I'm new to skin conductance analysis and to pspm.
I'm trying to analyse data from a within-subject experiment where we collect skin conductance response for each sbj in 2 sessions (day 1 and day 2). In both sessions, participants are presented with the same 4 experimental conditions. I would like to see if response across participants changes between day 1 and day 2, in the specified conditions.
I would like to know if it is possible to create a GLM model, where I enter both sessions in one model (under Data&Design). Or should I run 2 separate models per each day session?
Many thanks for your help.
Maria
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I made some progress on this, but I'm really not confident I'm doing it correctly. I now created a GLM model with 2 sessions, specifying a different condition file per each session (1 and 2). The names in each condition file are identical and they correspond to the 4 experimental conditions, but onsets are obviously different.
I would like to compare conditions across sessions, that is, if I have 4 conditions A B C D and E, I can compare A1vs A2, B1 vsB2, C1 vs C2, D1 vs D2, and E1 vs E2. Is the specification of the model correct, based on the contrasts I'd like to run?
Also, is it possible to extract the parameters per each condition per each sbj to run a repeated measure ANOVA?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The first important bit is to take care of data normalisation. Either you turn it off and run two different models, one for each session. Or you can turn it on or off just as you like, and run both sessions in one model. The latter is probably more sensitive as normalisation removes irrelevant between-subject variance in overall response amplitude (partly due to peripheral factors).
So, specifying one model with both sessions is good. But if you want to compare the two sessions, the events must have different names in the two sessions. That is, you will have events like "A1", "A2", "B1", "B2", ... where A1 and B1 occur only in session 1, and A2/B2 only in session 2.
Importantly, both condition files must be identical, and therefore the file for session 1 must contain definitions for A2, B2, and vice versa, but the onsets for these will be empty. That is, only the onsets corresponding to A1, B1, ... will be numeric vectors for the first session, and the onsets corresponding to A2, B2, ... will be empty vectors.
At the end, you can either run contrasts on the group level, or export the parameter estimates using the export function, and do a full ANOVA model in a statistics software of your choice.
Hope this helps
Best
Dominik
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Dear Dominik,
I have a similar question with regard to the inversion model. As my experiment also includes anticipatory SCR (with a cue onset before each stimulus onset), I'm trying to implement the dynamic causal modelling for aSCR. I applied something similar as the GLM, with 2 sessions and onsets as you suggested to do for the GLM. However, I get an error message saying "Error using cat
Dimensions of matrices being concatenated are not consistent.". Any idea why?
Many thanks for your help.
Best
Maria
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Dear Dr Bach
I'm new to skin conductance analysis and to pspm.
I'm trying to analyse data from a within-subject experiment where we collect skin conductance response for each sbj in 2 sessions (day 1 and day 2). In both sessions, participants are presented with the same 4 experimental conditions. I would like to see if response across participants changes between day 1 and day 2, in the specified conditions.
I would like to know if it is possible to create a GLM model, where I enter both sessions in one model (under Data&Design). Or should I run 2 separate models per each day session?
Many thanks for your help.
Maria
Dear Dr Bach,
I made some progress on this, but I'm really not confident I'm doing it correctly. I now created a GLM model with 2 sessions, specifying a different condition file per each session (1 and 2). The names in each condition file are identical and they correspond to the 4 experimental conditions, but onsets are obviously different.
I would like to compare conditions across sessions, that is, if I have 4 conditions A B C D and E, I can compare A1vs A2, B1 vsB2, C1 vs C2, D1 vs D2, and E1 vs E2. Is the specification of the model correct, based on the contrasts I'd like to run?
Also, is it possible to extract the parameters per each condition per each sbj to run a repeated measure ANOVA?
Hi Maria
thanks for your question.
The first important bit is to take care of data normalisation. Either you turn it off and run two different models, one for each session. Or you can turn it on or off just as you like, and run both sessions in one model. The latter is probably more sensitive as normalisation removes irrelevant between-subject variance in overall response amplitude (partly due to peripheral factors).
So, specifying one model with both sessions is good. But if you want to compare the two sessions, the events must have different names in the two sessions. That is, you will have events like "A1", "A2", "B1", "B2", ... where A1 and B1 occur only in session 1, and A2/B2 only in session 2.
Importantly, both condition files must be identical, and therefore the file for session 1 must contain definitions for A2, B2, and vice versa, but the onsets for these will be empty. That is, only the onsets corresponding to A1, B1, ... will be numeric vectors for the first session, and the onsets corresponding to A2, B2, ... will be empty vectors.
At the end, you can either run contrasts on the group level, or export the parameter estimates using the export function, and do a full ANOVA model in a statistics software of your choice.
Hope this helps
Best
Dominik
This is extremely helpful Dominik, thank you so much.
Best
Maria
Dear Dominik,
I have a similar question with regard to the inversion model. As my experiment also includes anticipatory SCR (with a cue onset before each stimulus onset), I'm trying to implement the dynamic causal modelling for aSCR. I applied something similar as the GLM, with 2 sessions and onsets as you suggested to do for the GLM. However, I get an error message saying "Error using cat
Dimensions of matrices being concatenated are not consistent.". Any idea why?
Many thanks for your help.
Best
Maria