I am investigating UK university performance, & comparing 96 disciplines on
proportion of students with highı performance.
There are 4 categories: 1st, 2.1, 2.2, <2.2
I produce 2 measure for each of the 96 disciplines:
Proportion achieving excellent degrees, i.e. 1st class, E; and proportion
achieving very good degrees, i.e. 1st OR 2.1, V.
One MIGHT expect that disciplines with high E, would also have high V.
WRONG. There are 10 disciplines with E>=20 and 27 disciplines with V> 66.7.
Medicine is the ONLY discipline which falls into the top 10 for E, AND the
top 27 for V.
This is an interesting and counterintuitive finding that I would like to
reinforce by giving the correlation between E and V.
BUT E and V are clearly NOT independent
Pearson r = .36, Kendallıs tau = .26, Spearmanıs rho = .38.
I also have students incoming high school grades, UCAS points scores, U.
Partial correlation E,V correcting for U = .023.
It seems to me realistic to claim that excellent and v. good performance
differences across disciplines depend differently on other factors.
BUT still worried about non-independence of E and V
PLEASE HELP with suggestions of better analyses
It is the case that multiple regression of E and V on relevant factors give
different results. BUT do not want to fall into fallacy of factor produces
different effects on E and V, therefore V and E are different. If one does a
multivariate regression with both E and V as response variables and looks
for interaction, the non-independence of E and V is again a problem.
HELP
Diana
Professor Diana Kornbrot
School of Psychology
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
email: [log in to unmask]
web: http://web.mac.com/kornbrot/iweb/KornbrotHome.html
voice: +44 (0) 170 728 4626
fax: +44 (0) 170 728 5073
Home
19 Elmhurst Avenue
London N2 0LT, UK
voice: +44 (0) 208 883 3657
mobile: +44 (0) 796 890 2102
fax: +44 (0) 870 706 4997
|