2

I have data on room assignments for students at a university, as well as certain characteristics on the students in the rooms, e.g. test scores, gpa, nationality, gender. I am told room assignments are random, but how can I test this? My thinking is if I can predict an individuals characteristics based on the average characteristics of a room than it is not random. So essentially I would have to run some regression where the right hand side are average characteristics of some of the room, and then do an F-Test to make sure coeffiecients are not significant by chance? I am not sure if this is correct or maybe there is smarter way to go about this.

Eric
  • 21
  • 1
  • What data do you have on the rooms, when you talk about room characteristics? – Silverfish Oct 18 '15 at 01:40
  • 1
    This is a good candidate for a permutation test. Its a monte carlo method where you create random room assignments from the students and look at the resulting distribution of the (as yet unspecified) assignment statistics. If your actual observed statistics are on the extreme end of the simulated statistics, then its unlikely that they were assigned randomly. –  Oct 18 '15 at 02:32
  • When I said average characteristics of the room, I meant average characteristics of the students in the room, that is gpa, SAT scores, toefl scores, gender, etc. I will look into the permutation test, thanks. – Eric Oct 18 '15 at 04:05

0 Answers0