Political Science Admission Algorithms (Gary King, Simon Jackman)

[TDBank]

This is how PhD admissions supposedly work at Harvard and Stanford.

Harvard: http://gking.harvard.edu/files/PS93.pdf

Stanford: http://jackman.stanford.edu/papers/pa04.pdf

--------------------------------------------------------------------------------------------------------------------------

Ranking of PhD political science programs based on placement: http://www.people.fas.harvard.edu/~chingos/rankings_paper.pdf

1. Harvard, 2. Stanford, 3. Michigan. 4. Rochester, 5. Chicago, 6. UC--Berkeley, 7. Duke, 8. WUSTL, 9. UCLA, 10. UCSD, 11. MIT, 12. Yale, 13. Princeton, 14. Cornell, 15. Columbia, 16. Northwestern, 17. Michigan State, 18. Ohio State, 19. Emory, 20. UNC--CH

[TDBank]

Gary King's article was written in 1993, Jackman's in 2004.

[Scalia]

I was wondering if someone could comment on the ranking methodology based on placement compared to that of USNEWS. Which one would you find more influential in choosing a program and why?

I've also noticed that some schools that are very strong overall like John Hopkins, Brown, and Penn don't do quite so hot in the USNEWS but seem to do okay in the rankings based on placement. Also, quite a few people who are applying to the very top schools have also cast some applications towards these schools despite their USNEWS rankings. Are these schools better than they are ranked or is it simply the "brand name" of their schools that attract so many apps?

As always, any information is greatly appreciated.

-Scalia

[TDBank]

I was wondering if someone could comment on the ranking methodology based on placement compared to that of USNEWS. Which one would you find more influential in choosing a program and why?

I've also noticed that some schools that are very strong overall like John Hopkins, Brown, and Penn don't do quite so hot in the USNEWS but seem to do okay in the rankings based on placement. Also, quite a few people who are applying to the very top schools have also cast some applications towards these schools despite their USNEWS rankings. Are these schools better than they are ranked or is it simply the "brand name" of their schools that attract so many apps?

As always, any information is greatly appreciated.

-Scalia

The US News ranking is based on reputation, and its input is feedback from chairs of political science departments from all over the country. The ranking cited in this thread is based on placement of PhD graduates, and the methodology is to assign an initial value to each program and placement into that program, which changes after each step of an iterative process, until the iteration converges. The authors do not talk about the how sensitive the final convergence is to the initially imputed values of the parameters, so that could be one methodological weakness of this ranking. But all in all, I think ranking based on placement is the one PhD students should care most about, and this ranking is the best placement-based one available at this point.

It could be just the brand name that attracts top applicants to schools like Penn, or it could be not. I can give an educated guess about this, but I really can't know for sure. The rankings certainly don't give us any decent answer to this puzzle.

[SuddenlyParanoid]

I think some of the reason that some schools with very good placement are not ranked as high in US News is that they are more focused programs. Rochester is one of the three main boutiques (along with CalTech and Stanford GSB). All Rochester students take formal and methods work, if I'm not mistaken, and tend to place very well. However, when faculty judge the reputation of a program they probably look at its strengths in all the subfields even if the school doesn't place much emphasis on them.

Similarly, I assume Chicage's cohorts has a lot more theorists than the average. Since Chicago is awesome at that, their students do well but the raters take into account Chicago's strength across everything without "weighting" the subfields by emphasis.

As for me, I looked at the Chingos study, schools' placement that they place online (although many are incomplete, and US News rank.

For outside of academia, say consulting, the government, or think tanks, does the brand name of the school matter? So a lower ranked Ivy would look better than higher ranked publics or less well known privates? I would guess that it does plus the name brand adds status at cocktail parties and high school reunions where people aren't up on poli sci.

[SBL]

I think some of the reason that some schools with very good placement are not ranked as high in US News is that they are more focused programs. Rochester is one of the three main boutiques (along with CalTech and Stanford GSB). All Rochester students take formal and methods work, if I'm not mistaken, and tend to place very well. However, when faculty judge the reputation of a program they probably look at its strengths in all the subfields even if the school doesn't place much emphasis on them.

Similarly, I assume Chicage's cohorts has a lot more theorists than the average. Since Chicago is awesome at that, their students do well but the raters take into account Chicago's strength across everything without "weighting" the subfields by emphasis.

