Re: To detrend or not to detrend

[Katie Gates]

Hi Henry, 

This question falls a bit outside the scope of how to best implement gimme or reporting of bugs. 

I would suggest consulting literature on non-stationarity in EMA studies and how others handle it when it is not linear.

All the best,

Katie

On Tue, Dec 16, 2025 at 2:39 PM Henry Whitfield <mindfuln...@gmail.com> wrote:

Hi Katie,

I tried you in another thread but didn't hear back. I'm sure you are busy - especially this time of year.

Hopefully I can put this as a quick question.

You might recall I've been writing a paper using S-GIMME and CS-GIMME to look at different treatment effects. I'm aware this violates the stationarity assumption.

My quick question is whether detrending is always the best option in this scenario, and does it fix the stationarity requirement? All psychotherapy is about change. 

Is it OK we ran GIMME on such a dataset?

I have detrended using the software you suggested in a different thread. It didn't change the clustering in the S-GIMME subgroups diagram (based on the similarities matrix). It did move the line that divides the two groups. 

I would like some guidance on what is the best option. The data I have don't follow continuous trends up or down - more the occasional small jump at certain points in the time series. 

Any guidance would be really appreciated.

Many thanks,

Henry