I'm running an ANCOVA in SPSS to determine the difference in a continuous DV between 3 groups (from one categorical IV).

The data also contains a continuous covariate (levels from 1-10), and the cases are weighted so that there is an equal number of cases based on the group type and the covariate (i.e. 5 cases from group 1 at level 1, 5 cases from group 2 at level 1, etc.)

Because of the weighing, I assumed that an ANCOVA wouldn't even be necessary, since the mean covariate level is the same across all 3 groups. However, when I compare the results between an ANCOVA and an ANOVA, there is a small difference, because in the ANCOVA, the covariate is evaluated above the mean value (i.e. instead of 5.5 it is evaluated at 5.9).

What could be the reason why this occurs? Also, am I right to assume that the analysis of covariance isn't necessary because of the preliminary weighing, and that an ANOVA would suffice in this case?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.