Dear experts,

I have 4 conditions (lets say : A B C D). Also, I have 2 groups (M and N).

For the PPIs of each subject I used this contrast : A>B (1 -1).

Then I performed one sample t-tests (for each group) and two-sample t-tests (group comparisons eg M > N using 1 -1 contrast for the comparison).

Then I visualised my results. 

I found some statistically significant activations at cluster level both for the one sample and two sample t-tests.

Question 1) What can I say about the significant results (clusters) at the one-sample t-tests? How I interpret my results? Can I say that the significant region increases its relationship with the seed region under the A vs B condition?

Question 2) At the two-sample t-tests, I found significant clusters for the M>N comparison. Can I say that the X region of M group increases its relationship with the seed region under the A vs B ?

Thank you in advance,
s.l

(1) For one-sample t-tests -- you can say that the slope of the relationship between two regions increases/decreases relative to baseline.

(2) For two-sample t-tests -- you can say that the increase/decrease in the slope of the relationship between two regions is greater in one group than the other.

Best,
Donald

Dear Donald,

Thank you for your response.

I know that I can plot the slope for 1 subject and see if there is increase or decrease but how can I see if there is an increase or decrease on the relationship based on the one-sample and two-sample t-test results?


Thank you again,
S

In the one-sample t-test:
A positive beta means the slope is positive.
A negative beta means the slope is negative.

In the two-sample t-test:
Regions that are positive for contrast [1 -1] means that group 1 slope > group 2 slope. You then have to go back to the one-sample t-tests to see if the slopes in each group are positive or negative.
Regions that are positive for contrast [-1 1] means that group 1 slope < group 2 slope. You then have to go back to the one-sample t-tests to see if the slopes in each group are positive or negative.

Hope this helps.

Best,
Donald

This is really helpful! thank you.

one last question: in order to see if the beta is positive or negative eg for the one sample t-test, I look inside SPM (which is generated by the one sample t-test) and then in SPM.Vbeta? If yes, which valus indicate the positive or negaative effect?

I'd just open up the beta images in checkreg or look at the spmT images. Positive values mean positive slope, Negative values mean negative slope.

thank you again!

Dear Donald,

I have one more question.

I want to see the strength of the PPI regressor in a specific ROI.

I was thinking about using Marsbar in order to extract beta weights of the PPI regressor (using the con image of each subject that corresponds to the PPI regressor or in other words the con image that was created using the 1 0 0 contrast).

By doing this, I will get the beta weights for the interaction term for each subject and then I could perform a regression analysis between these values (strength of interaction) and some behavioural data.

Do you think that this is reasonablee and correct?

Thank you in advance,
S.