Close

13 September, 2016

UChicago over Stanford?


UChicago is higher ranked than Stanford (US News) because UChicago happens to be better at generating the right stats for the methodology. You could call it gaming the system, but that’s unfair since it’s mostly not deliberate. According to Parchment.com’s statistics, about one in three students choose UChicago over Stanford, if admitted to both.

Here’s a walkthrough of the indicators that go into the ranking:

Graduation and retention rates

UChicago has a graduation rate of 86%—Stanford, only 75%. The difference is cultural. Stanford is a highly entrepreneurial and career-oriented school. Many people drop out of Stanford to pursue their business ideas or do other things.

Meanwhile, UChicago is known for sending a lot of people to grad school. If you’re trying to get into a good grad school, the last thing you want to do is drop out of undergrad. So the studious culture at UChicago lends itself to a higher graduation and retention rate.

Undergraduate academic reputation

This seems like a very fuzzy section. This section is supposed to account for “intangibles” via an opinion survey of various academics and high school counsellors.

I don’t have the raw scores to compare Chicago and Stanford, but I suspect UChicago would come out on top. Stanford has fantastic academics, no doubt. But historically UChicago has a very close knit and rosy relationship with academia. Everything from the Common Core curriculum to the anti-grade inflation teaching philosophy adds a subtle shade of approval in the ivory tower.

So the fact that academics were surveyed, rather than employers or the general population, gives UChicago an edge.

Another thing that had initially caught my eye in the methodology was this bit:

To reduce the impact of strategic voting by respondents, U.S. News eliminated the two highest and two lowest scores each school received before calculating the average score.

Although I initially suspected that this would benefit Chicago, I had apparently misunderstood this technique. In fact this is a fairly standard way of attempting to fix bias in surveys such as this one.

Faculty resources

Class size is 40% of this category. UChicago wins out on class size—78% of classes at U of C have <20 students, versus 71% at Stanford. UChicago places great emphasis on making classes smaller, and interactions with the professor more personal. I applaud this approach and believe that Chicago’s win here is well deserved.

Another 35% of this category is average faculty salary, adjusted by cost of living. There are two key parts here.

  1. Using average salary as opposed to median salary means that a college is higher ranked if its distribution is positively skewed, i.e. the professors at the very top make a lot of money. I don’t know about Stanford, but I do know that UChicago keeps several high profile professors on retainer by offering them huge sums of money. This drives up the average, even though the median would not be as affected. However this is just speculation.
  2. Cost of living is the big one. Living in Chicago isn’t cheap, but it’s nothing compared to Boston or the Bay Area. That fact alone would greatly skew Chicago’s faculty salaries to appear higher.

Is better paid faculty better faculty? I have no idea.

U.S. News also weighs the proportion of professors with the highest degree in their fields (15 percent)

As I said before, UChicago is very academically inclined. It wouldn’t surprise me if the professors there were higher up the ivory tower, especially compared to a school like Stanford which hires from more diverse backgrounds (possibly with fewer years of formal education).

Student selectivity

This section is worth less than the previous ones. It looks at SAT/ACT, which UChicago wins out on. I suspect this has to do with athletics. No offense to athletes, but their scores just aren’t as high.

Also considered are what percent of the student body was top 10% of their high school class, which is probably subject to the same effect as SAT score. Athletes would drive down Stanford’s numbers, plus UChicago’s reputation as a mega-brain school doesn’t help.

However, Stanford has a lower acceptance rate than UChicago. A rare win for Stanford in the rankings.

Other categories

Per-student spending, I’m sure both schools spend a lot. I can’t seem to find hard numbers.

“Graduation rate performance”, basically US News runs a regression and gets an expected graduation rate. UChicago’s graduation rate is probably higher than expected, due to the cultural emphasis on post-college education, so this nets U of C some more points.

And alumni giving. Personally after our family has paid tuition I feel like we’ve given enough, but for some reason people feel compelled to give even more. I don’t have access to the full data but at 5% of the total ranking this section isn’t that huge of a factor.

6 Comments on “UChicago over Stanford?

fireattack
27 November, 2016 at 2:34 am

>To reduce the impact of strategic voting by respondents, U.S. News eliminated the two highest and two lowest scores each school received before calculating the average score.
>I’m not sure how sound this methodology is. First of all, people could strategically rank their first-place college as third, now that this methodology is public. And second, this methodology benefits a school that comes in consistently third, rather than first or second. Here we can only speculate, but it wouldn’t surprise me if UChicago benefitted from this bit of the methodology as well.

I think you understand this part wrong. It means among all the score ONE school received (from all the surveyees), the highest two and the lowest two are removed before the data is averaged.

Say you have 7 people give scores to schools.

If Standard got 5,6,7,8,8,9,9, her average would be (7+8+8)/3=7.67.

This is a widely used scoring system, namely almost all the scoring Olympic sports, such as diving and gymnastics.

You can’t game it by give your first-place school as third… it’s not removing the highest two of a certain surveyee.

Reply
Sasha
27 November, 2016 at 7:12 pm

Interesting idea, but why remove only the top two scores? Wouldn’t there be far more than, say, 7 people who are scoring the schools?

Reply
josh
15 April, 2017 at 4:52 am

“To reduce the impact of strategic voting by respondents, U.S. News eliminated the two highest and two lowest scores each school received before calculating the average score.”

I think you are interpreting this wrong. It’s how learning curves and scoring is used in developing more accurate figures. Eliminating outliers who might otherwise warp the data: for example, people in CA could rate USC a 5 (from 1 to 5), but someone in North Carolina would rate it a 1. To get rid of geographic biases which could skew data, USNEWS eliminates both the 1 and 5 score. UChicago v Stanford rankings are probably not influenced by this method at all since they both are usually ranked 5 no matter where you are in the nation.

In addition, schools like Caltech, MIT, UChicago, all of which are NOTORIOUS for their grade deflation NEED to look for higher SAT scores to validate that the students they are accepting can handle the rigor. I thus disagree with your unsubstantiated reason that those schools accept higher scores to “game rankings.” On the other hand, Stanford is more susceptible to grade inflation as the culture of the school is different: the average GPA is a 3.6+ for heaven’s sake!

Finally, the main difference between these schools is applied (Stanford) vs theoretical (UChicago) approach to learning. I thought that would be so important to address when comparing these schools.

Reply
Sasha
20 April, 2017 at 6:37 pm

Thanks for the explanation Josh, I believe the other commenter had mentioned this as well. I will admit that I completely misunderstood this method of adjustment, I’ve edited the page accordingly.

As for SAT scores, I don’t believe we are in disagreement. I did not mean to imply that these schools accept high SATs just to game the rankings, and I’m not sure how you got that from the text. Most of these reasons are not due to deliberate “gaming” and thus I wouldn’t endorse using that term.

Reply
Gerry Elman
27 August, 2017 at 12:46 pm

Hi, Sasha,

About “gaming the university ratings system,” I can’t help but point to a recent book by a friend of mine on the Stanford law faculty. It’s by Paul Goldstein, entitled “Legal Asylum: A Comedy.” https://www.amazon.com/Legal-Asylum-Comedy-Paul-Goldstein/dp/1634256115 Not just “a comedy,” it’s really a farce … about a lawyer who becomes dean of the law school at a state university in the midwest. Seeking to raise the prominence of the school so she can be seen as a formidable candidate to serve on the U.S. Supreme Court, she figures out how to game the ratings of her law school, in ways that become increasingly hilarious as you read through the chapters.

Coincidentally, I came to your blog after reading the article about your Roman subway map in the email that the College just sent to UC alumni. I got my SB in chemistry from UC in 1963, and then an MS in chemistry from Stanford in 1964. So your posting on UC vs. Stanford intrigued me.

After that, I went to law school at Columbia and there was in the same class year as (later professor) Paul Goldstein, who’s an expert in copyright law as well as an accomplished novelist.

Separately, I recently was introduced to one of the platforms for generating visualizations of data, namely Tableau. I wonder whether you used Tableau for any of your work and would kindly share any comments about it.

Reply
Sasha
27 August, 2017 at 7:46 pm

Hi Gerry, thanks for the comment. Sounds like a great book. Do you think it’s possible for me to see the email somehow? My email address is trub@uchicago.edu — I’ve been noticing quite a bit of traffic coming from a uchicago.edu domain, but I’ve been unable to see the article itself!

Regarding Tableau, to be honest, I’ve never used it and I don’t know too much about it. My understanding is that it’s fairly powerful, but it’s geared mostly towards businesspeople without data science/info design experience. As a data tinkerer myself, I prefer having full control over every aspect of the visualization, which is why I typically do everything in python and then tweak the vis in Adobe Illustrator.

Reply

Leave a Reply

Your email address will not be published. Required fields are marked *