Wednesday, January 10, 2007

Academic analytics


Academic Analytics is for wonky deans who are paranoid about how well their departments compare with those at other universities [via]. Read this and this for more about their methodology for measuring productivity. While 'ranking' is not their main aim, their measure does produce a rank order. So, here are the 10 most productive departments in materials science and engineering:

  1. University of California - Santa Barbara
  2. Northwestern University
  3. University of North Carolina at Chapel Hill
  4. University of Massachusetts - Amherst
  5. Princeton University and The University of Texas at Austin
  6. University of Minnesota - Twin Cities
  7. Massachusetts Institute of Technology
  8. University of Illinois at Urbana-Champaign
  9. Drexel University

5 Comments:

  1. Tabula Rasa said...

    do the rankings make sense in your field? that link came up on a popular discussion board for my field today. the interesting thing was that numbers 4, 5, 8, 9, and 10 on their list wouldn't be considered in the top 30 on any other -- it was so ridiculous that the thread died without a whimper.

    reinforces my belief that rankings are meaningless.

  2. Abi said...

    TR: Two pointss: (1) Academic Analytics' measure seems to be meant to help Deans rather than prospective students or faculty. (2) Rankings are not the most important outcome; some quantitative measure of a department's 'productivity' is what this exercise appears to be about.

    Frankly, I was surprised to see MIT at No. 8. Illinois, MIT, Berkeley, Northwestern are the big guns in materials science. They are all large departments with excellent credentials, and I would expect them to occupy most of the top slots in any list. AA's measure of 'productivity' seems to weight certain factors (which ones? more paper/citation bang for each research buck? I don't know) which seems to promote Princeton and Minnesota, (which are good schools, but) whose materials program is small.

    In any event, I have not been up on the rankings game in materials science, but here's the NRC list (which I believe is from 1995):

    1. MIT
    2. Northwestern
    3. Cornell
    4. Berkeley
    5. Illinois
    6. Stanford
    7. UMass (Amherst)
    8. UCSB
    9. Penn State
    10. Penn

  3. Anonymous said...

    I'd like to add that UMass-Amherst does not actually have a "Materials Science and Engineering" department. It has Polymer Science and Engineering, Chemistry, Physics, Chem Engg, etc...and a Materials Research Science and Engineering center (MRSEC) which is housed in the PSE dept. I should know - I am a PSE grad student. PSE is probably the one that this ranking refers to. It is humbling and at the same time a little surprising that this dept is ranked so high, ahead of MIT, Cornell etc.

    Would be interesting to see what the criteria really is...

  4. Anonymous said...

    Just a quick reply to the first commenter; it seems that the rankings are based on quantitative data, which are kind of hard to pass off as ridiculuous, even if they show differences from common opinion (hard data may show that reputations are unfounded, no?). Here are the data they used, from their site (http://www.academicanalytics.com/variables2005.html):

    Percent faculty with a book publication
    The number of faculty in the Ph.D. program who have authored or co-authored a book, divided by the number of faculty in that program.

    Unique book publications per faculty member
    The total number of unique books attributed to the Ph.D. program (i.e., a book does not count twice if co-authored by two faculty members) divided by the total number of faculty in that program.

    Percent faculty with a journal publication
    The number of faculty in the Ph.D. program who have authored (or co-authored) a journal article, divided by the total number of faculty in that program.

    Unique journal publications per faculty member
    The total number of unique journal articles attributed to the Ph.D. program (i.e., a journal article does not count twice if co-authored by two faculty members) divided by the total number of faculty in that program.

    Percent faculty with a citation
    The number of faculty in the Ph.D. program whose journal articles have received at least one citation, divided by the total number of faculty in that program.

    Unique citations per faculty member
    The total number of unique citations attributed to the Ph.D. program (i.e., the total number of citations for each unique journal article; the citations of articles co-authored by members of the program are counted only once) divided by the total number of faculty in that program.

    Unique citations per journal publication
    The total number of unique citations attributed to the Ph.D. program (i.e., the total number of citations for each unique journal article; the citations of articles co-authored by members of the program are counted only once) divided by the total number of unique publications attributed to that program (i.e., a journal article does not count twice if co-authored by two faculty members).

    Percent faculty with a grant
    The number of faculty in the Ph.D. program who have won a federally-funded research grant, divided by the total number of faculty in that program.

    Unique grants per faculty
    The total number of unique federally funded research grants attributed to the Ph.D. program (i.e., a grant does not count twice if co-authored by two faculty members) divided by the total number of faculty in that program.

    Unique grant dollars per faculty
    The total number of unique federally funded research dollars attributed to the Ph.D. program (i.e., a grant’s research dollars do not count twice if co-authored by two faculty members) divided by the total number of faculty in that program.

    Unique dollars per unique grant
    The total number of unique research dollars attributed to the Ph.D. program (i.e., the total number of dollars for each unique research grant; the dollars associated with grants co-authored by members of the program are counted only once) divided by the total number of unique research grants attributed to that program (i.e., a research grant does not count twice if co-authored by two faculty members).

    Percent faculty with an award
    The number of faculty in the Ph.D. program who have won an honorific award, divided by the total number of faculty in that program.

    Total awards per faculty
    The total number of honorific awards attributed to the Ph.D. program, divided by the total number of faculty in that program.

  5. NONE said...

    UCSB and Northwestern are top MSE, but programs at Urbana and MIT are better.
    North Carolina, Princeton, Texas, Minnesota and Drexel shouldn't make top 10, in my opinion.

    Competition for top spot should be between Urbana and MIT, with Northwestern close second. Followed by UCSB, Berkeley, Stanford, UMass Amherst, CalTech, Penn State, Cornell etc... Drexel has a decent metallurgy program but no balance towards soft/nano materials, which are obviously crucial in today's science. It should be ranked 20-30th, at best.