Regression using Moderation Variables Analysis

[Zdavir Muhammad]

Hallo i wanted to say thank you again that you spent your time for willing to answer my question.

First of all i apologized for the limited word that i have but im trying to communicate with best and easy language .
Im having some problem to learning using Stata to regress my variables. I cant event found any videos about my problem on Youtube.


Now im conducting study about of banks violation on rules that affected their income. But with the theories said that it depend on the size of the bank and his LOA and CAR. This is in my study called moderating variables.

How to conduct it and run the command on Stata? In my problem, can i run the assumption test? How?

Thank You

[Attaullah Shah]

If you are interested in testing moderating effects, you need to introduce interaction terms between the variable of interest and the moderator. Stata has built-in support for interaction terms using ## symbols. Since all your variables seem to be continuous variables, you need to add letter c and dot as a pre-fix with such variables. For example,

reg income c.size##c.LOA
* if LOA is a dummy or factor variable, then
reg income c.size##i.LOA

Afterwards, you can use margin command to study the moderating effects more clearly.

[Zdavir Muhammad]

Thank you for the answer, Sir

[Zdavir Muhammad]

Im sorry for asking too many question sir.
This is my variables: indep var(violation bank on rules) and dep var is (1)bank income performance on Abnormal ROA with Abnormal ROE (2) bank income performance on NIM NPL BOPO and CAR.

Each of these dep var affected by moderating var: size of the banks and LAR CAR(as Risk).

Sir, how i make the command with these, sir? I common using just linear regression wiht one dep var with many indep var.. But havent met like these before..

Thanks a lot

Carlo Lazzaro Attaullah Shah

[Attaullah Shah]

I feel that there is omitted variable bias in your research design. There are many factors that might affect a bank's profitability. You are using size and a moderator as the only explanatory variables. A simple google scholar search shows that there are many papers available on this topic(see https://scholar.google.com.pk/schola...1%2C5&as_sdtp=). At least you should consider these papers and include relevant control variables in your regression.
Concerning the technical aspect of your question, if you have many dependent variables, you will need to estimate separate regressions for each dependent variable, unless you are using a system of equations.

[Carlo Lazzaro]

Zdavir:
see -help mvreg-.

[Zdavir Muhammad]

Zdavir:
see -help mvreg-.

Yes, sir i have tried it, sir..but i got confuse when i have to find how to put it in the Stata command when it put it on this:

indep var(violation bank on rules) dep var 1)bank income performance on Abnormal ROA with Abnormal ROE (2) bank income performance on NIM NPL BOPO and CAR.

dep var affected by moderating var: size of the banks(on both side of the dependent var) and LAR CAR(as Risk on just the Abnormal ROA with Abnormal ROE).

Thank you, sir, for your answer

[Carlo Lazzaro]

Zdavir:
I fail to get what you're after.
Anyway, you may want to run separate regressions and then compare them via -suest-.

[Zdavir Muhammad]

Zdavir:
I fail to get what you're after.
Anyway, you may want to run separate regressions and then compare them via -suest-.

Im sorry, sir, for my bad English

What im trying to say is i wanted to run a regression with two dependent variables..with on each dependent variables theres moderating variables affected the relation..

Let say this:

"I have one independent var (1) Vio

two moderating variables (1) BankSize (2) CAR-Ratio

and two dependent var: on one side (1) AbnROA and AbnROE ; and the other side (2) NIM NPL BOPO

Banksize as moderating variables is affecting dependent var on side (1) and the other moderating var CAR-Ratio affecting the other dependent var on side (2)

With what command i should run the regression, sir?

Many thanks for your attention