Sunday, July 23, 2006

July update

Again, I'd like to thank everybody who sent comments and corrections. Fortunately, there were not many of them. However, some of them prompted me to update the journal publications information for everybody on the spreadsheet. Here is the updated spreadsheet (download button on the bottom of the page).

The main changes from the previous spreadsheet:

- The political science departments at Alabama and New Orleans are now added.
- I removed three persons who either were associate professors in February 2006, or had not yet started as assistant professors at that time.
- I removed the School and PhDSchool worksheets, they are just too problematic (outliers, selection bias etc).
- I had significantly misrepresented the publication record for Jacob T. Levy, and this is now corrected.
- I cleaned the subfield designations up just a little, but still, they are rather inconsistent and thus don't rely on them.

There are other minor corrections, too insignificant to point out.

I might update my previous posts with "rankings" later.

Saturday, July 08, 2006

Comments

Thank you for all the comments. Instead of publishing and answering them individually, I have decided to prepare a Q & A post, where I will address questions and comments as they come up. I will try to update the post as often as necessary. However, I will most probably post updates on weekends only (and certainly not every weekend).

Mistakes and updates

Please let me know of any possible mistakes. Entering a lot of data into a spreadsheet quickly (no, I did not spend months on this...) produces mistakes. I have received some comments about possible mistakes, but after verifying I have only found one clear mistake: Jeff Yates was an associate professor in early 2006 and he should not be included. I will update the spreadsheet in the future, when more mistakes are brought to my attention.

I do repeat a warning on subfield designations: don’t rely on them, they are inaccurate (see also below why)! I don’t have the time and competence to change them, though.

Q & A on possible mistakes (based on actual comments)

Dr. [A] at [University B] should be included

Associate professors, tenured on untenured, are not included here. The CV on the particular person’s webpage says that the person has been an associate professor for a couple of years

Dr. [A] at [University B] has tenure [and should not be included]

If I could, I would certainly include tenure information (there are tenured assistants and non-tenured associates), but I just don’t have information on this. The CV on the particular person’s webpage, updated even after the data collection was completed, indicates that the person is still an assistant professor.

The database includes people who have tenure

See previous question. Also, the database is current as of early 2006, and surely a lot of tenure decisions have been made since then. Similarly, people might have moved since early 2006 (and certainly, many will have moved by September).

I did not receive my PhD from where the database indicates, and I have more publications

This is a tough one. These kinds of mistakes are the most embarrassing, and may hurt the particular person. However, there is nothing I can do until I receive more specific information. As regards the publications, the errors are most likely when the person did not have an updated CV on internet in early 2006. Also, please bear in mind that Web of Science is not perfect (in fact, in one case I noticed that Web of Science did not list a person as an author when that person clearly co-authored the piece - the database corresponds to the real authorship). Also, publication information is based on information as of early 2006, thus all later publications are omitted. If I end up updating the data in early 2007 (or early summer of 2007 when it makes more sense), they will be included then.

The reliability and validity of the data?

This is a good comment. I should have given a general indication of the safeguards undertaken.

Validity should not be a huge problem for this dataset. There are no measures of particularly indefinite concepts. Reliability is a problem, of course. And my safeguards were probably not sufficient for clearing publication standards in, let’s say, top political science journals. I did take a second look at several randomly chosen assistant professors, but that was all. Confidence in the reliability of the data should increase over time, though (especially the more downloads there are, the more people look at it, and the more comments I get).

Publications data

Are forthcoming publications taken into account?

No. This information is simply not available.

People usually count forthcoming publications when calculating their average yearly output, thus Pub/YearwPhD is inaccurate

True to some extent. Will this problem hurt anyone? Hardly, as long as hiring committees look at least briefly at individual CVs.

Also, having individuals’ Pub/YearwPhD right does not seem very important. A more interesting question seems to be the average output of a median assistant professor, especially of a median assistant professor about to go up for tenure. Considering that these assistants are out of graduate school for several years, the forthcoming publications should not increase their average output too much (one "fortchoming" for a person out for five years increases the average by only 0.2). And of course, if we wanted to know the average output of someone out for even a longer time, the forthcoming articles will not matter much at all.

Top 10 rankings

I am not a big fan of individual rankings; I hesitated to put them up. I repeat: please don’t take them too seriously. I believed they would spark some discussion (which they did on some other blogs). And they do provide some interesting information. Also, such rankings are very easy to construct for any other data items (“data” -> “sort” etc…).

Publications (especially when ranking individuals) should not have equal weight

This is a very good comment, and I completely agree with it. The million dollar question is, of course, what the appropriate weights should be.

