I had run a Hierarchical Multiple Regression in SPSS, by putting 2 control variables in Block 1, 2 predictors in Block 2 and an Interaction in Block 3. While the result show that Sig. F change of the model is not significant, the Sig. of each step is significant.
What does this mean? How can I interpret and report the results?
Also, do I need to report the Sig. F change? All of the Hierachical Regression Research that I read seems to didn't report it.
Your decription is not entirely correct. You have to distinguish between the chance in R^2 (table 1) and the overall R^2 (table 2). In both tables step one is identical (since both estimate the R^2 compared 0). This changes in for the next 2 steps: table one "R square change" and its signifance measure the change against the R^2 values from the steps before, how much variance is additionally explained by this individual step and is this change different from 0. Table 2 represents how much variance all of the included variables explain together and if this is different from 0.
For example, in step 3 table one tells you that the R^2 change =0.004 is not different from 0. But all variables together (all variables from steps 1 to 3) explain R^2 = .110, which is different from 0 (only statistically of course).
Your decription is not entirely correct. You have to distinguish between the chance in R^2 (table 1) and the overall R^2 (table 2). In both tables step one is identical (since both estimate the R^2 compared 0). This changes in for the next 2 steps: table one "R square change" and its signifance measure the change against the R^2 values from the steps before, how much variance is additionally explained by this individual step and is this change different from 0. Table 2 represents how much variance all of the included variables explain together and if this is different from 0.
For example, in step 3 table one tells you that the R^2 change =0.004 is not different from 0. But all variables together (all variables from steps 1 to 3) explain R^2 = .110, which is different from 0 (only statistically of course).
It looks like you are testing an interaction effect, and that the second regression, where you add the two main effects, shows a significant increase in R-sq, while the third regression shows a non-significant increase when you add the interaction effect.
Thanks Rainer Duesing and David L Morgan for your answers. So do I need to report the Sig. F change when I write up the results?
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation