One thing that I don't like about the US News college rankings is that they use objective criteria with arbitrary weightings. There's no credible rhyme or reason why these criteria matter, especially not in the proportion they do. Nor is there a good reason that these weightings should subtly change from year to year.

For this reason (plug coming), I created the algorithm that Parchment uses to rank colleges for the Student Choice Rankings [1]. It uses (voluntarily submitted) data from students who got into multiple colleges and chose one to attend. It treats this as a set of games in which one college "wins" and others lose. It's derived from the Elo chess ranking method [2].

As people will no doubt argue, voluntarily submitted data has its limitations. I acknowledge those but for this type of project, sample size is key, and I have yet to see another project with anywhere near the same quantity or recency of data. Perhaps in time that will change.

1 = https://www.parchment.com/c/college/college-rankings.php?pag...

2 = https://en.wikipedia.org/wiki/Elo_rating_system

A follow up on ranking strategies.

This approach was first used for college rankings by Avery, Glickman, Hoxby, and Metrick. I forgot to acknowledge them in the first post.

What we use is an algorithm that says, "For this desirable student (objectively 'desirable' to the schools that were impressed enough to admit them), which school did they choose?" I see this as the "virtuous" approach. In a sense, in this model, schools are rewarded for admitting desirable students and winning the competition for them.

This hinders gaming of the system because most things that you can do to game this generally helps students. Want to game it by admitting students who won't get admitted to your competitive peers? You'll fall in the rankings, because those students will pit you against 'weaker' opponents, giving you less of a chance to rise in the rankings even if you always win.

There's another ranking approach that is fun but naughty. I'll call it the "vicious" approach. This one takes all colleges that admitted a student and all colleges that rejected them. The ones who rejected the student are modeled as the winners, and the ones who accepted them are modeled as the losers. In this way, the colleges that are most selective in admitting a student rise to the top.

I've actually run this "vicious" algorithm, but we don't publish its results. (Gaming this one would hurt students: imagine you're City College. You could do well for yourself by getting all the future freshmen of your rival, State College, to apply to your institution, then rejecting all of them. The kids wasted money on their application, you probably rejected some great candidates just to game the rankings, and State College unfairly drops in the rankings.)

It seems like this would weight financial aid rather heavily? Maybe that's a good thing, but doesn't really deal with the quality of the school.

That's correct: anything that makes a college desirable vis-a-vis its competitors will factor in here. We don't attempt to control for that (that's, currently, intentional).

Avery, Glickman, Hoxby, and Metrick do control for that, as their goal is to use this to get an isolated notion of "best". For now, I prefer to answer the holistic question of, "Which schools do students prefer?"

If we get to a future in which some schools are paying students admitted to Harvard $100k to attend their institution just so they can rank at the top of the Parchment rankings, maybe we'd want to reconsider.

We are already in that situation vis a vis schools with merit scholarships vs. Ivy Leagues.

Harvard already gives financial aid to all of its lower-class and most of its middle-class students ... and they define middle class as family income in the $150k range, if I remember right. I don't think merit scholarships are really going to make anyone else directly comparable.

Of course, Harvard is rather unusual in this regard. I think only Princeton and a tiny handful of others have a similarly generous policy.

Not only are the weightings arbitrary, but many of the criteria bear no relevance to quality of undergraduate education. As an engineering undergrad, why would it matter that my alma mater's teaching hospitals were among the best in the world and raked billions in research grants?

Rankings are a terrible way to select schools, not just because they don't even claim to rate quality of education, but also because many top schools explicitly IGNORE any measures of success that aren't included in major ranking schemes, whereas schools with no hope of ever cracking top 20 are forced to craft a different niche for themselves.

Very interesting, but I think the flaw with this method is that many students will still make decisions based in large part on U.S. News.

For example, I would guess that a non-negligible amount of students that get into Harvard (#1) and Columbia (#4) will choose to go to Harvard in large part because they beat Columbia in U.S. News (and thus they perceive it has a better reputation).

