0
$\begingroup$

I am unsure whether my current choices of statistical tests are correct for what I am trying to explore. I am looking at whether there is a relationship between unhealthy eating behaviors and task-unrelated thoughts as my research question. I am also examining whether different social contexts affect that relationship or affect each of those variables individually and whether different emotionality levels affect the relationship or each variable individually as my two hypotheses.

I am using a linear regression to test if there is a relationship, a moderated regression to test social contexts' effects and another moderated regression to test emotionality's effects. All variables are measured numerically.

I have considered doing a mediation instead of a moderation but I am not sure if this is right, so I haven't used it. The more research I do the more I feel I have gotten all the tests wrong!

It is for my undergrad dissertation if that helps or makes a difference and my supervisor has not provided any support other than to say she doesn't think I've chosen the right tests so now I am near my deadline I'm very stressed its wrong! She mentioned doing a multiple regression but was quite vague about whether that would be better or not.

$\endgroup$
0

1 Answer 1

1
$\begingroup$

Distinguishing mediation and moderation can be a little confusing (and both may be happening!).

To be a little more general, let's call the two independent variables IV1 and IV2 and the dependent variable DV.

Moderation says that the relationship between IV1 and the DV is different at different levels of IV2 and that the relationship between IV2 and the DV is different at different levels of IV1. Moderation is usually evaluated by adding an interaction term. For example, if you want to model weight as a function of sex and height, there might be moderation, if additional height increases weight more (or less) in men than in women.

Mediation says that IV2 mediates the relationship between IV1 and DV. Most people define it as lowering or eliminating the relationship. An example of mediation is if you wanted to look at damage done by a fire based on number of firefighters who attend the fire. You would want to add size of fire as a mediator, because bigger fires both cause more damage and get more firefighters.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.