sdm-help-list > Question on covariates
Showing 1-3 of 3 posts
Display:
Results per page:
Jun 29, 2020  09:06 PM | Alex Rainer
Question on covariates
Hi Anton,

I would like to include subject age and sex across studies as two covariates (as I want to rule out the effects of sex and age in my meta-analysis findings). I added these variables as the 9th and 10th column in the sdm_table.txt. When I run the Mean portion of the analysis, I pick sex and age from the two drop-down boxes as the covariates. Could you tell me whether this procedure is correct?

A related question is that there appears to be a limit of 2 covariates. If users want to use 3 or more variables as covariates, how would they go about including them?

I have looked at other posts on covariates but am unsure whether the toolbox has changed regarding this function since the answers were posted.

Thank you for your help and thank you for the great toolbox!

Alex
Jun 30, 2020  08:06 AM | Anton Albajes-Eizagirre - FIDMAG - Germanes Hospitalaries
RE: Question on covariates
Hi Alex,

Although the GUI only provides interface for up to two covariates, using the command line interface you can use up to four (for now) covariates.
Alternatively, you can also write the command in the batch window and then run the batch.

Here's the brief explanation of the manual:
https://www.sdmproject.com/manual/?show=...

In order to see an example of how to write the command, you can first perform a mean with two covariates, see the format it uses to specify them, and then expand the command to include more covariates.

Hope this helps, best,

Anton

Originally posted by Alex Rainer:
Hi Anton,

I would like to include subject age and sex across studies as two covariates (as I want to rule out the effects of sex and age in my meta-analysis findings). I added these variables as the 9th and 10th column in the sdm_table.txt. When I run the Mean portion of the analysis, I pick sex and age from the two drop-down boxes as the covariates. Could you tell me whether this procedure is correct?

A related question is that there appears to be a limit of 2 covariates. If users want to use 3 or more variables as covariates, how would they go about including them?

I have looked at other posts on covariates but am unsure whether the toolbox has changed regarding this function since the answers were posted.

Thank you for your help and thank you for the great toolbox!

Alex
Jun 30, 2020  04:06 PM | Alex Rainer
RE: Question on covariates
Hi Anton,

Fantastic! Thank you so much for your quick response. Could you help me with a small thing regarding the number of threads? I'm running the Mean portion in the terminal (on a Linux machine). I have successfully run the preprocessing. I changed the number of threads in the sdmpsi_params.xml from 0 to 8. Somehow the terminal still displays "using 1 threads". Did I do something incorrectly? I ended up running the command in the GUI and it does use 7 threads.

Another thing is that the covariates accept both numeric and text values, is this correct? For instance, I have average age as one of the numeric covariates and smoothing parameters (e.g., 6mm, 8mm) entered as a text input.

** EDIT **

I was able to run the Mean analysis with 4 covariates. The results look different from the one without accounting for the covariates. My question is: what does including the covariates actually do? I was under the impression that it would ensure the effects of any of the covariates are removed (e.g., in the case of age as the covariate then the results would have the age effect regressed out and thus control for the age effect). Could you please let me know whether I am correct or mistaken here?

A related question is that I've looked at several studies using SDM and they all seem to have used meta-regression to examine the effects of their covariates but none appears to include the covariates as part of the main analysis. Is there a theoretical reason for this?


Alex