- Joined
- Nov 22, 2006
- Messages
- 673
- Reaction score
- 2
I'm testing a little "match list analyzer" program algorithm that essentially breaks down matches (IM only, including other specialties at this time would take a lot more research) into 4 "tiers" and normalizes all ratings to a 4.0 scale.
A super crude analysis of the IM matches of various universities from 2007/2006.
1. Harvard - 3.85
2. Yale - 3.67
3. Hopkins - 3.61
3. Mayo - 3.60
3. UPenn - 3.56
4. UCSF - 3.34
5. WashU - 3.25
5. Stanford - 3.23
6. Pritzker - 3.13
7. Baylor - 3.04
7. Vanderbilt - 3.00
8. Northwestern - 2.85
8. UCLA - 2.84
8. U Mich - 2.82
8. U Pitt - 2.80
9. AECOM - 2.57
10. RWJ - 2.45
10. UCI - 2.44
10. USC Keck - 2.43
11. UMDNJ-NJMS - 2.28
11. SUNY D - 2.24
12. NYMC - 2.13
These are the only schools I've checked, but does it match up with your personal assessments of their relative IM matches?
Any schools you might be particularly interested in seeing entered into the program?
*update* fine tuning algorithm, updated numbers that changed
*update* since a couple people have asked "how much of a difference each point makes" I did a couple (non-statistically rigorous) manual tests, and there's a base error of around +/- .05
Translation, if a school is within .05 of another school (technically up to .10, but that's highly unlikely) their match lists can be considered "about the same."
A super crude analysis of the IM matches of various universities from 2007/2006.
1. Harvard - 3.85
2. Yale - 3.67
3. Hopkins - 3.61
3. Mayo - 3.60
3. UPenn - 3.56
4. UCSF - 3.34
5. WashU - 3.25
5. Stanford - 3.23
6. Pritzker - 3.13
7. Baylor - 3.04
7. Vanderbilt - 3.00
8. Northwestern - 2.85
8. UCLA - 2.84
8. U Mich - 2.82
8. U Pitt - 2.80
9. AECOM - 2.57
10. RWJ - 2.45
10. UCI - 2.44
10. USC Keck - 2.43
11. UMDNJ-NJMS - 2.28
11. SUNY D - 2.24
12. NYMC - 2.13
These are the only schools I've checked, but does it match up with your personal assessments of their relative IM matches?
Any schools you might be particularly interested in seeing entered into the program?
*update* fine tuning algorithm, updated numbers that changed
*update* since a couple people have asked "how much of a difference each point makes" I did a couple (non-statistically rigorous) manual tests, and there's a base error of around +/- .05
Translation, if a school is within .05 of another school (technically up to .10, but that's highly unlikely) their match lists can be considered "about the same."