There are several options. The first is the Hix system (using journal impact factor, see his department rankings. I decided against using this approach (but anyone could try, it just requires extra coding!). First, relying on journal impact factor to rank individuals can be very misleading (I don't have a good link right now, but just enter "journal impact factor" in some search engine, e.g. Web of Science itself, and there are several articles that point out the problems). And I did not want to drop publications in other fields such as economics or psychology, but including them creates problems as impact factors may not be comparable between different fields.

Second option is just giving more weight to some very selected publications, especially the “top 3.” How much then? Triple for APSR and double for AJPS and JOP? Based on what? And, is an IR article in JOP really “worth” twice as much as an IR article in IO?

I just decided to have one raw “top” ranking; anyone could play around with the data and create their own top rankings. The easiest “custom” ranking is “data” -> “sort” -> “NoTop3/*”; I have done the calculations for this.

Break the rankings up by subfield

Again, an excellent comment. However, I don’t feel competent enough to judge everyone’s subfield.

I also want to give yet another very strong warning (repeated from above, and the notes on the data): if you want to get a list of scholars in a particular subfield, don’t rely on the subfield designations. They are unfortunately inaccurate and inconsistent. For example, americanists may be identified as “Behavior,” “Formal,” “Judicial,” “Law,” “Methods,” “PA,” “Policy,” “Positive Political Theory,” “Political Economy” or even something else. The departmental webpages and individual CVs do not give clear-cut designations. I did not want to leave subfield designation completely out, because it may be helpful when looking at a particular person. But again, don’t rely on them!!

Again, thank you for all the comments, and hopefully you find the data interesting.

Monday, July 03, 2006

Introduction

I have published here a database on assistant professors of political science in American PhD-granting departments. The link to data is in the next post. Comments are strictly moderated, and they will not appear promptly after you leave them (in fact, they may not appear at all). This is not a discussion forum, I enabled the comments so that I could get feedback if anyone wishes to provide me some. If you want to spread rumors and speculations about the job market, go to blogs such as americanandcomparativejobs.blogspot.com, irrumormill.blogspot.com or politicaltheoryrumormill.blogspot.com. For gossip on other things, go to irdepartments.blogspot.com, american-departments.blogspot.com, or comparativedepartments.blogspot.com.

I am not sure if and when I will update the data; I will almost certainly not do so before early 2007. When you notice any inaccuracies (there surely are some), or have any suggestions, leave a comment. I will gladly take care of them if I can.

Data

Here is the excel file (click the download button on the bottom of the page). Blogger unfortunately does not support file uploading/downloading. Variable names are explained below.

Basic information on the data

There were altogether 740 assistant professors in my sample. I am missing some (e.g. the whole department of University of New Orleans as their website was down when I compiled this). In some instances, I had to make judgment calls (when to include people who are affiliated with the political science department but not really there - I am not sure I was quite consistent in doing this; mostly these people were not included, but with some double appointments I just did not know). Only assistant professors are included, thus associates without tenure are "out." I hope I got all of the PhD-granting political science departments besides New Orleans. This excludes some excellent departments like Dartmouth; and it also excludes Public Affairs or International Affairs schools (except Maxwell School at Syracuse - I included their department of political science).

For the rest of the description, viewing the excel table alongside this post is probably helpful.

The work was done from late January until mid-February within approximately one month. The dataset indicates when the information was updated for each person. I don't think there are any major biases from the one-month time lag, the period is short enough. I made sure that no information on major journals was added to the ISI database during that time.

The information on the departmental affiliation, PhD schools, and graduation year is collected from various sources. Unfortunately, and surprisingly, it was really difficult to get this information for some people (there was a useful website/online cv for just over 500 people). When calculating the variable "years with PhD" (YearswPhD), I simply subtracted graduation date from 2006. For ABDs and those who got their PhD in 2006, "YearswPhD" field is arbitrarily "1."

Subfield designation is very arbitrary and inconsistent. I mostly relied on the departmental website, if such existed, and then my (mostly incorrect) assessment of what the person was doing. Please, do not rely on those designations. For example, you won't get a listing of people doing comparative politics by selecting fields where subfield designation is "comparative."

Publication data is gathered from ISI Social Sciences Citation index and ISI Arts & Humanities Citation Index (all searches were done in both of the two databases). Only those journal articles that were included in those two databases are counted. This creates serious problems, as anyone well knows. Some good publications are excluded, e.g. Presidential Studies Quarterly, State Politics and Policy Quarterly, and the early volumes of Political Analysis. Some non-peer-reviewed are included (some foreign policy journals come to my mind). Chapters in edited volumes are excluded.

Book reviews, editorial matters etc were not included in the database. Also, articles in the PS: Political Science and Politics were left out (this was a judgment call in the beginning to avoid counting "teaching" and "to the profession" type articles, and I decided to exclude all articles even though I discovered later that I have left out some substantive pieces).

Citation count is based on ISI Web of Science Cited Reference search. Everything was done manually person-by person, checking that the particular cited reference belonged to the person. There are certainly errors in this (there are just too many people called "Johnson" or "Kim" or "Lewis" to guarantee complete accuracy), but I believe I did the best I could. The citations are not discounted by the number of authors per reference. And of course, citations to everything (including books and edited volumes) are counted, if the cites appear in ISI articles. Sure, citation rankings are not too useful for very young assistant professors - there has been no time to gather the cites - but it does show something.

Book info is collected from persons' websites, when available. When not available, then I ran a search in the Library of Congress online catalogue.

Many people are obsessed with "top 3" publications (i.e. APSR, AJPS and JOP). I acknowledge that "top 3" is not particularly useful, but I still added the number of publications in "top 3" together for each individual. I did not add a Hix-style impact factor for each journal in order to get "Hix rankings for assistant professors."

The other basic calcuations I already did was discounting each publication by number of authors (and then summing these numbers together - "NoPub/Auth" variable); further dividing this by the number of years with PhD ("NoPub/Auth/YearswPhD"); and similar calculations with "top 3" articles.

For getting the school- and PhDSchool-level calculations, I summed up the YearswPhD variable for each school ("YearswPhD"); summed up the NoPub variable for each school ("NoPub"; note that this will create double entries for a school when two people from one school contribute to the same article); summed up the "discounted by number of authors" number of publications ("NoPub/Auth"); and did similar things to citations and "top 3" articles (everything should be self-explanatory).

Where do assistant professors come from?

The 740 assistant professors in the database come from 108 different PhD programs. The raw ranking is given below. Of course, this does not take into account the size of the program or where the graduates were placed.

1. Harvard (50)
2. UCBerkeley (46)
3. Michigan (36)
4. Princeton (33)
5. Columbia (32)
6. Stanford (29)
7. Yale (27)
8. Chicago (26)
9. UCLA (22)
10. Rochester (20)
11. Duke (19)
12. UC San Diego (18)
13. Cornell (16)
13. Ohio State (16)
15. Michigan State (15)
16. UNC (14)
17. Minnesota (13)
18. NYU (12)
19. MIT (11)
19. U. of Washington (11)
21. Washington U. (10)
22. Emory (9)
22. Indiana (9)
22. Texas A&M (9)
25. Arizona (8)
25. CalTech (8)
25. Georgetown (8)
25. Wisconsin (8)
29. Florida State (7)
29. Illinois (7)
29. Northwestern (7)
29. Rutgers (7)
29. UCDavis (7)
34.-38. George Washington, Iowa, SUNY Stony Brook, Texas-Austin, Virginia (6 each)
39.-44. Johns Hopkins, Maryland, Oxford, Pittsburgh, Rice, UC Irvine (5 each)
45.-51. Colorado, Georgia, Notre Dame, Penn State, Pennsylvania, SUNY Binghamton, Toronto (4 each)
52.-57. Claremont, Florida, Houston, Stanford (Business), UC Santa Barbara, Vanderbilt (3 each)
58.-70. Arizona State, Boston College, Cambridge, Missouri, MIT (Economics), New Mexico, New Orleans, New School for Social Research, North Texas, SUNY Buffalo, Syracuse, Syracuse (PA), Wisconsin-Milwaukee (2 each)
71.-108. Akron, American, Brandeis, Brown, Cambridge (History), Colorado-Denver (GSPA), Connecticut, EUI, Fordham, Georgia State, Hawaii, Kansas, Kentucky, Lund, Massachusetts, Melbourne, Michigan (Communication), Michigan (Philosophy), Missouri-St. Louis, Nebraska, North Carolina State, Ohio State (Psychology), Oregon, Pittsburgh (GSPIA), Purdue, Rutgers (PA), South Carolina, Stanford (Communication), Stanford (Education), Texas-Dallas, Toulouse (Economics), UC Riverside, UCBerkeley (Rhetoric), UCDavis (Ecology), UCLA (Economics), UCLA (Policy), UCLA (Psychology), Virginia Commonwealth (1 each)

Who has published the most articles?

The following only covers journal articles appearing in the Web of Science database, book publications and chapters in edited volumes are not taken into account (what can one do?).

The sum of “number of publications” variable provides a total of 1907 articles, which is 2.58 items per assistant professor. When using this method of calculation, any article for which more than one assistant professor has contributed is counted twice (or three times etc, depending on the number of assistants involved). When discounting the number of articles by co-authorship (e.g. when there are two authors, each get 0.5 “points”), then the total is 1343.7. This indicates that single authorship is still quite common. In fact, there are 916 single-authored publications in the database. I cannot tell exactly how many co-authored publications there are because of the double counting. The main conclusion – that the single-authored article publication is “alive” – does not change.

