Can I use moderators that are conditional to the presence of a certain IV outcome (larger than 0%)?

[Justin Peters]

Dear Stata Experts,
I am relatively new here and currently calculating my first research model. As part of this, I wanted to ask a general conceptual question:

My IV is a continuous variable ranging from 0 to 100 (investment score)

I would like to use a moderator / interaction term, however, the moderator would not be applicable to any companies with an investment score (thus IV) of 0 but only if the score is 1 or higher (cannot be calculated for them or would be 0 automatically).

Does this work and is commonly done or should a moderator always be applicable to all IV outcomes?

Thanks a lot for your time and support!

[Clyde Schechter]

I think it depends on what the moderator and the IV are. If, in the situation where IV = 0, the moderator is a variable whose actual value would be 0 if you had measured it, then it would be reasonable to do that.

But if you would just be arbitrarily coding the moderator as 0 when it is truly not calculatable (i.e. when IV = 0), then this would be a serious misspecification of the model. If the moderator truly is not definable when IV = 0, then the moderated analysis should be restricted to the data subset with IV > 0.

[Justin Peters]

Thanks a lot!

In my case the IV checks the % of the total funding a company received by a certain investor type (with focus on sustainability) and how this affects the likelihood that this venture is acquired by a company with a sustainability mission.

My moderator would check (either as count or binary) if the investor type "sustainable investor" had a successful green venture as part of the investment portfolio in the past, which should then make it even more likely due to experience, reputation etc. that a now new investments gets acquired.
However, if the % of total funding from sustainable investors is 0%, this means no sustainable investor provided funding, meaning the variable if they had another successful investment in the past would be irrelevant as they are not part of the investors.

Would highly appreciate your guidance here

[Clyde Schechter]

This strikes me as a clear case of a moderator that is undefinable when the funding from sustainable investors is zero. So an analysis with that moderator definitely ought to be restricted to those cases where sustainable investor funding > 0.