As for me, I looked at the Chingos study, schools' placement that they place online (although many are incomplete, and US News rank.

For outside of academia, say consulting, the government, or think tanks, does the brand name of the school matter? So a lower ranked Ivy would look better than higher ranked publics or less well known privates? I would guess that it does plus the name brand adds status at cocktail parties and high school reunions where people aren't up on poli sci.

On your last point, I think a lot would depend on what schools are specifically being compared and in what industries. For example in federal government circles, I imagine Georgetown has a greater reputation than UPenn.

[TDBank]

For outside of academia, say consulting, the government, or think tanks, does the brand name of the school matter? So a lower ranked Ivy would look better than higher ranked publics or less well known privates? I would guess that it does plus the name brand adds status at cocktail parties and high school reunions where people aren't up on poli sci.

This is indeed the educated guess I was thinking about. There is no way to empirically confirm it though, as the placement data are (perhaps unsurprisingly) not readily available at low-ranked political science departments within otherwise prestigious universities.

As far as reputational rankings are concerned, I really don't trust US News, as they don't seem to know what they're doing and their methodology is pretty sketchy. The NRC ranking would be the "gold-standard" reputational ranking to look at, but the old one is too old, and the new one hasn't come out yet.

Hey SuddenlyParanoid, that is an impressive list of acceptances you've got so far. Looks like you're gonna get it all. Good luck.

[ladedodaday]

This is indeed the educated guess I was thinking about. There is no way to empirically confirm it though, as the placement data are (perhaps unsurprisingly) not readily available at low-ranked political science departments within otherwise prestigious universities.

As far as reputational rankings are concerned, I really don't trust US News, as they don't seem to know what they're doing and their methodology is pretty sketchy. The NRC ranking would be the "gold-standard" reputational ranking to look at, but the old one is too old, and the new one hasn't come out yet.

Hey SuddenlyParanoid, that is an impressive list of acceptances you've got so far. Looks like you're gonna get it all. Good luck.

I think the problem with the Chingos study is that it doesn't include placements at non-political science PhD granting institutions, which means their results are a bit skewed. So, if you want a job at a top-100 research institution, by all means, that's your ranking, but that might not be everyone's goal.

[TDBank]

I think the problem with the Chingos study is that it doesn't include placements at non-political science PhD granting institutions, which means their results are a bit skewed. So, if you want a job at a top-100 research institution, by all means, that's your ranking, but that might not be everyone's goal.

It is evident that their results are skewed toward PhD-granting institutions. As for why, it's simply because their ranking methodology cannot accommodate non-PhD-granting institutions (for mathematical reasons which I'm sure you're not interested in).

[ladedodaday]

It is evident that their results are skewed toward PhD-granting institutions. As for why, it's simply because their ranking methodology cannot accommodate non-PhD-granting institutions (for mathematical reasons which I'm sure you're not interested in).

No, I understand the math perfectly well (it's very straightforward). It's fairly obvious that a reciprocal ranking can only include other PhD granting institutions! The problem is, as they mention, 55% of PhD recipients get tenure track jobs, but only 15% get jobs at PhD granting institutions. So, the placement ranking only includes a little over 1/4 of tenure track positions. I understand why this is the case, but the biases of rankings need to be explicitly stated before people can use them to make decisions.

[TDBank]

No, I understand the math perfectly well (it's very straightforward). It's fairly obvious that a reciprocal ranking can only include other PhD granting institutions! The problem is, as they mention, 55% of PhD recipients get tenure track jobs, but only 15% get jobs at PhD granting institutions. So, the placement ranking only includes a little over 1/4 of tenure track positions. I understand why this is the case, but the biases of rankings need to be explicitly stated before people can use them to make decisions.

I agree with you about the need to be explicit. Just to be fair to the authors, however, they did explicitly state the biases of their ranking. Specifically with respect to the bias against non-PhD-granting institutions, they wrote: "Although our method does not count those placed into positions at prestigious liberal arts colleges, the number of positions at such schools is limited compared to those at doctoral institutions. While our method also excludes graduates in other sectors of academic employment (comprehensive universities, two-year colleges, and high schools), it seems unlikely that a great number of those able to find tenure-track employment in doctoral universities would choose such positions instead.