So despite your interesting methodology, it will still be tied somewhat to the U.S. News rankings.

Yes, I agree that students' perception of how others would rank colleges is a serious factor that probably goes into their decision process. This type of ranking will tend to reflect all of those factors that go into the students' decisions.

Using just the data that we and US News make public, one could probably do a pretty interesting adjustment for US News rank. (We publish the Elo points for each college, so you wouldn't be limited to just rank-based methods.)

I don't think teenagers are the best people to ask when it comes to determining the best institute of higher education. It's no wonder that schools with lax requirements like Stanford and Brown seem to get a boost when kids are the ones choosing.

I hate to be 'that guy', but I think most would agree that those rankings are silly. They may paint a general picture, but the number and order of the rank is not a very accurate measurement of quality.

Observation: In the CS ranking, UBC is ranked above the University of Waterloo, though both ranked below U of T? I'm skeptical. (Disclaimer: I go to UBC).

While I agree that the rankings are contrived, I don't think this flaw obsoletes the purpose of the poll, which seems to be to get some picture of the distribution in the HN population.

Regardless of whether those rankings are accurate, most of us let them inform our college decision to one degree or another. Our affiliation with the various subsets of schools does suggest some traits about us and hence the HN population.

The exact ranking is somewhat arbitrary, but the rankings do generally put the schools most people consider as "top schools" at the top. Comparable schools have somewhat similar rankings.

But at the end of the day, the education is only as good as the effort you put in. You can graduate from a school that isn't ranked highly and still be incredibly successful

let's fight subjectivity with subjectivity

Touché

The rankings are more than a little silly.

I don't know american schools (I'm UWaterloo) but the difference in Canadian schools seems to be whether you went to a school that took their program seriously or one that just has it their to round out another program (computer science at Wilfred Laurier for instance).

Good point. That's pretty silly. It's fun to see these rankings, but they definitely have strange arbitrary weighting.

I'd love to see thoughts on ranking CS schools too. I always see MIT/Stanford posts but not much else.

I agree. Talking about schools and their rankings is very counterproductive. It measures the wrong metrics.

I went to Wharton/Penn for undergrad, which favors me in these kinds of surveys, but I'm still against them.

You should be prouder, UPenn is one of the best state schools in the country!

By what measure which we can trust?

It's an Ivy jab. UPenn is a private institution, hence the joke. The lore is that UPenn students are offended when people refer to their alma mater as Penn State, which is a state-funded institution.

Edit: I have no affiliation with either institution and think it's all irrelevant, other than from a cultural awareness perspective.

Once you actually study at multiple universities in these rankings you realize what a joke it is. I can't believe Penn is ranked so highly, outside of Wharton. I was recently there for CS coming from a UK top 5 in CS and Penn CIS was a joke. Also, Oxford > Cambridge for CS? Wat. When you get to about the top 50 universities, it becomes so specific to the strengths of the particular department that these universities really can't be ranked in a single field.

Oh whoops totally missed the state school phrase! (and I know the joke too!)

(imo Penn State is a fine - as in very good - engineering school)

RIP JoePa.

People who don't care about rankings don't usually go to the trouble of distinguishing Wharton from Penn (i.e. "Penn wasn't the highest ranked school I got into.") /light hearted jab

You missed a massive category: the schools ranks 20-100. I would describe a school ranked #40 or #60 in a major as really good. Not top 20 good, but really good. There is a bigger difference between the #50 school and the #200 school than there is between the #1 school and the #50.

In retrospect, I wish I added that initially. But at this stage, enough people voted that the number wouldn't be perfect.

Tiny technical school in Hoboken, NJ. Just outside of NYC. I've ran into a few alumni on here (small world).

I like it here, I'll admit - I was bummed that I was waitlisted for CMU, and it's still something I think about to this day. But I really, really like it here, and it's led to some interesting professional opportunities.

