simple slopes

General help and assistance with jamovi. Bug reports can be made at our issues page: https://github.com/jamovi/jamovi/issues . (If you're unsure feel free to discuss it here)

by Lucie » Mon May 17, 2021 2:33 pm

I have ran a mixed-effects model using GAMlj. I also ran simple effects analysis which showed that my continuous IV significantly predicted my DV in one condition but not the other. How do I test whether the difference between these simple effects slopes is significant in Jamovi?

Thanks!
Lucie
 
Posts: 11
Joined: Thu Feb 25, 2021 5:31 pm

by mcfanda@gmail.com » Tue May 18, 2021 6:56 am

Hi,
if you have two conditions, the F-test of the interaction between your IV (independent variables) is the test you're looking for. The interaction test null hypothesis is that the simple slopes are all equal, so if the test is significant, you can say that the simple slopes are not all equal. In the case of two conditions, this means that the two simple slopes are different.

If you have more than two conditions, saying that the simple slopes are not all equal does not mean that they are all different. Thus, you need to look at the interaction coefficients, in the "Parameter Estimates" table. For instance, if your conditions are 1,2,3, and the continuous IV is called X, the interaction coefficient labelled "1-2 * X" informs you on whether the simple slope of X is different between condition 1 and 2, whereas the interaction coefficient "1-3 * X" informs you on whether the simple slope of X is different between condition 1 and 3.
User avatar
mcfanda@gmail.com
 
Posts: 260
Joined: Thu Mar 23, 2017 9:24 pm

by Lucie » Tue May 18, 2021 2:02 pm

Thank you for the quick reply - this is really helpful! So just to confirm, the F statistic of the interaction between the continuous and categorical variables (when there is only two conditions) is testing whether the simple slopes of each condition are significantly different? Would you only need to run a further test if you ran two separate simple effects models for each condition?
Lucie
 
Posts: 11
Joined: Thu Feb 25, 2021 5:31 pm

by mcfanda@gmail.com » Tue May 18, 2021 10:08 pm

It's not clear what the question is. Can you elaborate on that?
User avatar
mcfanda@gmail.com
 
Posts: 260
Joined: Thu Mar 23, 2017 9:24 pm

by Lucie » Thu May 20, 2021 1:08 pm

I have noticed that in some papers have assessed the difference between simple slopes as well as stating the F statistics. Is this because they ran two separate simple effects models?
Last edited by Lucie on Tue May 25, 2021 1:43 pm, edited 1 time in total.
Lucie
 
Posts: 11
Joined: Thu Feb 25, 2021 5:31 pm

by mcfanda@gmail.com » Mon May 24, 2021 4:37 pm

Comparing the simple slopes over and beyond the F-test is needed only if one has a categorical moderator with more than two groups. Otherwise, the test for comparing two simple slopes is the F-test (or t.test if you look at the interaction coefficient).
This is well explained in Aiken & West (1984) book or in Cohen, Cohen, Aiken & West (2003 and newer version)
User avatar
mcfanda@gmail.com
 
Posts: 260
Joined: Thu Mar 23, 2017 9:24 pm

by Lucie » Tue May 25, 2021 1:42 pm

Thank you for explaining this, this is really helpful!
Lucie
 
Posts: 11
Joined: Thu Feb 25, 2021 5:31 pm


Return to Help

cron