mediation regression command?

[Kaas de vries]

Dear stata community,

I am struggeling a bit with the following stata exercise. I am looking at the relationship between perceived peer income and household's borrowings. Further, I would like to test if the big five personality traits influence this relationship.This is what it looks like

	big five personality factor
perceived peer income	>>>>>>>>>>>>>>>>	collaterized debt

Now I want to use OLS regression in Stata. What is the correct command for performing this in stata?

Yours sincerely,

MrLlewisme.

[Roman Mostazir]

Welcome to the Statalist. A little google search with "regression with Stata" terms brings up plenty of materials to start with and you are expected to do your own research first. You are also expected to read the FAQ section thoroughly to learn how to make a meaningful post, provide data example using -dataex- and use of code delimiters while posting and rules about explaining unfamiliar terms (i.e. big five personality factors !!).

On your question, assuming -'income' is your outcome variable, Stata has a very simple OLS regression command:

regress income debt

If you want to explore how 'characteristics' influence the relationship, you need an interaction term:

regress income c.debt##c.characteristics

As you have not provided any data example, I am only assuming that both of your independent variables are continuous. My advice is go through the Stata help file for regress: type help regress and learn through. Also learn the use of post regression -margins- command which will help you to explore the relationships.

By the way your 'mediation' term in title and goal to build OLS are misleading. Influence in relationship and mediation have separate meanings. If you want to explore mediation, you need different model rather simple OLS. Google mediation in Stata . For mediation you may need programmes like ivregress or Stata's sem/gsem suits. There are other user written commands available too.

[Kaas de vries]

Dear Roman,

I am sorry for not having read thoroughly the forums guidelines. I did do some research on google, but I was unsure on how to implement the moderation (yes, I made a mistake there as you noticed), aspect.

Thanks for your time and advice. As you can tell, I am really new to stata. You gave me some good leads to work with.

Yours sincerely,

[Roman Mostazir]

I think you also need to have clear concepts about your research questions. Moderation (#3) is different than mediation. This paper might be of your interest:

Addressing Moderated Mediation Hypotheses: Theory, Methods, and Prescriptions., Preacher KJ, et al. 2007

[ericmelse]

Dear Klaas,
For a paper in The Stata Journal see: R. Hicks & D. Tingley Causal mediation analysis.
You could also study the UCLA web pages on this subject: How can I do moderated mediation in Stata?

[Kaas de vries]

Dear Roman and Eric,

Thanks you both for your suggestions. Eric, I especially found this suggestion helpful "You could also study the UCLA web pages on this subject: How can I do moderated mediation in Stata?".

Yours sincerely,

[ericmelse]

Here is another (open access) reference for those who are interested this subject, published in Frontiers in Psychology: 2017 R Agler P de Boeck On the Interpretation and Use of Mediation Multiple Perspectives on Mediation Analysis.