search

berkeley-stat159 / project-iota

BSD 3-Clause "New" or "Revised" License
1 stars 6 forks source link

working on linear modeling #57

Closed Jay4869 closed 8 years ago

Jay4869 commented 8 years ago

Re-working on the linear modeling because we generated new convo-response

  1. Only doing simple regression on the block test cond005, and figure out the difference between 2-black and 0-black, which shows how the long term memory works in the brain. Looking for what result is
  2. Adding block test and contrast test cond003 in the linear regression and apply hypothesis test to check how the betas work, and make p-value map for selecting sub-region of brain
Jay4869 commented 8 years ago

@matthew-brett For the simple regression on the block test, I think I also need to add start and end instruction cond001 and cond004, is that correct? I just want to double check

matthew-brett commented 8 years ago

You can add these, if you like, but I suspect they won't change the result very much. The general idea should be to add regressors that you believe may explain variance, and then test to see if they do explain a large amount of variance. In this case, you'd have to look, but I suspect they would have rather small influence compared to the block regressor.

Jay4869 commented 8 years ago

done!