By a Nose

Jockeys wearing academic regalia racing on horseback toward "Top Ranked University" finish line

Appendix to:

Graduate School and College Excellence
Does research reputation influence undergraduate rankings?

Alexander Hicks, Professor of Sociology


Vol. 8 No. 2
October/November 2005

Return to Contents


By A Nose
Jockeying in the Rankings Race

The Current Standings

Whither the NRC Study?

"I am not going to change our methods of calculation just in order to try and achieve a ranking higher than another institution."

"Part of the reason educational reputation is so important is because people—students, faculty, and administrators—derive much of their status from the status of their institution."

Graduate School and College Excellence
Does research reputation influence undergraduate rankings?

Peer Scorings and Rankings of Colleges and Graduate Programs and Research

Appendix


The “Lecture Track” Reconsidered
Professional identity and aspiration among non-tenure-path faculty

Tales from the Lecture Track: Kristin Wendland, Music

Tales from the Lecture Track: Sheila Tefft, Journalism


Virtue and the Stewardship
of Academic Freedom

Reflections on ambition, conversation, and community

Endnotes

 

I construct an adjusted program score that sums scores for the same programs but then divides the sum by the number of the 12potential programs actually present at each school. I adjust two more scores because the two preceding ones are flawed by their lack of humanities programs beyond English and History (not to speak of the hybrid humanities/social sciences quality of history). To adjust for this thinness in the U.S. News’ attention to the humanities, I construct what I call a balanced program proxy as follows: I average (a) the four social science scores, (b) science scores for Biology, Chemistry, Mathematics, and Physics,   and (c)  a simulation of four humanities scores. This last component consists of History counted once and English added in three times, the extra two as indicators for missing humanities programs (in the languages, etc.).  This proxy takes two forms. One is the simple balanced program proxy, which is divided by 12 for the number of program scores in the average. The second is adjusted program proxy, which is divided by the number of actual programs present at a school (with English always counted three times). These measures all correlate at least 0.812 with each other, on average about 0.895. .  I also construct an average score for these four scores, which correlates on average 0.957 with the four.

Note that correlations of the simple program score, adjusted program score, balanced program proxy, and adjusted program proxy scores and peer college rankings are 0.839, 0.849, 0.777 and 0.865, respectively

Note that public Universities in the data set (with their rankings on the average of Average are : UC-Berkeley (1), U-Mich (7), Wisconsin (10), UNC (11), UCLA (13),  Illinois-Champaign-Urbana (17) UCSD(19), UVA (22), Cal Tech (23),  and William and Mary (37).

Note regarding correlations between the dimensions of college ranking  and  “peer assessment”  that correlation in the -.660 to -.368 range are considerably lower than the 0.85-ish levels of correlations for graduate/research and student selectivity because comparisons of squared correlation coefficients provide a better basis for comparing the relative magnitudes of correlations than the raw correlations. For example, -.837 squared and -.660 squared are  0.701 and .436, respectively: although the -.837 is only 1.27 times -.660 (in absolute value),  0.701 is 1.61 times as large as 0.436.


Table A1: Peer Assessment and Graduate Programs Scores for
Top 40 NURs 

NUR

USN RANK

simple

adjusted

balanced

AdjBal

Average

Stanford

4.90

4.79

4.79

4.81

4.81

4.80

Harvard

4.90

4.58

4.58

4.78

4.78

4.68

Princeton

4.90

4.60

4.60

4.68

4.68

4.64

Yale

4.90

4.35

4.35

4.57

4.57

4.46

MIT

4.90

3.55

4.73

2.75

4.71

3.94

UC-Berkeley

4.80

4.82

4.82

4.86

4.86

4.84

Columbia

4.70

4.17

4.17

4.34

4.34

4.25

Cal Tech

4.70

2.66

4.56

1.93

4.64

3.45

U. Chicago

4.60

4.34

4.34

4.54

4.54

4.44

Michigan

4.60

4.34

4.34

4.43

4.43

4.39

Cornell

4.60

4.28

4.28

4.34

4.34

4.31

U. Penn

4.60

3.97

3.97

4.15

4.15

4.06

Duke

4.60

3.88

3.88

4.00

4.00

3.94

J. Hopkins

4.60

3.83

3.83

4.04

4.04

3.94

Northwestern

4.40

3.88

3.88

3.96

3.96

3.92

Brown

4.40

3.71

3.71

3.73

3.73

3.72

Dartmouth

4.40

1.74

2.99

1.28

3.08

2.27

UCLA j

4.30

3.97

3.97

4.03

4.03

4.00

U. VA

4.30

3.44

3.44

3.64

3.64

3.54

Carnegie Mel

4.30

3.05

3.05

2.83

2.83

2.94

Wisconsin

4.20

4.24

4.24

4.26

4.26

4.25

UNC-Chapel Hill

4.20

4.16

4.16

4.25

4.25

4.20

Rice

4.20

3.35

3.35

3.26

3.26

3.30

Vanderbilt

4.10

3.10

3.10

3.27

3.27

3.18

Wash. U.

4.10

2.79

2.79

2.84

3.41

2.96

U. Ill.-CU

4.00

3.94

3.94

3.93

3.93

3.93

Emory

4.00

2.75

3.00

3.14

3.77

3.17

Georgetown

4.00

1.23

2.94

1.23

2.94

2.08

USC

3.90

3.13

3.13

3.17

3.17

3.15

Notre Dame

3.90

2.48

2.98

2.83

3.09

2.85

UCSD

3.80

3.82

3.82

3.77

3.77

3.79

NYU

3.80

3.66

3.66

3.64

3.64

3.65

Will & Mary

3.80

.94

2.83

.94

2.83

1.88

Brandeis

3.60

2.30

3.07

2.60

3.12

2.77

Tufts

3.60

1.43

2.87

1.92

2.88

2.27

Case West’n

3.60

1.66

2.84

1.24

2.98

2.18

Boston Col

3.60

.97

1.66

1.47

1.96

1.51

Wake Forrest

3.50

.23

2.70

.23

2.70

1.46

Rochester

3.40

2.76

3.31

3.02

3.29

3.09

Lehigh

3.10

.21

2.50

.21

2.50

1.35

simple --simple program score
adjusted --average program score
balanced --balanced program proxy
AjdBal. --adjusted balanced program proxy
average –average of the above scores

 

Note that (as came to my attention in late September after the article corresponding to this appendix was done was done) a measure of the “depth and breadth of academic quality of programs ” at the “The Top 25 Research Universities in the U.S., 2002-2003” (plus a “runner up” and “other” schools) is available at http://www.philosophicalgourmet.com/topresearch.htm.

My average measure correlates 0.881 with the measure from “philosophicalgourmet” for the 25 schools of  Table 1. It correlates 0.865 for these schools plus the public universities of Table A1; 0.848 with the complete Table SA1, including private schools not among the top 25 of Table A1 and omitted from Table 1 (i.e., William and Mary, Brandeis, Tufts, Case Western Reserve, Wake Forest, Rochester and Lehigh). My average measure correlates 0.822 with the measure from “philosophicalgourmet”for all schools in Table A1 without missing data.
Scores from the “philosophicalgourmet” are missing for for nine schools of Table A1 (i.e., Dartmouth, USC, Notre Dame, William and Mary, Brandeis, Tufts, Case Western Reserve, and Lehigh).