F-test including interaction term

[Jord Sluis]

Dear reader,

I have some trouble running the F-test including an interaction term. I have to run an F-test on the ESG score and the interaction term of the following regression:

reg ROA i.Law##c.ESG Size Leverage sales_growth PPE i.Industry_n i.year, robust

When I try to run the following command:

test ESG=i.Law##c.ESG=0

I get the following error: i: operator invalid

Thought I did not have to include ESG as a variable in the main regression since when running i.Law##c.ESG, stata automatically adds the ESG, and the Law variable as well as the interaction term ESG*Law.

Is there anyone who knows how I can perform the F-test?

Thanks in advance!

[Jord Sluis]

Made a small edit to the question above.

[Rich Goldstein]

you need the testparm command; here is an example:

sysuse auto
regress price i.for##c.weight
testparm i.for##c.weight

also, you might want to see

h testparm

[Leonardo Guizzetti]

Your question is somewhat unclear. Which F-test are you looking for (there's more than one). I will assume you want a test for the entirety of the interaction terms between Law and ESG -- a test which will tell you if there are any differences beyond their additive effects.

if this is your estimation command:

reg ROA i.Law##c.ESG Size Leverage sales_growth PPE i.Industry_n i.year, robust

Then you can use -testparm- to perform the multiple testing for you automatically (or else you will need to spell out each interaction level with i.Law using -test-):

testparm i.Law#c.ESG

Note: there is only one hash mark (#) in the -testparm- statement because Stata will interpret this to mean just the interaction terms (e.g., 1.Law#c.ESG, 2.Law#c.ESG, etc). The use of two hash marks (##) in you estimation command does the same, and also adds the individual effects involved in the interaction (i.Law and c.ESG). This is explained in -help fvvarlist-.

[Jord Sluis]

Thanks for the quick responses, Rich and Leonardo.

Indeed, I want to test for the interaction terms between Law and ESG.
Thanks for your help!