I use hierarchical regression to test whether or not an independent variable moderates the relation between another independent variable and the dependent variable! The last model has an R² change of .000 though the "change" is significant (F = 56.940, p < .001). How do I interpret this if I have the hypothesis "Y moderates the relation between X and the dependent variable"? Does this hypothesis hold since the model is significant?
See the following document, it might help
http://orsp.kean.edu/documents/Moderation_Meditation.pdf
No, I wouldn't consider this any good. Your data might be broken somewhere. The R²-change indicates the change of R² between your two models: the one before adding the mediator and the one after. If your change is smaller than .000 but is still Significant at such a high rate, then this means that your Variance is very odd.
F = variance of one group / variance of other group
If this value is 56... then variance of one group is 56 times higher than the other. Yet if their change is way below .000 then both variances must be very small.
Like this maybe?
56.940 = 0.0005694 / 0.00001
Such low variances are just unpractical. Why? Because R² reports how well your independent variable predicts your dependent variable. If your R² is 0.0005694 then it means that of all the chaos and variability in your dependent variable, the independent variable explains only 0.05694%. That's much less than one percent!
Thanks Alexander for valuable answers! I agree with you it's a very odd! however I tested the regression assumptions before running my model to ensure the quality of data gathered! but i got these results when running hierarchical regression! R-square for model 1 = 0.493 also R-square for model 2 = 0.493 with no R² change = 000! and my model still significant with (F = 56.940, p < .001) also the beta value for model 2 Beta = (0.17-) with T= (0.212-, sig. 0.832)! How do I interpret this results!!??
I can not explain the R² change significance.
The second part says: t-Test is not signiicant and the beta is likely caused by random variation.
Hi,
ΔR2=R2current−R2previous (taken from https://www.ibm.com/support/knowledgecenter/en/SSLVMB_22.0.0/com.ibm.spss.statistics.algorithms/alg_regression_rsquarechange.htm ).
In your case R-square for model one is the same as for model two, therefore, SPSS couldn't give you any other value than 0.
To my understanding both of your models show some strength in the relationship, but the change itself (0) is essentially meaningless.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS