The most important thing to ask for during campus visit

[notcoachrjc]

In the next few weeks and months, those of you who have been accepted will begin to visit departments. There is another thread about what to ask for during these visits, the advice there mostly suggests asking rather trivial questions about aspects of graduate school that will only marginally impact your time there (course offerings, travel funding, even the stipend itself). 

There should be one thing in particular that you ask for over anything else: for a complete record of graduate student progress from enrollment to graduation and placement. Ultimately, the end goal of graduate school is an academic job (as I highlighted earlier, this is a fairly rare outcome). You want to put yourself in the best possible position to succeed and the only way to do so is with hard data, not a brief glimpse at a department on its best possible behavior. The former approach is fine, but rife with selection bias. Departments will highlight their best placements, their happiest students. Faculty will be the most active in both engaging with graduate students and in recruitment. Ultimately, recruitment is a way to one-up someone at a rival department or get a competent research assistant. You should discount these interactions as much as possible. 

A complete record of graduate progress gives you several key pieces of information:

- How many people in your entering cohort are actually likely to get tenure-track jobs. This percentage is a good benchmark.

- How does your subfield do compared to others? What about your adviser? If IR's rate of placement from enrollment is 50% and American's is 5%, the overall record may look alright, but a large proportion of students in the department may lack opportunity to get jobs. This is usually true for Theory (some highly-ranked departments haven't placed a theorist in over a decade). If your prospective adviser has never placed anyone into a TT job (especially if he/she's had multiple opportunities), this is obviously problematic.

- What is the attrition rate? If it is high, it would be good to investigate why people drop out.

- Where do graduates get academic jobs? A job is a job, but if a department is mostly placing students in visiting positions or post-docs with few tenure track jobs (and those are in community colleges, or whatnot), it may speak to underlying problems. 

If a department can't or won't share this information, you have to immediately be suspicious. Are they hiding behind a high attrition rate? A poor placement rate among PhD recipients? Are students prevented from getting degrees because they don't get academic jobs (you'd be surprised how common this is) to keep the PhD holder-to-academic job rate high? 

Again, asking other questions at these visits is fine, but for those looking at an academic career (and that should be all of you), this is the only hard data you can get and the only objective measurement of quality outside of the very noisy US News rankings. Do yourself a favor and ask. 

[wb3060]

1 hour ago, notcoachrjc said:

Are students prevented from getting degrees because they don't get academic jobs (you'd be surprised how common this is) to keep the PhD holder-to-academic job rate high? 

Can you elaborate on this? 

[elw]

OP -

Honest question: From what position of authority do you speak from?

[joseon4th]

6 minutes ago, elwright said:

OP -

Honest question: From what position of authority do you speak from?

Judging by OP's posting style, I believe notcoazhrjc would be the first to encourage us to question the origin of their commentary.

Edited February 12, 2016 by joseon4th

[brent09]

As a current grad student (and successful TT placement, for whatever that's worth), I would largely agree with OP. I don't think departments are nefariously deceptive, but they will want to present the best set of statistics they can. (I've never heard of a department failing a defense because the student did not land an academic position. That seems a bit apocryphal.)

The OP is right, though, that a placement list can be deceptive. You'll want to know (a) how your subfield places, since even at top departments there can be wide variation across subfields; and (b) how your would-be advisors place students. You can also ask about completion rates, but from my limited experience, grad students who drop out do so for many reasons. At my department, I cannot think of a single attrition that stemmed from a systematic, departmental failure. This is not universal, though, so if attrition is very high it should give you pause.

But also remember this: there's wide variation in placement across students. One or two strike-outs can dramatically lower a department's placement rate in a year. And people strike-out at Princeton and get R1 placements from Emory. You should weight a department's placement heavily---perhaps primarily---but it's not the only predictor of your career success.

[rising_star]

Not in your field but, what @brent09 said. Placement record matters and, if possible, you want to get a sense of what things led to attrition. As brent09 has said, some of that is personal and due to the students rather than the program. But, if your POI has a track record of their students quitting or taking a long time to finish or not placing well, that's something you'll want to know going in. And note that by "placing well", I mean placing their students at the kinds of places you (think you) want to work at. If you want to be at a top-25 R1 after graduation, then working with someone whose best students place at R2s (and not just because that's where those students want to be) is something you may want to reconsider. Placement, like attrition, is tricky because a lot of it is personal and not up to the advisor. Still, it's information you should seek out.

[elw]

I take issue with the tone of the original poster, not the importance of past placement.

[brent09]

47 minutes ago, elwright said:

I take issue with the tone of the original poster, not the importance of past placement.

Yea, I tend to agree.

And to clarify: I disagree with OP about the other details. Travel funding, summer funding, course offerings... these are not 'trivial.' This is a big decision, it contains multitudes.

[TakeruK]

I think this is a very important question, but not "the most important". But I don't even know what "the most important" would be....I don't like absolutes/superlatives because it depends so much on each situation!

However, I want to say that I came into grad school applications thinking something similar, that I would ask every school for a complete record of their student placement. But none of the schools kept these records. Maybe it's just a difference in field, but I did originally think the first school was being shady by not being able to answer this question until I talked to more and more and none of them knew. Now that I've been in my school a few more years, I understand it a bit better now. It's really hard to keep this information (and to maintain student privacy and confidentiality).

It is especially hard because you don't want to just know where they went immediately after graduation (most students would have the next thing lined up so this can be a simple thing to track at graduation). But you want to know where they are 4-6 years past graduation. But people lose touch over this time span.

Also, the data is sensitive. I don't think a school should tell other people where their students went unless their students gave them permission to do so. At least not on like a formal report of student post-graduation plans (of course, an advisor might know where their own students are and might tell you informally). So, the school needs a way to keep in touch with its graduates and survey them and ask them for permission etc.

There's the problem of small numbers. My program graduates 3-4 students per year. I don't think this is enough to really look for trends, since the ability of students to find work would also depend on how the economy is doing, how much the government is funding basic science etc. And, what specific interests did each student have! Maybe 50% of the graduates here want to go into academia, and when there's 4 graduates per year, that's 2 per year. If you see a few years with 3 into academia (or 1 into academia), you don't know if the change is due to program's strengths or interests changing or just random.

One thing that is more useful at my school is the Exit Survey that all graduates (across campus) complete. They ask direct questions like do you have an offer, how much are you being paid, and where. Working with the school, I've seen the data and I know which places hire most of our graduates. But, this is confidential information--we asked the school to publish it to help prospective students decide but they are afraid of compromising the graduates' anonymity. Some companies might not like it that we say they hire our graduates. The data was collected on the promise of anonymity for internal institute use. We're hoping that some more aggregated data can still be released (for example, average post-graduation salary by department or something).

[alphazeta]

I think we can all agree placement is important (but this hardly means that placement data is the most important thing to ask for).  Some cautions though.

1) Attrition data is not very useful.  Many people decide they don't actually want a PhD or an academic job and quit.  That's not indicative of any departmental failings or problems.  Further, the "outside options" for students are generally best at the best departments, which means that non-academic careers are much more attractive to these people.  Any inference drawn from attrition is terribly confounded.  

2) People talk about departments/advisors "placing" their students.  This isn't how it happens.  You get a job based on what you do (and a healthy margin of luck).  Your outcome is a function of your effort, the resources available to you, and how you use those resources.   Different people need different resources and will tend to flourish in different environments.  Placement data is useful because it tells you how similarly-positioned people have fared, but a narrow focus on placement should never supplant consideration of the actual resources.  Everyone seems to understand this at the subfield level, but it's even more specific than that.  Obviously, it's a red flag if placements are very poor from a department where your evaluation is that the resources available are very good (the reverse is also a red flag), but you always need to consider both.

3) Students fail when they ignore the things they know about themselves.  A friend of mine was accepted to a number of departments; he attended the same one as me because it had the highest ranking/best placement.  Even then, he knew that his interests weren't a good fit for the department.  He also hated the location.  He was in our department for four years, was very unhappy, and very unproductive.  He then transferred to another department with lower rank/worse average placement (where in essence he had to start from scratch).  He's happy now; he's also productive.  Ultimately, he will place much better out of department #2.  He could, should, and almost did recognize this initially, but a narrow focus on placement led him to waste four unhappy years of his life.*

*Footnote: You have to do this evaluation carefully.  As a general rule, attending the higher-ranked department is a good choice (especially if the difference in rank is large).  I've talked to a lot of people who overweight fit in their decision (and think they can evaluate fit more accurately than is actually the case).  It's a balancing act.  The whole point of my comment here is that you need to strike a balance.  

[brent09]

1 hour ago, alphazeta said:

I2) People talk about departments/advisors "placing" their students.  This isn't how it happens.  You get a job based on what you do (and a healthy margin of luck).  Your outcome is a function of your effort, the resources available to you, and how you use those resources.   Different people need different resources and will tend to flourish in different environments.  Placement data is useful because it tells you how similarly-positioned people have fared, but a narrow focus on placement should never supplant consideration of the actual resources.  Everyone seems to understand this at the subfield level, but it's even more specific than that.  Obviously, it's a red flag if placements are very poor from a department where your evaluation is that the resources available are very good (the reverse is also a red flag), but you always need to consider both.

 

I agree generally, but take some issue with this part. You're right that placement is mostly up to the student. But it's undeniable that some advisors do a better job of placing than others. It's not how 'famous' the advisor is, either. Some advisors are highly engaged---they make time to talk, they provide good and prompt feedback on research and market materials, they write good letters, put you in touch with the right people and lobby on your behalf.

Advisors matter a lot on the market. A distant, hands-off, checked-out or otherwise unresponsive advisor can hurt your development as a scholar, and your chances when applying for positions. They can't make up for personal failings, but they matter.

[alphazeta]

I may have put my point too strongly.  Yes, advisors matter a great deal.  They are one of the most important resources available to you, but your fate is still largely in your own hands.  More importantly, while I agree that some advisors are better than others, there's also a dyadic component.  That is, the person who would be my ideal advisor probably wouldn't be your ideal advisor.  

[whatwasithinking]

@alphazeta My research interest match to a professor who is fairly new out of columbia. While she is engaged in a few projects, she has yet to publish a lot of substantial papers. I know there is a book in progress. How do you guys feel about this? I am sure she's not had a single student graduate because it's just been 3 years since she joined the department. When I talked to her, she seemed excited and so am I but now I am confused.

[notcoachrjc]

On 2/12/2016 at 9:19 AM, alphazeta said:

I think we can all agree placement is important (but this hardly means that placement data is the most important thing to ask for).  Some cautions though.

1) Attrition data is not very useful.  Many people decide they don't actually want a PhD or an academic job and quit.  That's not indicative of any departmental failings or problems.  Further, the "outside options" for students are generally best at the best departments, which means that non-academic careers are much more attractive to these people.  Any inference drawn from attrition is terribly confounded.  

2) People talk about departments/advisors "placing" their students.  This isn't how it happens.  You get a job based on what you do (and a healthy margin of luck).  Your outcome is a function of your effort, the resources available to you, and how you use those resources.   Different people need different resources and will tend to flourish in different environments.  Placement data is useful because it tells you how similarly-positioned people have fared, but a narrow focus on placement should never supplant consideration of the actual resources.  Everyone seems to understand this at the subfield level, but it's even more specific than that.  Obviously, it's a red flag if placements are very poor from a department where your evaluation is that the resources available are very good (the reverse is also a red flag), but you always need to consider both.

 

These are two rather strong takes, but are ultimately not accurate. Regarding 1), I really don't see how you can infer that people deciding they don't want a PhD or academia is independent of the department's actions. Foremost, the only experience a graduate student really has with academia is through his/her department. It is without doubt that a toxic department culture, evidence of poor performance in one's subfield, being ignored by one's adviser, poor training, overburdening through TA/RA work or any number of other department-related reasons can lead to attrition or a decision that academia is not "right for me." Someone is far more likely to think academia is not right for them when their department is a mess and treats them poorly. Second, attrition rates are generally higher at lower ranked departments where funding is more scarce, so the "grass is greener" for higher ranked departments argument falls flat. It's simply not true.

Regarding 2), again, there's fundamental misunderstanding of how the tenure-track (and to an extent, post-doctoral market) works here. Of course there are things you can do that are independent of your adviser that help you get a job: publish, get grants (that don't require your adviser to be a PI) and network independently (which is difficult for most grad students). Otherwise, your success, for better or for worse is tied to the person chairing your dissertation committee. Throughout your graduate career, a good adviser will put you in position to succeed by not only giving feedback, but by providing access to invitation-only conferences, introducing you to big-name scholars in the field, participating in panels you organize for APSA, etc,. These intangibles are not trivial because they provides advantages and you can be that your competition on the job market is getting these. Moreover, once it comes time to apply for jobs, it's night-and-day between good and bad advisers. Good advisers will make calls on your behalf, attend your practice job talks and write not only good letters, but impactful ones for the positions you're applying for. Bad advisers will write a letter and then wonder, months later, why you haven't gotten any calls.

Remember, every tenure track job you apply for gets at least 100 applications. Many of the applicants will have publications AND pedigree. You need to stand out somehow. One way is for someone on the search committee to have already heard of you or to have heard of (or from) your adviser and vouch for you. This is why it's important to compare placement, not only across, but within departments. Are some faculty member's students finishing and getting jobs and others' aren't? Is it really plausible to say this disparity is because their students are lazy or unqualified or lined up six-figure consulting gigs or whatever? 

3 hours ago, whatwasithinking said:

@alphazeta My research interest match to a professor who is fairly new out of columbia. While she is engaged in a few projects, she has yet to publish a lot of substantial papers. I know there is a book in progress. How do you guys feel about this? I am sure she's not had a single student graduate because it's just been 3 years since she joined the department. When I talked to her, she seemed excited and so am I but now I am confused.

This is risky. There are three reasons not to have junior faculty as your advisers/chairs. One, they might publish out and leave in a couple of years and then you're out of luck. Two, they might not publish at all (sounds like the case here), and not get tenure. Again, out of luck. Three, even if everything goes well on the first two, you want your advocate when you're on the market to have the most impactful voice possible. This is generally not a junior scholar or even an early associate. It's good to have someone like this on your committee (or even as a co-chair), because they'll be more in touch with current research and read your work, but it's also a good idea to have a more senior person as chair for all of the reasons listed above. 

[alphazeta]

I don't want to start a flamewar here, but. 

On #1: Of course the decision to leave academia can be related to department culture, etc.  On the other hand, it often isn't.  In my own cohort, we experienced our first attrition on day 3 of classes.  If it were true that propensity to leave academia were distributed evenly across departments, then it would be valid to compare attrition across departments in an effort to measure toxic culture, etc.  This is not the case, though, where the most important out of a large number of confounds is the fact that some departments admit students with much better outside options.  Some departments also do a better job of guiding the (substantial number of) students who would actually be best off outside academia onto another path.  The worst outcome for a student is to spend 5-7 years, finish the degree, fail to place academically, and end up in either serial VAPs or a non-academic job that they could have gone into years ago.  

On #2: I never said your advisor doesn't matter.  Of course your advisor matters.  Anyone at the dissertation stage spends 95+% of their time working independently (aside from whatever TA/RA responsibilities they might have).  Your advisor doesn't write your dissertation, you do.  What matters is the quality of this product and any additional publications.  Networking can give you a leg up, but:

1) The advantages of networking, having your advisors make calls, etc. function as a multiplier.  A well-networked advisor who makes calls on your behalf can convince people to read your file and consider it; that is, they can get you on the "long shortlist."*  The rest is up to you - if people look at your file and it's crap, no advisor can save you.

