Moderation and mediation

Discuss statistics related things

by Mojtaba » Thu Jan 25, 2018 6:17 pm

Hi,

Thanks for the great software!

I have following questions about moderation and mediation:

1. is the output of indirect effect based on Sobel test?

2. it would be very useful to have the N in result of the simple slope analysis ;)

Thanks ahead,
Mojtaba
Mojtaba
 
Posts: 3
Joined: Mon Jun 26, 2017 10:32 am

by Ravi » Fri Jan 26, 2018 6:48 pm

Hi Mojtaba,

To answer your questions:

  1. Yes, by default the same test as Sobel is used (Delta method). However, it is often recommended to use the bootstrapped method instead because it gives better estimates of the standard errors.
  2. Why would it be useful to include N in the result of the simple slope analysis?

Cheers,
Ravi
User avatar
Ravi
 
Posts: 141
Joined: Sat Jan 28, 2017 11:18 am

by Mojtaba » Thu Feb 01, 2018 10:57 am

Hi Ravi,

Thanks for the reply.

Similar to other analyses, it is essential to know how many subjects have been assigned to each 3 groups (low, intermediate, high) because as you know N influences the effect size and this report shows the reliability and power of the results. In simple word, how trustable is the finding. For instance, in my data, I had only 4 participants in low grou, which means any comparison with this group is extremely weak.

Cheers,
Mojtaba
Mojtaba
 
Posts: 3
Joined: Mon Jun 26, 2017 10:32 am

by Ravi » Tue Feb 06, 2018 12:41 am

These values (-1SD, Average, +1SD) are based on the model, not on the individual cases. These are basically just linear formulas where we use a different value of the moderator (-1SD, Average, +1SD) to calculate the individual estimates. So there is not really such a thing as N here.
User avatar
Ravi
 
Posts: 141
Joined: Sat Jan 28, 2017 11:18 am


Return to Statistics