Two-sample t-test¶

In this tutorial we will look at comparing two groups of participants. To do this we will use a two-sample t-test, using the dataset previously introduced.

We will focus on testing for a difference in overall task activation (contrast con_0009.nii) depending on whether participants responded with their left or right hand.

Bạn đang xem: Two-sample t-test¶

Specifying the model¶

1. Navigate to derivatives/second_level and make an empty directory where you will save your results. Name it something meaningful to you, e.g. two_sample_ttest_task.
2. Switch to this new directory in Matlab, using the file navigator (typically on the left hand side of the Matlab window).
3. We will now define our groups based on the information in participants.mat.
4. In your Matlab command window, load the file: load(‘participants.mat’).
5. You can view the contents of the file in the Matlab window. You will see that there is a variable called response hand which codes for which hand the participant used to make task responses. 0 corresponds to left, 1 to right.

6. We will define a variable selecting all participants who responses with their left hand. Copy and paste the following code into the Matlab command window and press enter:

   img_group1 holds all the directories pointing to left-handed responder’s con_0009.nii.

7. Xem thêm : ICD-9-CM to ICD-10-CM Conversion

   Now we’ll do the same for the right-hand responders:

   img_group2 holds all directories pointing to right-hand responder’s con_0009.nii.

8. Return to the main SPM menu window and select Specify 2nd level.

9. In the pop-up batch editor window, select your newly created output directory by clicking Directory and navigating to derivatives/second_level/two_sample_ttest_task in the selection box.
10. Define your statistical model by selecting Design Two-sample t-test
11. Select Group 1 scans Specify…. Bring up the edit window by clicking Ed in the file selection window.
12. From the edit window, you can now insert the variables you have saved in Matlab. Type in img_group1 and press Eval. This should fill the window with full paths to each left-hand responder’s con_0009.nii file. Click Accept Done.
13. Let’s do the same thing for right-hand responders. Group 2 scans Specify… Ed img_group2 Eval Accept Done.
14. From the drop-down menu, select SPM Stats Model estimation.
15. Navigate to Model estimation in the left-hand panel of the batch window.
16. Press Select SPM.mat Dependency Factorial design specification: SPM.mat file OK.
17. From the drop-down menu panel, select SPM Stats Contrast manager.
18. Within the Contrast manager, click on Select SPM.mat Dependency Model estimation: SPM.mat file OK.
19. You can now start specifying your contrasts of interest in Contrast sessions comparing left to right hand responders and vice versa.
20. Select Contrast sessions New: T-contrast.
21. Name your contrast, Name Specify… left > right.
22. Specify your contrast weight, Weights vector Specify… 1 -1.
23. Now, do the same for the reverse contrast – Contrast sessions New: T-contrast. Name Specify… right > left. Weights vector Specify… -1 1.
24. When you’re ready, save your batch and press to run your analysis.

Your design matrix should have two columns specifying left-hand and right-hand responders:

Viewing the results¶

Xem thêm : Brachial plexus injury

To view the results of your analysis, select Results from the SPM menu and select the SPM.mat file corresponding to your analysis. In our case, this will be in derivatives/second_level/two_sample_ttest_task. Choose a contrast to view, e.g. left > right.

SPM will now let you select masking and multiple comparisons correction. Select the following in the SPM window:

• Apply masking none
• P-value adjustment to control FWE
• P-value 0.05
• Extent threshold (voxels) 0

After going through these steps, SPM will display the results as an activation map and summary table:

We can see that participants who used their left-hand to make responses had stronger (more positive) activation in the contralateral (right) motor cortex.

Nguồn: https://buycookiesonline.eu
Danh mục: Info