Rankings are bogus, this place felt right.

Along with CMU, Berkeley EECS is also not Top-20. Perhaps this list is not the best match for HN priorities.

Out of curiosity, why do you say that?

I ask because I'm planning on going there next year for computer science.

Nobody should pay any attention to these rankings.

I didn't know that US News were considered a reliable source of ranking schools (if schools can be ranked at all). Isn't US News a for-profit magazine that has figured out SEO really well?

None of the rankings are "reliable" at all, but since they're all flawed, might as well go with the one that is the de facto "standard" within the US.

Their ranking has been around (and well-known) since at least the 80s.

I dropped out of college, having obtained most of a BS. Thank the stars I didn't load up on student loans. I went to a public university that isn't notable for much (Univ. of Houston). My own opinion is that it was mostly pretty average with a few areas of quite-goodness. I don't regret going. I do wish I hadn't arrived so unprepared.

What about using the global rankings[1] - though they are admittedly biased against liberal arts school, they are less manipulated than the US News rankings.

1 - http://www.timeshighereducation.co.uk/world-university-ranki...

Since I'm pretty un-connected to this account, going to start undergraduate at Cornell. While it may not be Harvard/Yale/Princeton, I'm not going to complain (however there is still some element of jealousy/annoyance).

What is HN's opinion of Cornell, especially vs other Ivies?

PG went to Cornell IIRC

I actually prefer this college ranking: http://toptestprep.com/best-colleges-rankings/

It combines both liberal arts colleges and national universities. It's less biased, and more objective.

Most of the stuff that I learned for my job happened either during internships, self study outside of classroom education or on the job. College Education for CS Majors is years behind the industry requirements.

That's pretty much the way it is with nearly every major. Plus depending CS degrees exist on a sliding scale between Math, and Programming.

I think a very math/theory heavy approach is better overall, but it's not a vocational training program

I would argue against making CS classes heavier on Math/Theory side of things unless you are going for a PhD. In the industry we need to follow the footsteps of Stanford and offer more practical classes such as iPhone Programming, Facebook App Development, Machine Learning...I would add to the list new technologies such as HTML5 Single Page App Development, NOSQL Based programming, Android Programming etc

I went to a university that is largely unknown outside of the UK. Not that it matters particularly, because I didn't even study Computer Science.

'What did you study at college' might make for an equally interesting poll.

Since when is U Chicago #4? Not that it isn't totally deserved, but when I was applying to colleges I don't even remember it being in the top 15, and changes in the top 10 aren't exactly common...

Since the new administration decided to start gaming the rankings. Admissions rate now stands at 8.8%, down from mid-forties when I was admitted only eight years ago. Part of that was from switching to the Common App, part of it was riding the coattails of Obama.

It is a great school. But until recently, rankings weren't really something that the administration paid much attention to.

UChicago was top-10 when I was applying for colleges, and to be honest I don't give much credence to the exact ranking. The set of schools in the top 20 match roughly what I would expect (saying top 5, for example, would exclude schools like MIT and Stanford)

I am still studying (last year) at the University of Gothenburg, Sweden.

I am annoyed, my school was #20 when I went, now tied for #24, so I still picked the first option. University of Virginia (at the time was tied for top ranked public university in country)

School of hard knocks

If we are currently undergraduates, should we answer "attended" or "do not hold a degree." It is important to be consistent.

Question has been edited

Though I want to note that the US News and World Rankings have gotten a lot of flack for the issues with their methodology.

I don't disagree, but the entire process is pretty subjective (there are many variables to take into account, and weights are somewhat arbitrary). USNWR provides a concrete list for the discussion and I think the top 20 captures the spirit of the original question.

For the purposes of this question you are probably correct.

I went to a tiny liberal arts school no one outside Texas has ever heard of. Majored in CS.

Caltech

The Open University.