Incremental Validity through Hierarchical Regression is a common practice in psychological research. This is generally conducted by adding the existing measure in step 1, then the existing and new measure in step 2.
I was wondering how to interpret the results when change R-Squared is significant (adding the new measure adds, above and beyond, the old measure), yet the standardised regression coefficients in the second step show STRONGER associations between the old measure and the criterion variable, than between the new measure and the criterion measure?
Cheers,
Conal
Hello Conal,
For the scenario you describe, the interpretation would be that: (a) the new variable added to the model significantly added to the explanatory power of the first independent variable; and (b) of the two independent variables now in the model, the first is given more emphasis than the second.
No paradox, just one possible outcome.
Good luck with your work.
Hi Conal,
this will happen if either a) both regression weights are positive but the predictors show a negative correlation, or b) the predictors have opposing weights (wrt. to their sign/direction) and both correlate positively.
This is simple path tracing.
Best,
Holger
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression