irfannooruddin

I can not and will not disagree with any of this. I think you're correct in most of what you say. It's one of the reasons I place perhaps outsized importance on the statement: it's the one part of the file that should arguably be untainted by things like pedigree. Believe me also when I tell you that those of us who serve on admissions committee, and, more generally, work with PhD students, spend a lot of time worrying about the fact that our admissions processes are less than perfect. With attrition rates as high as 60-70%, graduate school is a very inefficient allocation of resources. If inventing a better process would allow us to "predict" better and therefore to allocate resources more "efficiently", we'd jump. One option, which I know schools like Emory and Colorado use in some form or the other, is to incorporate in-person interviews for some shortlist of candidates, either to determine final admission status or final funding. Do people have other ideas?

+1000 Two other thoughts: 1) Dan: It's the Michigan, not OSU, mafia. We're just expanding our turf. 2) On the LOR question: Does the identity of the letter writer matter? Possibly. There is such a thing as source credibility after all. I mean if I get a letter from, say, Sam Huntington (yes, I know he's passed) saying that student X is the best student he's taught at Harvard in 50 years, I'm guessing it would make a much stronger impression on me than if I got a letter making the same relative claim from a professor whose name I didn't recognize at a school whose reputation I don't know as well. But disentangling the Huntington effect from the Harvard effect from the fact that said student is clearly a standout is very hard, and you can decide for yourself how you want to attribute that causality. More generally, we can't control the identity of our letter writers in ways that would truly matter. I attended Ohio Wesleyan, a decent liberal arts university. My letter writers would not have been recognizable to the average committee member (for one thing, none of them were political scientists). It's irrelevant whether a letter from Huntington would have been better for me; that wasn't an option. So my goal was to give my letter writers the most ammunition I could so they could write the best letter they were willing to write (in other words, I followed Dan's advice above to the extent possible).

+1 Go blue.

Pretty sure there isn't that much detail for the subfield rankings, or at least there wasn't the last time I bothered to check. The NRC rankings, even if dated, might be more informative.

Bing! Or, more geekily, we're trying to predict something really really hard: your probability of succeeding in graduate school. But the data we have is crappy, and the variables we can access are few and limited: 1) Academic background (which institution, what major, what GPA) 2) GRE scores 3) LORs 4) Statement Let's assume for sake of argument that few applicants to any given grad program are complete non-starters on (1) since typically only decent/good students apply to PhD programs. Further, let's assume that the GRE scores are mainly used to determine funding eligibility (I certainly don't care about them). That leaves LORs and the statement. I've said this elsewhere, so won't repeat myself here, but LORs are limited, albeit still important, because most letters are positive. Where they vary informatively is in the level of detail the letterwriter chooses to provide. But you don't control that, and you can't necessarily predict who will write a "good" good letter and who will write a "bad" good letter. So you're hoping, and it's fair to assume that some great candidates are undermined by less than great letters (again, where great means informatively great). So the statement looms very large, and here's the blunt truth: Most statements suck, and it's the single area where even good candidates could make the greatest improvement to their probability of acceptance. When my undergrads apply to graduate school, I make them write up to a dozen drafts and then help them rewrite it once more. But that certainly wasn't done for me when I was a college senior. I wrote my statement; I sent it off. No feedback, no edits, no revisions mandated by my advisors. And I'm quite sure that my statement was far less effective as a result.

Given that BFB has invoked the eternal wisdom of Bill Z., there's little to add, but this is the Internet and so I won't let that stop me. It's not that LORs are useless. It's that most of them contain little useful information because they vary so little. Ask yourself: would you ask someone to write your letter if you thought they thought you an idiot? No. Neither would I. So we all ask people we know like us, which means all the LORs tell us that their student is well above average. At this level, the only informative signal is a negative one (for instance, when someone ticks the box that you're in the top 50% of students they've ever taught) -- trust me, this happens, but even when it does it's striking how at odds this "objective" ranking is with the qualitative assessment in the letter which is often stuffed with superlatives. So the signal-to-noise ratio is low. What I'm looking for is detail. A letter from someone who supervised your senior thesis, or for whom you served as a RA, or who knows exactly what research you did when on your study abroad -- that helps. A letter from someone who recalls you fondly as a fine student who always came to class prepared -- well, that's less helpful. Finally, because YMMV as BFB points out, I'll simply say this: beware the moosehead letter.

BFB summarizes my position on rec letters well. And I agree with dnexon too that SOPs are lousy predictors of eventual dissertations, but, in a very low-information setting, the ability of a student to pose a question clearly and reveal their analytic chops is useful info about their ability to do PhD-level work. What that question is is far less relevant.

I honestly don't think I looked at that question even once. A lot of our applicants left it blank; others wrote in a set of top-5 departments; yet others wrote in a greater variety of departments. I don't care (though I accept the notion that some might). AFAIK, that question exists so the bean-counters in the graduate school can use the answers to generate a set of "peer" institutions, i.e., where do people applying to us also apply? It is not a factor in admissions. Here's how I read the files: 1) Get basic info: name, previous ed institutions, GPA, GRE scores (the latter two simply to ascertain if you're above the arbitrary thresholds established by our graduate school since that has implications for funding, as BFB has explained previously). 2) Read personal statement closely; skim writing sample if provided (OSU doesn't require one) 3) If statement is any good, and if "fit" is plausible, glance at letters to see if any flags pop up. 4) Next file. In short, for me, it's all about the statement and/or writing sample if one was provided. Frankly I found letters of rec fairly uninformative for the obvious selection bias reasons. As my list formed, I had pretty strong priors as to who I thought would get a lot of attention and whom I thought we'd have the inside track on. But that didn't bear on my decision at all, and an admit's "gettability" never came up at the admission committee meeting.

Not a chance that this is true, especially for those two departments. Do we assign weights to the probabilities a given admit will come (i.e., do we have models of the yield?)? Of course, we do. But top departments believe they can compete and do.

Couldn't have said it better myself. +1

I urge you to go for the official visit if you can make it work. This is especially (perhaps only) true if you're genuinely trying to decide between multiple plausible options. You're talking about spending 5-7 years of your life in this place and with these people. Getting to know them, and feeling the intangible "fit" from your end, is important. I recall as if it's yesterday my visit to Ann Arbor. The people I met that weekend clicked; it ended up being a very special cohort (16 of 19 finished their PhDs, which is an unheard-of completion rate), and I truly believe the seeds of that success were planted that recruitment weekend. But, if going that weekend is a hassle, don't sweat it if you can't make it. Just try and go another time.

As I mentioned earlier in response to someone else, you're not alone in being rejected. Some people run the table no doubt, but most don't. I went 3/15 myself, including rejections at 2 of the institutions your sig indicates you're in at. Fortunately for me, I got in to the 3rd and that's all that I needed. One thing that I now realize, that I didn't 17 years ago when I was in your shoes, is how vital "fit" is, and how differently faculty evaluate fit than graduate students and especially prospective graduate students. When I applied, I indicated an interest in studying India as well as more general interests in comparative political economy. Looking back, I realize now that most of the schools to which I applied (essentially the top 10 plus a few safeties) had no one who studied India. I am sure they had plenty of reasons to reject me, but that fact would likely have been sufficient in itself.

Congratulations.

I understand. It's 17 years since my spring of admission decisions, but it still hurts that I went just 3/15. But 1 is all that you need.

One of my students has been accepted to Princeton, fwiw.

For what it's worth, my own take on this is that it's always to your advantage for schools to know what other schools you're in. I know of at least two cases where knowing what our "competition" had offered allowed us to "match" the funding offer. Had that information not been shared, those two individuals would have had less attractive packages.