Hello All,
I am working on CSR.
one study i made to see impact of CSR on financial inclusion behavior of rural people.
for that i have taken financial literacy and income both as a moderating variables for my study.
In results (Hierarchical Regression Model) tells. (DV was Financial Inclusion)
Step 1: all the demographic variables (gender, age and income) came not significant, also the R square value came .02 (2% Only)
Step 2: CSR and financial literacy were entered as an independent variables, came with positive coefficient and highly significant at .001, R square this time .40 (40%)
Step 3: Interaction term (CSR x Financial Literacy) and (CSR x Income) were entered, both came with negative coefficient but (CSR x FIN-LIT) was significant, (CSR x Income) was not significant.
R square now .42 (42%)
Now i am not able to understand how to interpret this (negative but significant interaction term).
Kindly help me regarding this.
Thanks
Kuldeep Singh
Assuming that both CSR and FIN-LIT are interval variables, then you should look up how to interpret "linear by linear" interaction.
Hello Kuldeep,
In addition to the suggestions by Cristian Ramos-Vera and David L Morgan , you'll need to consider what, if anything, happened to the "main effect" scores of CSR and financial literacy in the third step (when interactions were added). As well, here's an excellent resource on understanding of mediation/moderation effects:
Hayes, A. F. (2017). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach (2nd ed.). Guilford Press.
Good luck with your work!
Respected David Morse Sir,
Thanks for your kind reply.
Sir, here i am attaching my results (moderation analysis), Hierarchical Regression Model as well as the graph for conditional effect
Please take a look on this, and guide me regarding main effect.
Thanks Sir,
Sincerely,
Kuldeep Singh
Dear Cristian Ramos-Vera
I am attaching my results
Kindly take a look and guide me for moderating effects
Thanks
Respected David L Morgan Sir,
Thanks for your kind reply,
in my study, i have measured CSR and FIN-LIT on five point likert scale.
should i assume them interval on the basis of Conditioning effect results or in any other manner?
Please guide me regarding this. I have attached my results here.
Thanking you
Sincerely yours,
Kuldeep Singh
There is a difference between using single Likert-scored items and combining multiple Likert-scored items into a scale. If your measures are based on single-Likert scored items, then they are ordinal rather than interval variables and should not be used in a regression.
Respected David L Morgan sir,
My measures are based on multiple-Likert scored items.
I have taken imputed scores from Amos where I run all the items for validity tests and model fit ( CFA)
Than I have made regression
Since you are using Structural Equation Modeling, having Likert-scored items won't affect anything related to your interactions.
FYI, there is nothing wrong with a negative interaction terms, it is just an indication of how the IVs go together to affect the DV. Here is a link to a discussion of interaction that includes an example of graphing "continuous by continuous" interaction.'
https://www3.nd.edu/~rwilliam/stats2/l55.pdf
Thanks David L Morgan sir,
Sir, i have made an interpretation for my interaction effect of CSR x FIN LIT (negative but significant)
"A negative beta coefficient for the interaction term indicates that the effect of CSR on financial inclusion is stronger under the conditions of low financial literacy and vice-versa. "
Is it correct sir?
Thanking you
Respected David L Morgan Sir,
Yes sir, Two main effects are positive.
I have also send you the attached file of my results table. (moderation effect, Hierarchical Regression Model )
I request you to please take a look on that table, and guide me accordingly.
Many thanks to you sir.
Sincerely Yours,
Kuldeep
"A negative beta coefficient for the interaction term indicates that the effect of CSR on financial inclusion is stronger under the conditions of low financial literacy and vice-versa."
Based on the graph you supplied, the only part of this statement that I wonder part is the final, "and vice versa." Of course, interaction effects can always be summarized from the standpoint of either of the two variables involved. But because the graph only emphasizes one of these two possible points of view, I think the "vice versa" might be confusing to readers.
Instead of vice versa, I believe the alternative formulation is that "the effect of financial literacy on financial inclusion is stronger when CSR is lower.
Between these two versions, I would pick the one that is a better match to a meaningful substantive hypothesis, and report just that version of the interaction effect, along with its associated graph.
Either way, as the graph makes clear, the effect of the interaction is notably small than either of the two main effects (which reinforces what is indicated by the size of the coefficients).
Thanks David L Morgan Sir,
I have noted your points.
I will choose that version, which is a better match to a meaningful substantive hypothesis.
Thanks
Sincerely Yours,
Kuldeep Singh
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression