Troubleshooting
Problem
I plan to run a factorial Analysis of Covariance (ANCOVA) using SPSS GLM. Before doing so, I need to test the assumption of Homogeneity of Slopes. This assumption is also known as homogeneity of regression or homogeneity of regression slopes. Can you show me how to do this in SPSS GLM?
Resolving The Problem
This is easy to do following a few simple steps.
1. Set up the GLM Univariate dialog box just as you would for performing the ANCOVA. However, instead of clicking OK, click PASTE. This action opens an SPSS Command Syntax Editor window in which the command equivalents of your dialog box selections appear.
2. In the SPSS Command Syntax Editor window, modify the DESIGN subcommand by adding a term, which is the crossproduct of the highest-order interaction and the covariate. For example, suppose you want to run a two-way ANCOVA in which the factors are named A and B, the covariate is named X, and the dependent variable is Y. The pasted command syntax will appear as follows.
UNIANOVA
y BY a b WITH x
/METHOD = SSTYPE(3)
/INTERCEPT = INCLUDE
/CRITERIA = ALPHA(.05)
/DESIGN = x a b a*b .
The highest order interaction is A*B, so you want to modify the DESIGN subcommand to read
/DESIGN = x a b a*b a*b*x.
3. To run the commands, go to the menu in the SPSS Command Syntax Editor window and click Run->All.
In the resulting output, the test of HOS is based on the significance level of the covariate crossproduct in the Between Subjects Effects table. If the p-value is <.05 (or some other level that you have selected), then you should reject the assumption. Otherwise, the assumption stands.
The covariate crossproduct reflects the collected sums-of-squares and degrees-of-freedom for all possible factor-by-covariate interactions. For the example given above, this means that the F test for A*B*X represents an omnibus test for the A*X, B*X, and A*B*X effects.
Related Information
Historical Number
18757
Was this topic helpful?
Document Information
Modified date:
16 April 2020
UID
swg21476217