Choosing Stats PhD Program: Harvard, Stanford, Columbia, MIT EECS, MIT ECON

[spectral]

My research interests are:

• Causal inference and machine learning
• Robust (non-parametric) inference
• Learning under distributional shift
• Learning in strategic settings
• Causal inference in competitive settings

In terms of pure fit I _think_ Harvard (especially the EconCS lab) and MIT Econ are the best for the last 2, and Stanford seems the best for the first 3.

I am also interested in doing applied work in the social sciences.

Any advice would be super helpful!

[bayessays]

If you want to be a statistician, it's hard to go wrong with Stanford.  If you want to be an economist or a computer scientist, the MIT programs would of course be good options too.  But if you can be guaranteed to work with one of the big professors at Harvard/Columbia, they have great people too.  All great choices, congrats! 

[StatsG0d]

All really great programs, and a pretty difficult choice!

Stanford might be a good option because Wager (stats) and Athey (economics) collaborate often. Both have worked at the intersection of causal / ML and Wager has done some interesting stuff in policy learning under constraints.

Don't know much about MIT's econ program other than it's the best one. Overall, I feel like if you're more interested in probability / statistics stats departments are better. If you really, really like causal inference and linear models, then economics might be better (and this might be an oversimplification, but hopefully a useful one).

One thing to consider is if you're really interested in statistics / ML, do you really want to take economic theory courses, or would you rather take stats courses?

[statsguy]

It's hard to go wrong with Stanford. Assuming you actually got in to Stanford... go to Stanford. 

[StatsG0d]

10 hours ago, statsguy said:

It's hard to go wrong with Stanford. Assuming you actually got in to Stanford... go to Stanford. 

Interesting. I don't think it's that clear. MIT for econ is like what Stanford is for stats, and typically econ programs are more competitive than stats / biostats (although for these programs the competition is probably similar).

[spectral]

fwiw: the admit site shows MIT Econ admitted 22/700 applicants this year, stanford has never been that competitive.

this is not really something i'm deciding on the basis of right now. i feel like in econ itll be hard to pursue a bunch of my interests and ill have to write almost exclusively 90 page econometrika submissions. 

i guess my main q is harvard where between kosuke imai and yilling chen there's very good coverage of my interests or stanford where theres Wager but not many other folks in my areas

[bayessays]

Do you have a preference for jobs after?  I feel like being a professor in an economics department is a very different career path than being a professor in a statistics department.  You mention Yiling Chen, but she is not a statistician and doesn't write any papers in statistics journals - would you even be able to work with her as a student in their stats department?

In addition to Kosuke Imai, I'd also look at Susan Murphy at Harvard who seems to overlap with a few of your interests.

[spectral]

I would be able to work with Chen, people are co-advised in CS and Stats reasonably often. And I think in terms of jobs I am interested in being faculty in either CS or Stats, ideally with an affiliation in some social science (polisci, sociology, econ, tho the first two seem more likely!)

[DanielWarlock]

46 minutes ago, spectral said:

I would be able to work with Chen, people are co-advised in CS and Stats reasonably often. And I think in terms of jobs I am interested in being faculty in either CS or Stats, ideally with an affiliation in some social science (polisci, sociology, econ, tho the first two seem more likely!)

Given your interest, I think Harvard is best fit. Imai has affiliation at Kennedy school, Neil Shephard is affiliated at economics. Murphy is also a big name here doing causal inference and reinforcement learning affiliated to CS department. There is no problem that you seek additional advisors at MIT or other Harvard departments. You can easily find someone at MIT to supplement for (3) (4). Everyone is saying Stanford stats but they are mainly about highly mathematical/theoretical high-dimensional stats and probability theory. So I guess you will need to go to their CS department to find advisors? Stanford probably is not that of a good fit for you. 

[bayessays]

It definitely sounds like Harvard and Stanford stats would be best for those career options.  I know MIT econ has some econometricians and they publish in journals like JASA, but it seems like a less direct route of getting to where you want.  I'm not sure if Columbia has anyone besides Blei who you're interested in.

There's really no one-size-fits-all answer here.  Stanford is always a safe choice.  But look at where Kosuke Imai's students have gotten jobs, Murphy has had successful students that overlap with a few of your interests, they have other good profs too... to me it sounds like you want to go to Harvard.

[trynagetby]

Your tier of choices are way above my pay grade, but I'll throw in my two cents that it seems that Stanford doesn't seem to be the best choice. I'd recommend taking a look at their stats phd dissertations to see if you're interested in that type of work. Stanford posts dissertations publicly.

Edited April 6, 2021 by trynagetby

[StatsG0d]

IMO if you're interested in potentially a joint appointment with a social science, economics is probably the better way to go.