thank you very much Andrew for explaining this , clear and useful!

Hello Andrew, thank you for the very helpful videos! I am wondering how to create manually a .mat or .txt file I could use as Time Course for multiple regression (taking into account for example reaction time, screen changes, etc.); for now the experiment script only records the timing of all these events but from what i understood it needs to be related to each scan in each run ? how do i get from the experiment's timings to the actual file with as many raws as there are scans in each run ? i hope my question is comprehensible thank you Julie

Hello Adnrew, first of all thanks for the great video. Is the adding motion regressors after alignment(motion correction) showing better result?

@AndrewJahn

8 жыл бұрын

+Woo Hyun Shim Hi Woo, usually they should yield a better result, since any activation correlated with motion (and, presumably, ONLY due to motion) should load onto those regressors; it should make the parameter estimates for your regressors of interest purer, since variance associated with motion will be accounted for in your model. That being said, there is a slight loss of power by including six additional motion regressors, but since your degrees of freedom are based on the number of time points across the session, and since the number of time points in a session is typically in the hundreds, this often isn't a problem. -Andy

@OrangeBlancKR

8 жыл бұрын

+Andrew Jahn Thank you much!!!!!!

Hi Andrew, Would you happen to have an instructional video about the structure of the .mat (multiple conditions) files? Thank you!

@AndrewJahn

4 жыл бұрын

Hi Lien, I don't have a video about that, but I did write a post about what's inside the SPM.mat structure: andysbrainblog.blogspot.com/2013/10/whats-in-spmmat-file.html -Andy

Hi Andrew, thank you very much. I have a stupid question- does first-level analysis equal to one sample T test? When do we use first-level analysis?

@AndrewJahn

4 жыл бұрын

Hi Jane, yes, you can create t-statistic images at the first-level, but they're not the same as the population-level inferences you are thinking of. Think of the 1st-level t-statistics as tests about the subject's average response to a condition, and the variability of that response. The group-level t-statistic, on the other hand, takes the average of the averages for each voxel for each subject, and compares that against the variability of each subject's mean. This is done for each voxel of the brain to generate a group-level t-statistic image. Best, -Andy

@janeli8937

4 жыл бұрын

@@AndrewJahn thank you very much