Against All Odds: Stories of Grad Admissions Hope

[necessarily possible]

1 hour ago, Turretin said:

Yeah, I've been thinking of this as well! I applied to 17 programs, in which I figured that the "average" acceptance rate was 6.5% (Michigan is 2%, but Michigan State is 15%), if chosen randomly. 17 x 6.5% = 110.5%. There's something wrong with the probability, but I haven't worked it out. Because 100% means certainty, and you can't have it go above that; moreover, it is a priori uncertain, so it can't be the upper bound, 100%, either.

The right way to do this is to calculate the probability of being rejected from all and then subtract that from 100%. That is, 1-(0.935^17) = 68%.

Edited January 20, 2017 by quine

[dgswaim]

There are a lot of problems with thinking about "chances" in terms of raw probability calculus. For instance, a 6% admission rate does not imply that some person s has a 6/100 chance of being admitted. That would be true if admissions just involved random selections of 6 individuals from a 100 person sample space, which is not the case. S's likelihood of being admitted is conditioned by the results for applicants whose pool of applications overlap with that of s, whether or not schools divide applicants by area, whether s's advisor is taking new students, considerations of diversity and general department make-up, etc. 

Edited January 20, 2017 by dgswaim

[matchamatcha]

What I'm taking away from this thread is that I probably have better chances of being accepted were the process actually a totally random pick a name out of hat sort of scenario.

[Duns Eith]

4 hours ago, Kierkegaardashian said:

I think the issue is that the probabilities don't "stack" in this way. If you have a 6% chance of admission at each of the ten schools you apply to, you have a 6% chance of being admitted, not a 60% chance.

The probability I was considering was the event being accepted anywhere. Imagine you've got 10 different lotteries or raffles, each with a probability of winning 10%, you've got a probability increasing for each ticket purchased, because you may win multiple times, for one of them, or not at all. But the probability of winning at least once is increased with every ticket.

3 hours ago, quine said:

The right way to do this is to calculate the probability of being rejected from all and then subtract that from 100%. That is, 1-(0.935^17) = 68%.

If they are statistically independent, this is exactly right: we're interested in the denial of the probability "I will get in nowhere". This also intuitively means that the more applications you put in, the resultant likelihood of being rejected everywhere approaches zero (asymptotic curve).

The intersection is at 10 applications when the average is 6.5% acceptance, when applicants are chosen at random.

1 hour ago, dgswaim said:

There are a lot of problems with thinking about "chances" in terms of raw probability calculus. For instance, a 6% admission rate does not imply that some person s has a 6/100 chance of being admitted. That would be true if admissions just involved random selections of 6 individuals from a 100 person sample space, which is not the case. S's likelihood of being admitted is conditioned by the results for applicants whose pool of applications overlap with that of s, whether or not schools divide applicants by area, whether s's advisor is taking new students, considerations of diversity and general department make-up, etc. 

Right. There are a lot of conditional probabilities we may figure that make the "acceptance rate" figure less than helpful/meaningful.

Edited January 20, 2017 by Turretin

[dgswaim]

24 minutes ago, Turretin said:

The probability I was considering was the event being accepted anywhere. Imagine you've got 10 different lotteries or raffles, each with a probability of winning 10%, you've got a probability increasing for each ticket purchased, because you may win multiple times, for one of them, or not at all. But the probability of winning at least once is increased with every ticket.

 

So you're saying I should buy lots of lottery tickets. Got it. Will do.

[TK2]

1 hour ago, dgswaim said:

There are a lot of problems with thinking about "chances" in terms of raw probability calculus. For instance, a 6% admission rate does not imply that some person s has a 6/100 chance of being admitted. That would be true if admissions just involved random selections of 6 individuals from a 100 person sample space, which is not the case. S's likelihood of being admitted is conditioned by the results for applicants whose pool of applications overlap with that of s, whether or not schools divide applicants by area, whether s's advisor is taking new students, considerations of diversity and general department make-up, etc. 

Right, but you don't know any of that, or how to assess it, or how to position yourself on the list of applicants. All you know is the X of Y are admitted, which makes it essentially a lottery as far as you're concerned.

[Schopenhauerfanboy]

2 hours ago, TK2 said:

Right, but you don't know any of that, or how to assess it, or how to position yourself on the list of applicants. All you know is the X of Y are admitted, which makes it essentially a lottery as far as you're concerned.

I see where you're coming from.

My position is that needing "knowledge" seems too restrictive. Seems like having adequate supportive reasons would be enough to make at least a reasonable estimation. Every year, some number of untenable applications are submitted. For instance, students who barely make GPA cutoff, no research experience, subpar GRE, an all around "vanilla" application or worse. Why would you assume your chances of getting in are equal to that applicant's if you have superior stats? I mean, of course, you don't have knowledge -- there are cases in which an inferior applicant in other respects has an outstanding writing sample or whatever the case may be. But suppose you have 90th percentile scores, perfect GPA, letters from famous profs, etc. Are you really suggesting that such an applicant's chances are just as likely as every others' ? Sure, unbeknowst to you, an applicant pool could be particularly good one year (everyone has perfect stats!), but that seems like an extremely rare occurrence that shouldn't figure in your considerations. Maybe I am missing something. 

[TK2]

oops double. messed up the quote.

Edited January 21, 2017 by TK2

[TK2]

7 hours ago, Schopenhauerfanboy said:

I see where you're coming from.

My position is that needing "knowledge" seems too restrictive. Seems like having adequate supportive reasons would be enough to make at least a reasonable estimation. Every year, some number of untenable applications are submitted. For instance, students who barely make GPA cutoff, no research experience, subpar GRE, an all around "vanilla" application or worse. Why would you assume your chances of getting in are equal to that applicant's if you have superior stats? I mean, of course, you don't have knowledge -- there are cases in which an inferior applicant in other respects has an outstanding writing sample or whatever the case may be. But suppose you have 90th percentile scores, perfect GPA, letters from famous profs, etc. Are you really suggesting that such an applicant's chances are just as likely as every others' ? Sure, unbeknowst to you, an applicant pool could be particularly good one year (everyone has perfect stats!), but that seems like an extremely rare occurrence that shouldn't figure in your considerations. Maybe I am missing something. 

Well, you said it yourself - there are a lot of confounding variables which make 'the stronger applicant' an almost nonexistant entity. Maybe you get a commitee member with a particular disdain for GREs. Maybe the person who would have been the ideal supervisor just decided to quit academia and move to Madagascar to run a beach bar. Maybe your SoP is a grating mess and you can't tell because you've been over it so many times. Maybe people really don't apply if they don't feel ready. Maybe no one is actually 'vanilla'. You don't know. Why assume oneself to be advantaged in some way for the purpose of dry odds?

(For example, my more-or-less top choice program actually has one of the higher-middle admission rates in my field, mostly, I would guess, due to unusually large admitted cohort sizes (relatively). So I can be optimistic, right? I've applied to technically more selective programs where I feel I'm a strong candidate. But it also has an official GRE cutoff of 5.0 on the AWA. That's the 93%. Are they for real? Is no one with a 4.5 applying? Do they actually throw out those 4.5s? When they report 100 or whatever applications, are they getting those 100 entirely from the 7% of the would-be-grad-student population getting past their crazy cutoff? And...who the hell knows, therefore, how I measure up in a crowd of people who all have 93% GREs. So, all I know is that they have a roughly 8/100 admission rate, and that's what I work with in the long, dark compulsive unstructured snacktime of the soul over here while procrastinating by playing the odds game.)

[TwoTimesTolstoy]

I think it's a bit obsessive and pointless to try to work out one's "odds".

The biggest point to be made here is that the process is far from cut and dry. Stats really don't mean shit. Nobody cares about your GRE and GPA if other applicants are a much better fit for the program. Nobody cares about how famous your letter writers are if they consider your writing sample to be uninteresting and/or underdeveloped. Nobody cares about your publications/institutional prestige if your statement of purpose is blunderous or you don't seem like a good match for the people in the department, both in terms of research and personality (I've talked to people on hiring committees who swore up and down that the person who eventually gets the job is often the person, of the finalists, who got on with the current faculty best -- they're building an intellectual community, and they want people who seem like they'd get along great).

Moreover, I think this obsession about an applicant's "stats" has grown out of the "How exactly does this process work and what exactly do I need to be successful?" mentality, which is, in many aspects, absurd and dangerously obsessive. GPA/GRE/University prestige (and even publications!) give very little, if any, indication that one is a good philosopher; one shows their competency as a philosopher through their writing sample, a well-written statement of purpose, and writing interesting papers and making interesting contributions in classes that your letter-writers will mention. There are too many very gray aspects of consideration that hugely weigh in on the determination of admissions decisions to say that "one is likely or more likely than most to be successful with x, y, and z." As in myriad situations, the attempt to quantify things is not only unhelpful, but also serves to cover up other aspects of the situation that matter quite a lot.

In the fictional words of Albert Camus (taken from my favorite edition of Existential Comics, "The Analytics at the Bar"), "It's more of an art than a science, old boy."

Edited January 21, 2017 by TwoTimesTolstoy

[Dialectica]

I created this thread assuming (as I think we all should) that the odds are low at worst, or, at best, completely unknown. The idea is to think of anecdotes that give hope to those who feel self-conscious about some portion of their file (e.g., low GREs) or how their season is starting out (e.g., only a few waitlists, or only rejections). There is hope even with low stats or unknown undergrads. There is hope even if you are rejected from low-ranked institutions (see the first story posted). 

[Duns Eith]

9 hours ago, TwoTimesTolstoy said:

I think it's a bit obsessive and pointless to try to work out one's "odds".

Yes.

And that's why we are doing it.

[dgswaim]

22 hours ago, TK2 said:

Right, but you don't know any of that, or how to assess it, or how to position yourself on the list of applicants. All you know is the X of Y are admitted, which makes it essentially a lottery as far as you're concerned.

I don't need to know what criteria condition the probability of admission, however labyrinthine. I merely need to know that these criteria are in play, whatever they are. This is enough to see that admissions is not like a lottery.

[TK2]

21 minutes ago, dgswaim said:

I don't need to know what criteria condition the probability of admission, however labyrinthine. I merely need to know that these criteria are in play, whatever they are. This is enough to see that admissions is not like a lottery.

But you don't really know how they're in play. Even the easily quantifiable parts - GPA, GRE - are relative to the entire pool. Maybe a score you think is good is actually middle of the road. Maybe one you think is poor is actually pretty good. Maybe this particular committee decides to only use GREs as an intial strainer and genuinely throws them out afterwards, making that part of the application partiallyh moot. And the qualitative parts - work, statements, letters - you really have no way to assess at all, since they need to fit with the entirely idiosyncratic preferences of each particular program and comittee. So you know what factors are in play, but the best you can do in asessing them is to roughly assume some weaker stuff balances out some stronger stuff, feel reasonably confident that a high GPA is better than a low GPA, take the value of the GRE with a grain of salt, and assume your attractiveness basically averages out across programs given that you meet the minimal criteria.

[dgswaim]

1 hour ago, TK2 said:

But you don't really know how they're in play. Even the easily quantifiable parts - GPA, GRE - are relative to the entire pool. Maybe a score you think is good is actually middle of the road. Maybe one you think is poor is actually pretty good. Maybe this particular committee decides to only use GREs as an intial strainer and genuinely throws them out afterwards, making that part of the application partiallyh moot. And the qualitative parts - work, statements, letters - you really have no way to assess at all, since they need to fit with the entirely idiosyncratic preferences of each particular program and comittee. So you know what factors are in play, but the best you can do in asessing them is to roughly assume some weaker stuff balances out some stronger stuff, feel reasonably confident that a high GPA is better than a low GPA, take the value of the GRE with a grain of salt, and assume your attractiveness basically averages out across programs given that you meet the minimal criteria.

Sure, but this is all relative to the epistemic situation of the applicant. If I'm an applicant and I want to perform a calculation of my odds of being admitted, then it might be reasonable to make some simplifying assumptions about the process given my limited knowledge of what precisely is being considered. That said, we all certainly know enough about the admissions process in general to be reasonably certain that ceteris peribus clauses like these don't accurately capture what's happening. 

[Schopenhauerfanboy]

14 hours ago, TK2 said:

Well, you said it yourself - there are a lot of confounding variables which make 'the stronger applicant' an almost nonexistant entity. Maybe you get a commitee member with a particular disdain for GREs. Maybe the person who would have been the ideal supervisor just decided to quit academia and move to Madagascar to run a beach bar. Maybe your SoP is a grating mess and you can't tell because you've been over it so many times. Maybe people really don't apply if they don't feel ready. Maybe no one is actually 'vanilla'. You don't know. Why assume oneself to be advantaged in some way for the purpose of dry odds?

(For example, my more-or-less top choice program actually has one of the higher-middle admission rates in my field, mostly, I would guess, due to unusually large admitted cohort sizes (relatively). So I can be optimistic, right? I've applied to technically more selective programs where I feel I'm a strong candidate. But it also has an official GRE cutoff of 5.0 on the AWA. That's the 93%. Are they for real? Is no one with a 4.5 applying? Do they actually throw out those 4.5s? When they report 100 or whatever applications, are they getting those 100 entirely from the 7% of the would-be-grad-student population getting past their crazy cutoff? And...who the hell knows, therefore, how I measure up in a crowd of people who all have 93% GREs. So, all I know is that they have a roughly 8/100 admission rate, and that's what I work with in the long, dark compulsive unstructured snacktime of the soul over here while procrastinating by playing the odds game.)

The points are well-taken. 

My point is that if you have perfect stats or close to perfect stats you have very reasonable grounds to believe your applications are more propitious than someone who doesn't. I don't see how that's controversial. 

Here's another reason to think your chances might be higher than the baseline probability of 5% or whatever it might be.
I was told by a number of DGS that you have better chances of getting in if you're working in an area to which the department receives less applications. One department I was told consistently gets an overwhelming majority of applications to Philosophy of Science, M&E, etc, and far less in the history of philosophy, but the department still allots a number of spots to historical students. So if you're working in history, you'd have good grounds to think there's a lower applicant-to-spot ratio.

[The_Last_Thylacine]

Any cool acceptance stories? 

[dgswaim]

14 hours ago, desolesiii said:

Any cool acceptance stories? 

There was a chap on here during the 2014 (don't remember his handle) admissions cycle that submitted 10 applications, all to the top 10 PGR programs. He was admitted to all 10.

[Dialectica]

3 hours ago, dgswaim said:

There was a chap on here during the 2014 (don't remember his handle) admissions cycle that submitted 10 applications, all to the top 10 PGR programs. He was admitted to all 10.

Somehow I have the feeling this story *won't* give me hope when the decisions start rolling in. 

[The_Last_Thylacine]

3 hours ago, dgswaim said:

There was a chap on here during the 2014 (don't remember his handle) admissions cycle that submitted 10 applications, all to the top 10 PGR programs. He was admitted to all 10.

Admittedly, I'm slightly envious of people like that---who have already attained that level of obvious talent. Given the choice between every top 10 program, I would pick USC every time. 

That will be a difficult rejection letter to read if (when) it comes! 

[The_Last_Thylacine]

I read about a guy in the previous years' posts who turned down Harvard so he could stay close to his girlfriend. What a selfless man. 

[dgswaim]

3 hours ago, desolesiii said:

Admittedly, I'm slightly envious of people like that---who have already attained that level of obvious talent. Given the choice between every top 10 program, I would pick USC every time. 

That will be a difficult rejection letter to read if (when) it comes! 

The only top 10 program I applied to was Princeton, and I'm sure I didn't even come within a whiff of getting in. 

[lyellgeo]

Apologies to responding to an old thread, but I just wanted to write my thoughts on it. 

The original idea was that you take acceptance rate and calculate the odds of a shutout (applying to 15 programs with 5% acceptance rate, odds—assuming independence—would be roughly 50/50 of a shutout.) That, combined with the fact that one's "numbers" may well be better than average, seems to make the chances of being shutout fairly low. 

I think that approach is methodologically flawed, though not for the reasons others were saying. There's nothing wrong in principle with trying to calculate the odds of getting shut out. What one should do, however, is calculate the odds of themselves being accepted at each individual program. In order to be accepted, one has to be accepted outright or off of the waitlist—in other words, you need to be one of the top 10 or 15 (depending on the program) students applying to that program, when factoring in for program fit and ability to do graduate study. (Obviously this will be highly dependent on yourself and the specific program.) 

This difference may seem small but I think it's important. On the one hand, someone with good all-around stats (say, in the top 20%) may have an extremely low chance of actually making it through the final cut. I'm not sure what this application would look like exactly, but you could have someone (call them A) who, for example, consistently makes it to the final round before not being admitted—even if A were consistently one of the top 20 or 30 applicants, they could still easily be shut out from all the programs they applied to. On the other hand, you could have someone (call them B ) who has relatively bad stats and who frequently gets tossed in the initial cut (say, half the time) but who, after surviving the initial cut, often makes it to the top 10 students under consideration. By contrast, B could still get accepted in at several programs that match their interests. 

I think this is a source for both optimism and pessimism. On the one hand, being a "good" applicant by the numbers basically guarantees nothing—even with exceptionally good stats, someone could easily get shutout if their application doesn't stand out further on in the process. On the other hand, someone with rather average stats could still have a good chance of getting accepted somewhere (even multiple places) if they think that they have a decent chance (assuming they can survive the initial cut) of standing out later in the process. Again, the question of whether someone can "stand out" in order to be seen as a top applicant really depends on a) the non-quantifiable aspects of that person's application (writing sample and personal statement) and b ) the specific program (its values, specializations, style of doing things, the extent to which it cares about undergraduate program prestige, and so on). So, not that that helps anyone wanting specific numbers, but I think that can help explain both why the application process can be so frustrating, with good applicants seemingly getting rejected from everywhere, as well as why there are still so many stories of someone who gets in "against the odds" at a top program.

Edited January 10, 2018 by lyellgeo