There are 165 assistant professors who don’t have a single entry in the Web of Science database, and 195 who have more than 3. The “median” assistant professor has two entries and 1.33 “co-authorship discounted” publications. The “median” assistant professor also has 0.45 publications per year with PhD (notwithstanding co-authorship), and 0.33 “co-authorship discounted” publications per year with PhD.

The “rankings” or “top lists” have huge problems, so please don’t take them seriously. The “top 10” of each category is given below (for “years with PhD” lists, the assistant professor must have a value of “years with PhD” of at least three; you can construct the lists for one- or two-years from the data yourself very easily). The employer (as of January-February 2006) is given first, then the PhD school and year in parentheses.

Most publications

1T. Jeffery A. Jenkins (18), Northwestern (Illinois 1999)
1T. Jeffrey A. Karp (18), Texas Tech (UC Santa Barbara 1995)
3. David Tewksbury (14), Illinois (Michigan (Communication) 1996)
4T. Kristian S. Gleditsch (13) UC San Diego (Colorado 1999)
4T. Susan A. Banducci (13), Texas Tech (UC Santa Barbara 1995)
6T. Sean Nicholson-Crotty (12), Missouri (Texas A&M 2003)
6T. Lucan Way (12), Temple (UCBerkeley 2001)
6T. Peter Andreas (12), Brown (Cornell 1999)
6T. J. Tobin Grant (12), Southern Illinois (Ohio State 2001)
10T. Scott W. Desposato (11), UCSan Diego (UCLA 2001)
10T. Daniel Drezner (11), Chicago (Stanford 1996)
10T. John Bohte (11), Wisconsin-Milwaukee (Texas A&M 1997)

Most publications discounted by co-authorship
When two authors, a person gets 0.5 “points”, when three authors, a person gets 0.33 “points” etc.

1. Jeffery A. Jenkins (12.67), Northwestern (Illinois 1999)
2. Daniel Drezner (11), Chicago (Stanford 1996)
3. Peter Andreas (10.75), Brown (Cornell 1999)
4. Lucan Way (10), Temple (UCBerkeley 2001)
5. Donna Lee Van Cott (9), Tulane (Georgetown 1998)
6. Paul Goren (8.5), Arizona State (Pittsburgh 1998)
7. Jeffrey A. Karp (8.42), Texas Tech (UC Santa Barbara 1995)
8. Venelin Ganev (8), Miami (Ohio) (Chicago 2000)
9. David Tewksbury (7.87), Illinois (Michigan (Communication) 1996)
10. Scott W. Desposato (7.83), UC San Diego (UCLA 2001)

Most publications per year with PhD
Only those assistant professors who received their PhD in 2003 or before are included

1. Sean Nicholson-Crotty (4), Missouri (Texas A&M 2003)
2. Jamie Carson (3.33), Georgia (Michigan State 2003)
3. Jennifer Lawless (2.67), Brown (Stanford 2003)
4. Jeffery A. Jenkins (2.57), Northwestern (Illinois 1999)
5. Adam H. Meirowitz (2.5), Princeton (Stanford GSB 2002)
6T. Lucan Way (2.4), Temple (UCBerkeley 2001)
6T. J. Tobin Grant (2.4), Southern Illinois (Ohio State 2001)
8T. Ronald R. Krebs (2.33), Minnesota (Columbia 2003)
8T. Ethan Bueno de Mesquita (2.33), WashU (Harvard 2003)
8T. Kevin Arceneaux (2.33), Temple (Rice 2003)

Most publications per year with PhD discounted by co-authorship
Only those assistant professors who received their PhD in 2003 or before are included. When two authors, a person gets 0.5 “points”, when three authors, a person gets 0.33 “points” etc.

1. Sean Nicholson-Crotty (2.5), Missouri (Texas A&M 2003)
2T. Lucan Way (2), Temple (UCBerkeley 2001)
2T. Ronald R. Krebs (2), Minnesota (Columbia 2003)
4. Ethan Bueno de Mesquita (1.83), WashU (Harvard 2003)
5. Jeffery A. Jenkins (1.81), Northwestern (Illinois 1999)
6. Adam H. Meirowitz (1.71), Princeton (Stanford GSB 2002)
7T. Andrew L. Roberts (1.67), Northwestern (Princeton 2003)
7T. Jamie Carson (1.67), Georgia (Michigan State 2003)
9. Branislav Slantchev (1.58), UC San Diego (Rochester 2002)10. Scott W. Desposato (1.57), UC San Diego (UCLA 2001)