2) Publications, publications, publications.  There are a handful of advisors who do  genuinely "place" unpublished students.  That is, they have enough of a track record and are prominent enough that search committees will believe them when they say someone has star potential (but the committee has to agree with this take when actually reading the material).  But publications are the coin of the realm.  An ABD with a solo top publication is far better off than an unpublished Gary King student.

3) There are a handful of truly selfless advisors in this profession.  That is, advisors who work hard on behalf of every one of their students.  This isn't the norm.  Advising a student is an investment.  If you're a good investment (i.e., someone who's work is good), your advisor will invest a lot more in you.  That is, there are some advisors that are just lazy/bad advisors but a dysfunctional advising relationship generally says as much about the student as the advisor.  

*Footnote: For those unfamiliar with the process, academic hiring proceeds roughly as follows.  An ad is posted, hundreds of applicants send in files.  Your file consists of a cover letter, CV, 3-4 letters of recommendation (of which the most important is the one from your advisor), and 1-3 writing samples (published/publishable papers or dissertation chapters).  Most committees make a very quick first pass over the files (i.e., in the range of 60 seconds per application).  This is generally just a look at the CV - at many programs, applicants who don't have either publications or pedigree (which is a function of department + committee) are tossed.  From here, a slightly more intensive winnowing begins, during which you might expect search committee members to read the letters and skim the writing samples.  Eventually, this gets you to a "long shortlist" of about a dozen top candidates.  At your average department, every member of the committee will review all of the files on the long shortlist in detail (including reading some or all of the writing samples).  From here, they narrow down to a list of candidates who are brought to fly out (usually 3-5 candidates).  The flyout consists of a series of one-on-one meetings and interviews and a job talk (1.5 hour presentation of your best paper) after which a final selection is made.

[alphazeta]

On 2/14/2016 at 7:08 AM, whatwasithinking said:

@alphazeta My research interest match to a professor who is fairly new out of columbia. While she is engaged in a few projects, she has yet to publish a lot of substantial papers. I know there is a book in progress. How do you guys feel about this? I am sure she's not had a single student graduate because it's just been 3 years since she joined the department. When I talked to her, she seemed excited and so am I but now I am confused.

The general rule of thumb: If you like a senior person's work, try to work with them.  If you like a junior person's work, try to work with their (former) advisors.

It's always better to work with more senior/established people.  At many universities, you actually must have a tenured chair (you can skirt this if you use the junior person as a de facto chair while a senior person is the de jure chair, but that can have problems).  As notcoachrjc points out, a junior person will usually not have the same networks and reputation, which is possibly detrimental.  Further, a letter from a senior scholar saying "X is one of the best students I have ever worked with" is a much stronger signal than a letter from a junior scholar saying the same thing.

At this stage in the process, I assume you're choosing between concrete alternatives.  I would recommend against going somewhere to work with a junior scholar if that's the only draw for the department in question.  I have observed that people often dramatically overestimate how important it is to "match" with an advisor's interests.  This doesn't need to be a very precise match.  

Bob Keohane, for example, who is essentially a scholar of international organization/global governance (perhaps best known as the father of so-called neoliberalism) has advised students in much the same mold (e.g., Moravcsik), but some of his notable advisees include J. Ann Tickner (noted feminist IR scholar), Gwyneth McClendon (comparative politics scholar, specializing in Africa and ethnic politics), Eddy Malesky (comparativist specializing in Vietnam, although also known for some work in IPE), Terry Karl (comparativist specializing in petro-politics, especially in Latin America), etc. He's advised students who work on both security issues (John Owen, Page Fortna) and IPE (Lisa Martin, Randy Stone) and much much more.  He's a brilliant scholar, but by no means an expert in those areas.

Obviously, the point of an advisor is to have someone with relevant expertise, but that doesn't need to be defined narrowly.

[notcoachrjc]

12 hours ago, alphazeta said:

I don't want to start a flamewar here, but. 

On #1: Of course the decision to leave academia can be related to department culture, etc.  On the other hand, it often isn't.  In my own cohort, we experienced our first attrition on day 3 of classes.  If it were true that propensity to leave academia were distributed evenly across departments, then it would be valid to compare attrition across departments in an effort to measure toxic culture, etc.  This is not the case, though, where the most important out of a large number of confounds is the fact that some departments admit students with much better outside options.  Some departments also do a better job of guiding the (substantial number of) students who would actually be best off outside academia onto another path.  The worst outcome for a student is to spend 5-7 years, finish the degree, fail to place academically, and end up in either serial VAPs or a non-academic job that they could have gone into years ago.  

On #2: I never said your advisor doesn't matter.  Of course your advisor matters.  Anyone at the dissertation stage spends 95+% of their time working independently (aside from whatever TA/RA responsibilities they might have).  Your advisor doesn't write your dissertation, you do.  What matters is the quality of this product and any additional publications.  Networking can give you a leg up, but:

1) The advantages of networking, having your advisors make calls, etc. function as a multiplier.  A well-networked advisor who makes calls on your behalf can convince people to read your file and consider it; that is, they can get you on the "long shortlist."*  The rest is up to you - if people look at your file and it's crap, no advisor can save you.

2) Publications, publications, publications.  There are a handful of advisors who do  genuinely "place" unpublished students.  That is, they have enough of a track record and are prominent enough that search committees will believe them when they say someone has star potential (but the committee has to agree with this take when actually reading the material).  But publications are the coin of the realm.  An ABD with a solo top publication is far better off than an unpublished Gary King student.

3) There are a handful of truly selfless advisors in this profession.  That is, advisors who work hard on behalf of every one of their students.  This isn't the norm.  Advising a student is an investment.  If you're a good investment (i.e., someone who's work is good), your advisor will invest a lot more in you.  That is, there are some advisors that are just lazy/bad advisors but a dysfunctional advising relationship generally says as much about the student as the advisor.  

*Footnote: For those unfamiliar with the process, academic hiring proceeds roughly as follows.  An ad is posted, hundreds of applicants send in files.  Your file consists of a cover letter, CV, 3-4 letters of recommendation (of which the most important is the one from your advisor), and 1-3 writing samples (published/publishable papers or dissertation chapters).  Most committees make a very quick first pass over the files (i.e., in the range of 60 seconds per application).  This is generally just a look at the CV - at many programs, applicants who don't have either publications or pedigree (which is a function of department + committee) are tossed.  From here, a slightly more intensive winnowing begins, during which you might expect search committee members to read the letters and skim the writing samples.  Eventually, this gets you to a "long shortlist" of about a dozen top candidates.  At your average department, every member of the committee will review all of the files on the long shortlist in detail (including reading some or all of the writing samples).  From here, they narrow down to a list of candidates who are brought to fly out (usually 3-5 candidates).  The flyout consists of a series of one-on-one meetings and interviews and a job talk (1.5 hour presentation of your best paper) after which a final selection is made.

Not a flamewar, just a disagreement. But there are some issues with the premise of your argument. They're commonly made, so it's an important debate to have. When you say:

" If it were true that propensity to leave academia were distributed evenly across departments, then it would be valid to compare attrition across departments in an effort to measure toxic culture, etc. "

I would think it would be just the opposite. If we see a pattern of attrition that is unevenly distributed toward a certain group of schools we can infer an association. You know, stuff we do in our research. Your claim is just that: we should see more attrition at top schools because those entering have better outside options...but we don't. NYU's attrition is in the 30% range (http://politics.as.nyu.edu/docs/IO/4610/nyu_placement_record2013.pdf), for example. Given their methods training, students should have plenty of outside options. Stanford's comparable page is down, but from what I remember, they placed about 50% of their admits into tenure-track jobs. Attrition, with the limited data available, seems to skew toward lower-ranked schools for people with fewer outside options (and those are available online as well: https://sites.google.com/site/honestgraduatenumbers/. Whether departments somehow do a better job of weeding out unqualified students or students leave on their own volition out of frustration with academic induced by the department is observationally equivalent. I just think it's unreasonable to discount the latter in favor of the former. 

