]

d.For the second potential mediator you identified in (2), again use the bootstrapping technique to test for mediation; note that here you just need to change the name of your "Med" in the SOBEL line of syntax, and run it again. Report the estimate of a*b and the 95% confidence interval's lower limit and upper limit.

4. Interpret the results you obtained from question 3, i.e., explain what the analysis tells you about the teaching techniques and what does or doesn't mediate their relationship to math exam performance.

To run syntax in SPSS, go to File -> New -> Syntax to open a syntax window. Enter (type or paste) the syntax you want to run (as in 3(c) above) into the window. Then go to the menu and choose Run -> All (or whichever option is appropriate). Note that "Run" is only a menu option from the syntax window, and does not otherwise appear in the menu. Output of the commands appears in the familiar output window.

Sobel test calculator:

Bootstrapping SPSS syntax "macro" instructions:

Bootstrapping SPSS syntax:

Dave Kenny's mediation web page including four steps: