Wednesday, March 11, 2009

Theo gets Quantitative about grad school

As the result of multiple requests, I've been attempting to list and rank the graduate school options that I have, as I near the point of having to make up my mind. Being who I am, I think probably I've gone wayyy overboard (that's what you get for asking), but the results are provided below if you're interested/have enough stamina. I wrote it up as a mock lab report/paper, mainly to amuse myself.

- Theo

Quantitative grad school selecion: Or, mathematical objectivity abused


Selecting the right grad school is important if you want a PhD and a career as an academic. I want a PhD, and hope to have a career as an academic. Therefore, it would be good to make the best possible choice of graduate schools, from the available options, based somewhat objectively on a set of criteria.


I brainstormed a list of a bunch of separate features of a graduate school that I think are important. I grouped them into hierarchical groups, and assigned them weights within groups and levels, based on their comparative importance to me. These numerical weights, with higher values corresponding to increased importance, were converted to proportion weights, multiplied down the hierarchical trees. These trees and integer weights (for readability) are reproduced below:

Generally, these features were grouped based on whether they were academic, non-academic, or "discretionary" - ie, my chance to subjectively incorporate my gut feelings, without assigning a particular descriptor to them. (This included such things as the presence of a vet school Eugenie might want to go to - Shhhhh; she doesn't like it when I think that way).

Having previously eliminated 2 of 5 schools (one for general uptightness and location, the other for location and insecurity of funding), I had 3 remaining schools to choose between. For each school and each important feature, I subjectively assigned a score between 0 and 4. Results were determined by taking the product of the weight and the score for each category and school.


It is with some chagrin that I report the results by way of excel figures, but I'm too tired/busy to do any differently, and it's better this way than showing the numbers.

By sub-category then:
Major take-homes here - School2 gets the highest ranking for the academic category, followed by School1. In the non-academic category, things get shuffled around; School1 takes the lead by a good bit over School2

And now the overall results:

For the quantitative folks out there, this corresponds to:

To help this make sense, a perfect school would score a 4 in every category, and have a raw score of 4, or "percentage" of 1.0. (Yes of course these calculations require ~6 sig figs!!! Naturally.)

This, based on this, I should go to School1 (followed by School2 and School3. It's coincidence (?) that I numbered them in that order, actually).


I tried as much as possible to be very objective about this ranking, and to trick myself. That's why I tried to add so many categories, and make the weighting system rather complex, so that I couldn't predict ahead of time what the result would be. I also made the categories and weights first, and then later rated the schools all at once, before doing any calculations. That being said, the points assigned to each school in each category are necessarily subjective. For example, I much prefer rural to urban areas, so in the "Urban/Suburban/Rural" category, rural corresponds to 4, and, for example, Los Angeles would be a 0, or 1. Not all of the categories are independent of each other; I'm not sure how big of a problem this causes, but whatever (ie, rural, a good thing, corresponds by necessity to less public transportation, a bad thing).

All that being said, these results do a fairly good job of reflecting the more intuitive response/ranking I've been accumulating over the last week or so of trying not to think about it all.

School2 would be a new place for me, I got along fantastically with the advisor, and enjoyed meeting and talking with the various members of the lab group. The lab does a solid mixture of theoretical and mathematical work. It's part of a big department, at a university where collaboration between faculty of different departments is very common and encouraged. I'd be free to do pretty much anything I wanted to, including seeking out collaborations with other students and faculty. And, rather excitingly, there are some students doing theoretical work. I'm not used to being able to discuss what I do with fellow students with a similar level of understanding (usually I have to talk to prof's or post-docs, and there aren't that many of them either). It was really nice to talk theory with people who were roughly my peers. Could be a lot of fun to be around more theory people. While I'm sure the funding situation is easily sufficient for me to live on happily, it's not as solid as the offer I have from School1 (more on this later), and probably would require more teaching. My advisor there regularly obtains grants, but if the work I was doing wasn't related to those grants, I'd have to secure my own funding, or TA. [Another intangible, that bothered me a bit, but doesn't make sense to the logical part of my head... every one of this advisor's grad students is either married (with kids), or in a serious long-term relationship, and several years older than me... makes me a little worried that no one would want to hang out with a "single" youngster; "single" because at least for the first year of graduate school, Eugenie will still be at SLAC and I'll be bach'ing it.]

School1 is the school/biological station where I currently work. So I'm very familiar with it: I know the graduate students, several of the faculty, and the area quite well. It's a rural location, which I like. I've been offered a set of 3 fellowships, and I'd be certain of solid funding during my PhD here, including two years guaranteed to be free of teaching or research duties. The cost of living here is also significantly less than at School2 (check out this fascinating website for COL comparisons: link). I'd work with the same advisor that I have right now. I get along with him well; he's very very sharp and I learn a lot. He does almost exclusively theoretical work, and thinks in terms of a system that I'm not terribly enthusiastic about. However, he's said before that he'd be supportive of a co-advising situation, where I'd pick out a second faculty member (such as the one next door) that does empirical plant ecology, something that I want to do more of. His lab group is small, but because of the small size of the station and associated student community, the graduate students behave like an extended lab group and there's quite a lot of interaction. And people reallly want me to come/stay here. After my first two years, I'd relocate to the bio station, so auditing math courses would be more difficult, as the main campus is over an hour's drive away.

School2 would be meeting new people and making new connections; I've already spent 2 summers plus the last 3 months at School1's bio station. Both universities have vet programs, unlike school3, although Eugenie will be glad to know that even when I removed that from the calculations, the rankings of schools stayed qualitatively the same. So there.

School3 is the closest to home, and reminded me a lot of places that are very dear to me. I have a good friend there, and it's a smaller city than School2. The academic fit isn't as good, although realistically, I could probably do what I wanted to there as well. The advisor is a partial member of the stats department, and quite mathy, so I could get my math fix. Funding is intermediate between the other schools. Lab group is pretty focused on a particular topic. I only met 2 of 3 grad students, and none of the post-docs, whereas at School 2 I met pretty much everyone (3 grad students, 2 postdocs). No vet program. Pretty campus. I'm actually surprised it didn't rank higher, but maybe it's paying the price for not being the most recently visited school (School2), or the place where I live (School3). People sounded quite excited to have me there, which was awesome. I dunno.


What do you all think? How important are academic vs. non-academic concerns in making a decision? If I weighted academic concerns even more heavily, School2 would win; currently it's the "best" academic fit, and the "worst" non-academic fit. As it is, giving non-academic concerns 35% and academic 55%, School1 "wins".

I'm not certain that any of the numerical differences in rankings is sufficiently large enough to reach a solid conclusion, I don't know. I know I could make any of these places work happily.

Feed back would be most welcome....

No comments: