methodological question about s-GIMME

[Romi Adiel]

Hi all,

I have a methodological question about s-GIMME.
Can s-GIMME accommodate, within the same model, a single 3-level ordinal variable together with several continuous variables in an EMA design?

Concretely, I have intensive longitudinal (EMA) data collected ~8 times per day over several weeks. At each prompt, participants report:

• Repetitive negative thinking (RNT) type: a single ordinal variable with three categories (rumination / worry / both)

• Negative emotions: multiple continuous variables rated on a 0–100 scale

The goal is to model within-person temporal and contemporaneous associations between RNT type and negative emotions.

From a technical perspective, I am unsure whether s-GIMME:

• can directly handle mixed measurement levels,

• requires treating the ordinal variable as continuous,

• or would require binarization or another workaround.

I would appreciate any advice on whether this is feasible within s-GIMME, or whether a different modeling framework would be more appropriate.

Thank you very much in advance!

[Katie Gates]

This is a great question! 

Currently, S-GIMME (and any form of GIMME) can't handle ordinal or categorical variables if they are allowed to be outcomes (i.e., have an arrow pointing towards it). 

If you are OK with having them only be predictors, such that RNT type can predict negative emotions but not the reverse, then gimme performs well if you set the categorical variables to be exogenous. This can be done using this argument; 'exogenous = c("rumination", "worry", "both") where each variable is a 0/1 indicator of the specific type of RNT. Please note that a reference category of "no RNT" assumed here. If there is no time points where they are outside of RNT thinking, then you'd have to choose which of the 3 to use as a reference category. 

I wish we had a better solution where RNT could also be an outcome, but unpublished simulation studies have shown that the path selection becomes noisy and unreliable when categorical variables are the outcomes. 

Best,

Katie