Henry's Physical Education

Steps to Run Regression Types

Simultaneous Regression

1.     Analyze > Regression > Linear

2.     Enter the variables into the regression model

a.     Place history in the DV box

b.     Move all other variables over into the IV box

3.     Click Ok

Hierarchical Regression

1.     Analyze > Regression > Linear

2.     Enter the variables into the regression model

a.     Place history in the DV box

b.     Place SES in the IV box -> click next

c.     Place race dummy codes in IV box ->click next

d.     Place grades in IV box ->click next

e.     Place Locus and SC in IV box ->click next

3.     Select statistics box

a.     Mark “R square change” and part and partial correlations

4.     Click continue then OK

Stepwise Regression

1.     Analyze > Regression > Linear

2.     Enter variables into regression model

a.     Place history in the DV box

b.     Place grades, SES, self concept and locus in the IV box

c.     Do not move race variables into the IV box (they are a structural set, and do not make sense alone)

3.     Under IV box, select “stepwise” for the method

4.     Click OK

Note: In between each regression type, you will need to click “reset” in the regression window

Compare Outputs of Different Regressions

1.     Things to pay attention to

a.       Simultaneous regression –

                               i.       R2 and associated statistics used to determine statistical significance and importance of overall regression

                              ii.       Regression coefficients to determine magnitude of each variable

b.       Hierarchical/Sequential Regression

                               i.       Change in R2

c.        Stepwise Regression

                                 i.     Change in R2

                                ii.     Regression coefficients

2.     Compare regression outputs

a.       How does the R2 change across each model?

b.       Are the regression coefficients (and significance values) similar across the models?

c.        What conclusions can you draw from the different model types?