1
$\begingroup$

I am currently analysing data for my Bachelor thesis. I want to find out if people's decision-making ability (X) has a direct and/or indirect effect through their tendency to overeat (M) on their BMI (Y).

I have conducted a simple mediation analysis using the macro PROCESS in SPSS. As far as I have (hopefully rightfully) concluded the effects are not significant. However, since the age range of the tested participants is very broad, I decided to include the age as a covariate.

I am not sure how to exactly interpret the results now. I have included the SPSS output in a Word document below to make things more visual. In the part where it says outcome variable “BMI”, “alter” (=Age) has a coefficient of 0.086 with a p value of .0103. Does that mean that the age has a significant effect on the BMI?

Also in the total effect model found below it says that money_sum (=decision-making ability) has a coefficient of -.0005 (p= .3271) and age has a coefficient of .0860 (p= .0081). Again does that mean that age has a significant influence on the BMI?

But how can I interpret then the total (effect = -.0005), direct (-.0005) and indirect models?

Does age not have an effect after all? I am a bit confused now and I really appreciate your answers to help me get rid of this confusion.

enter image description here

enter image description here

enter image description here

enter image description here

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

The path a and path b of your model are not significant. Your indirect effect (path a*b) also included zero, indicating no mediated effect. If removing age and adding age to your model gives you different path coefficients, then that should be of an interest. Otherwise significant age in your model doesn't change your conclusion from the mediation model.

Good luck.

Improve this answer
$\endgroup$
3
  • $\begingroup$ Thanks a lot for your answer. Would you say that it does NOT make sense to include age as a covariate since the paths are not significant anyway? Are covariates only included if there is a significant relationship or can it be reasonable to use in my case (not significant paths) as well? Also, you say that significant age in my model does not change my conclusion from the mediation model. Now, I get that but what can it then mean that age is significant? Does it mean that another model fits better. Sorry for so many questions and thanks for your help $\endgroup$
    – NoName
    Commented May 31, 2018 at 19:00
  • $\begingroup$ No to the first and No to the second. Significant covariates (age in your case) in mediation model means nothing. Do not interpret it. Just interpret the mediation model as I explained above. Adding and removing covariates to your mediation model is like adjusted and unadjusted mediation analysis. if you found two different results from your mediation analysis after adding and removing age then that is another story, otherwise forget it. $\endgroup$
    – Million
    Commented May 31, 2018 at 22:08
  • $\begingroup$ Thank you so much for your help. Now I can write my conclusion. $\endgroup$
    – NoName
    Commented Jun 1, 2018 at 10:37

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.