Regarding the second point: Of course competence is a prerequisite, but there are certain advantages afforded to those with famous or well-known advisers by osmosis that other students simply do not get. Part of it may be self-selection of good students to famous advisers, but the good students get to show off much better: opportunities to meet other big names on the field, co-authorships with scholars at other institutions, invited publications that help their research get exposure. Solo publications at top places for ABDs are still rare enough that you can make a splash with a few co-authored ones (not with your adviser, of course -- these are discounted heavily). Advisers don't have to be selfless for this, they just have to be willing to have their students tag along to things, which these usually are. Contrast this to an adviser that barely communicates with their student or even one that's perfectly willing to provide feedback, but doesn't have any pull and there is a real distinction, if not in placement itself, but possibly the quality of the placement. It's why top-10 stars end up at top-10s and top-30 stars end up at other top-30s and all the way down from there. 

 

[alphazeta]

8 hours ago, notcoachrjc said:

Not a flamewar, just a disagreement. But there are some issues with the premise of your argument. They're commonly made, so it's an important debate to have. When you say:

" If it were true that propensity to leave academia were distributed evenly across departments, then it would be valid to compare attrition across departments in an effort to measure toxic culture, etc. "

I would think it would be just the opposite. If we see a pattern of attrition that is unevenly distributed toward a certain group of schools we can infer an association. You know, stuff we do in our research. Your claim is just that: we should see more attrition at top schools because those entering have better outside options...but we don't. NYU's attrition is in the 30% range (http://politics.as.nyu.edu/docs/IO/4610/nyu_placement_record2013.pdf), for example. Given their methods training, students should have plenty of outside options. Stanford's comparable page is down, but from what I remember, they placed about 50% of their admits into tenure-track jobs. Attrition, with the limited data available, seems to skew toward lower-ranked schools for people with fewer outside options (and those are available online as well: https://sites.google.com/site/honestgraduatenumbers/. Whether departments somehow do a better job of weeding out unqualified students or students leave on their own volition out of frustration with academic induced by the department is observationally equivalent. I just think it's unreasonable to discount the latter in favor of the former. 

I'm not sure we're even disagreeing on point 2, so much as emphasizing things differently.

On #1, I don't think I implied that attrition should be highest at top departments.  Rather, I meant that comparing attrition from a top department to attrition from a lower department will tend to make the top department look worse in comparison than it should.  That is, in the counterfactual where Stanford and Low-Ranked U admit the same students, I would imagine a much higher attrition rate at Low-Ranked U relative to what we observe at present.

But more fundamentally, there are students for whom attrition is a good outcome.  That is, these students are better off personally leaving academically than staying in it for any number of reasons.  A nudge out the door does these students a favor.  To repeat a point above, the absolute worst thing a department can do is string these students along until they finish and then condemn them to years of adjuncting/VAPing.  Some departments (mostly in the low-ranked U category) do everything they can to keep students, who the department knows will not succeed, in the program in order to report lower attrition to the administration and keep a pool of TAs available. 

 

[PhD4metoo]

Statistics are important, and some grad schools and programs do publish them - like Duke grad school, which uses the data to determine program sizes, and Yale, which puts its statistical profiles for each program online- for admissions, completion and career outcomes- http://gsas.yale.edu/academics/departments - for example, here's the data for Political science phd program: http://gsas.yale.edu/sites/default/files/department-files/political_science.pdf