help
help > RE: Longitudinal Analysis Error
Jun 5, 2018 01:06 PM | Alfonso Nieto-Castanon - Boston University
RE: Longitudinal Analysis Error
Hi Pavel,
Yes, my impression is that this indicates some problem in the condition specification. In your case I imagine that you would need to define only three conditions (one for each time-point). It would be something like:
1) In Setup.Basic enter the actual number of subjects, and the actual number of sessions (timepoints) that you scanned for each subject
2) In Setup.Conditions create three conditions (e.g. T1, T2, and T3), and associate each condition for each subject to the corresponding session (e.g. if subject 7 only had T1 and T2 data, then the number of sessions in step (1) above already specified that you only had two session worth of data, so in Setup.Conditions you will only seed session#1 and session#2 for this subject, and you just need to associate condition T1 with session #1 and condition T2 with session #2)
3) also in Setup.Conditions select the 'allow missing data' option, in order to let CONN know that it is allright that some subjects may have some missing conditions (e.g. not all subjects may have data associated with the T3 condition)
4) In Setup.SecondLevelCovariates create two group covariates (e.g. GroupA and GroupB), each identifying using 1/0's the subjects within each group
Let me know if this makes sense
Best
Alfonso
Originally posted by Pavel :
Yes, my impression is that this indicates some problem in the condition specification. In your case I imagine that you would need to define only three conditions (one for each time-point). It would be something like:
1) In Setup.Basic enter the actual number of subjects, and the actual number of sessions (timepoints) that you scanned for each subject
2) In Setup.Conditions create three conditions (e.g. T1, T2, and T3), and associate each condition for each subject to the corresponding session (e.g. if subject 7 only had T1 and T2 data, then the number of sessions in step (1) above already specified that you only had two session worth of data, so in Setup.Conditions you will only seed session#1 and session#2 for this subject, and you just need to associate condition T1 with session #1 and condition T2 with session #2)
3) also in Setup.Conditions select the 'allow missing data' option, in order to let CONN know that it is allright that some subjects may have some missing conditions (e.g. not all subjects may have data associated with the T3 condition)
4) In Setup.SecondLevelCovariates create two group covariates (e.g. GroupA and GroupB), each identifying using 1/0's the subjects within each group
Let me know if this makes sense
Best
Alfonso
Originally posted by Pavel :
Hi Admins,
Thank you for creating such a nice tool!
I’m running an analysis of a longitudinal study. There are two groups with three time points and an attrition rate that causes the groups to be uneven all time points. They’re different by a few subjects at each point.
I setup 6 Conditions to assign to each group (3 time points x 2 groups) and assign them the functional data accordingly. I check the “allow missing data” box in the setup as well.
The preprocessing, denoising and 1st level was error free. However, I get an error when I run the 2nd level analysis to compare BETWEEN groups. For example, group 2 > group 1 for any of the time points, I get an “exceed matrix size” error. I also notice that the design matrix counter in the tool goes to 0. The tool works fine when I compare within groups.
Can you help me? Did I make a mistake in settings up the conditions?
Thank you,
Pavel
Thank you for creating such a nice tool!
I’m running an analysis of a longitudinal study. There are two groups with three time points and an attrition rate that causes the groups to be uneven all time points. They’re different by a few subjects at each point.
I setup 6 Conditions to assign to each group (3 time points x 2 groups) and assign them the functional data accordingly. I check the “allow missing data” box in the setup as well.
The preprocessing, denoising and 1st level was error free. However, I get an error when I run the 2nd level analysis to compare BETWEEN groups. For example, group 2 > group 1 for any of the time points, I get an “exceed matrix size” error. I also notice that the design matrix counter in the tool goes to 0. The tool works fine when I compare within groups.
Can you help me? Did I make a mistake in settings up the conditions?
Thank you,
Pavel
Threaded View
| Title | Author | Date |
|---|---|---|
| Pavel | May 25, 2018 | |
| Alfonso Nieto-Castanon | Jun 5, 2018 | |
| Pavel | Jun 6, 2018 | |