Logit regression with Moderating variable

[Roel statistics]

Hey everyone, I was not able to find a topic about this subject (could be wrong), but I was hoping someone could help me out.

For my thesis I would like to do a logit regression in stata. My dependent variable is 'success' (1 = success, 0 is no success). I have 1 independent variable, some control variables and 1 moderating variable. My question is: How do I do a logit regression in stata, with a moderating variable. If more information is needed, please let me know.

Kind regards

[Carlo Lazzaro]

Roel statistics:
welcome to this forum.
1) as per FAQ, real given and family names are appreciated on this forum. Thanks;
2) as per your (scant) description you're probably referring to an interaction (see -fvvarlist-) between the independent variable and the moderator;
3) please note that (as per FAQ again) it's up to the OP to provide potential interested listers with all the details necessary to receive a positive reply. Thanks.

[Roel statistics]

Dear Carlo, thanks for your reply. As for my name, I am sorry did not know this was necessary, not sure how to change it now.

Maybe a little better description: I want to test whether my dependent variable success (yes or no) is significantly influenced by independent variable team heterogeneity (this is a value ranging from 0 to 1, i.e. 0.38). Next to that, I want to test whether this relationship is moderated by the industry in which it operates. I am thinking about making this variable also binary (high-tech or low-tech). How do I run this logit regression?

[Roel statistics]

To add upon this: for the first logit regression I use the following command:

logit Successdum Externalratio Match_stage Scan_stage Test_stage Build_stage Grow_stage

In which the Successdum is my dependent, externalratio my independent and all the 'stage' variables are control variables. Now I want to add 'Industrydum' as a moderator variable.

[Carlo Lazzaro]

Roel:
1) see https://www.statalist.org/forums/help#realnames about how to re-register wityh real nemes. Thanks;
2) if the predictor you're interested is continuous, interaction loses its relevance. hence, you can probably code something alon the following lines, creating a two-level categorical regressor for -industrydum-:

logit Successdum Externalratio i.industrydum Match_stage Scan_stage Test_stage Build_stage Grow_stage

[Roel statistics]

Thanks a lot Carlo for your help!

1) Will contact asap to change my name
2) Allright, so if I understand correctly I basically add the industry variable as a 'normal' predicting variable and see whether significance of the 'real' predicting variable changes?

[Carlo Lazzaro]

Roel:
not quite.
By adding -i.industrydum- you will see the contribution of that predictor (when adjusted for the other predictors) to variation of the regressand.
It's likely that the contribution of -Externalratio- will change when you add -i.industrydum-.