Moderation vs Mediation

[sedki zn]

Dear all,
I hope you are doing good
I'm studying the non monotonic relationships between Financial constraints ( FC), CEO stock options (SO) and strategic risk taking as follow:

1st chapter : the impact of FC on SO using threshold regression

2d chapter : the impact of SO on risk taking using quantile regression

3d chapter : the impact of FC on risk taking using threshold regression

Then I try to investigate the moderating effect of SO in the relation between FC and risk taking ( here SO is the threshold variable )

Is this econometrically doable ?

Because my supervisor says this can be treated as mediation and not moderation ! ( because SO is once dependent variable and once independent variable); however mediation can take place only with linear regression.

So my question is it correct to study the moderating effect of SO through the Threshold model in this case ?

Kind regards
Sedki

[sedki zn]

@Dear professors River Huang Richard Williams Clyde Schechter

I would appreciate your help

Best regards

[sedki zn]

Dear all,

anybody could help me please !

[Fei Wang]

The moderating effect of SO in the relation between FC and risk taking is technically doable, but you need to check if it's conceptually meaningful. Moreover, the arrangement of your chapters 1-3 may make readers feel that you're analyzing mediation rather than moderation -- I guess that's why your supervisor suggested mediation analysis. If you plan to focus on the moderating effect of SO, then chapters need to be reorganized to cater to your research purpose.

[sedki zn]

Dear Fei Wang
I would like to thank you for your answer.
1: concerning the arrangement of chapters, I tried to sheld light on the importance of SO in the relationship between FC and risk. That's why First i have studied this latter after investigating the 3 main relations.
2: I studied the moderation effect and not the mediation effect because for all three relationships I have used a non-uniform techniques and to my moderate knowledge there is not a model which studies the mediation with non-linear regression. Correct me if I'm wrong please.
3: for your opinion which is the best organisation I should follow ?

finally, I would like to know if i keep such organisation and method, can i defend my work ??!

best regards,

[Fei Wang]

I think your supervisor is the right person to give guidance on the structure of your thesis. I'm not familiar with mediation analysis (it's not a standard toolkit in economics.), but I'm able to find many papers on mediation in non-linear models, so I guess you may need to update your second argument. Anyway, if your focus is moderation, I don't think the latest development of mediation quite matters. I suggest you tell your supervisor you'd like to spotlight on moderation and see what's her/his advice.

[sedki zn]

Dear professor Fei Wang

thank you very much

actually my supervisor doesn't not matter if I go for moderation. He just have some concerns about the Jury the day of defending my dissertation.

so if I got you well, it is appropriate to keep the moderation analysis? If yes what argument could I use if the jury ask me why I did not use mediation instead?

Again, thank you for your kind answers

best regards

[Fei Wang]

I'd argue that looking into the moderation effect of SO in the relation between FC and risk taking is theoretically important (not familiar with your field, but there must be arguments about the importance of analyzing specific moderation effects) and empirically feasible -- That does NOT deny the validity of treating SO as a mediator, and it's simply another promising research paper and not your current topic.

[sedki zn]

Dear professor Fei Wang

thank you very much

So I can sum up, moderation is VALID method in my case ( even SO is once dependent variable and once independent variable ) . Technically is doable and there is not methodology issue such as multicollinearity or correlation ?

I really appreciate your kind replies
I'm grateful

kind regards

[Fei Wang]

Yes, you may need to justify your research from concepts and theories. It's not a technical issue.

[sedki zn]

Dear professor,

I can not find words to thank you
I really appreciate your help and kind answers

best regards,
sedki

[sedki zn]

Dear professor Fei Wang

i am really sorry bothering you again; but i have got another answer (following) that makes me really confused

here, i am begging you to clarify this to me. because as this sir says, the SO is a causal result of the IV (FC) AND a causal antecedent of th DV (risk taking) and this is the definition of the Mediation.
even you already told me that i can treat this as a moderation, i have got confused.

So please, even after this sir answer, moderation is VALID in my work ? ( i know it is valid if i ONLY treat this relationship BUT as a work of 3 chapters ( where first 2 chapter treat the causality) is still VALID ? in other words, the Jury will not ask me why i studied the moderation while in first two chapters i studied the causality.??

sorry for all these questions but i really need your help

Kind regards,

[Fei Wang]

My views are not contradicting the one you cited. I've said clearly in #4 that, your chapters 1-3 are preparing for mediation analysis, not for moderation. Basically you have two options. First, maintain the three chapters and do mediation. Second, do moderation after conceptually justifying why SO can be a moderator and reorganizing your chapters 1-3 such that they seem to prepare for moderation analysis.

Conceptually, a variable can be a mediator and a moderator at the same time. For example, I'm studying the effects of gender on wages. Education can be a mediator: girls gain less education in many developing countries, and thus attain lower wages. Education can be a moderator: relations between wages and gender are weaker for those better educated, implying that gender discrimination of wages is less serious as education grows.

As your chapters 1-3 pick the mediation role of SO, you'd better stick to it. But it doesn't rule out the possibility that SO can be a moderator in a DIFFERENT study -- That's why I suggest reorganizing your chapters if you insist on moderation.

[sedki zn]

thank you dear Fei Wang

i got it now.
actually i am not insisting on moderation as much as i'm running out of time and i need to submit my thesis in few days. thus, i cannot go for a mediation work now and reorganizing the chapters would be kinda complicated too in this short period.

So if i keep all like this with a theoretical justification for moderation, will be legitimate ?

kind regards

[Fei Wang]

I would stick to mediation and further complete what chapters 1-3 have missed for a comprehensive mediation analysis, and, at most, state with one paragraph in the conclusion that why SO can be a moderator and suggests future research on this.

If you are not familiar with mediation in nonlinear models, then just do a standard linear mediation analysis in chapter 4 -- at least keep logical coherence through the thesis.