Help me make a decision

[Casorati]

I am fortunate to have been admitted to the Statistics PhD program at UBC, Waterloo and Biostatistics PhD program at McGill, UNC (unfunded) and Florida and I am having a hard time making a decision, so I would like to seek out advice from you guys. It seems that the academic placements at UBC and UNC are very good, and McGill is a better research fit as they have a group of people doing cutting-edge research in causal inference. Florida is relatively new but there are a few faculties I would like to work with. Waterloo has a large statistics department with a focus on applied side. Obviously UNC has the best reputation but I don't think it sensible to pay for the tuition and living cost for 5 years.

[Casorati]

Seems that many people vote for UNC. Is it worth it to turn down a funded offer and accept UNC, which does not guarantee funding?

[Spaghettini Plot]

I don't think it is worth it to pick UNC just because of the prestige if you are not going to be funded.  If you are dead set on Causal Inference then you'll plenty of options at McGill which is also a great school. I think the poll could be misleading if people answer it before knowing the options that you have to decide between

[Spaghettini Plot]

Personally, if you are dead set on Causal Inference I think McGill would be the best fit. Otherwise I think UBC has plenty of research areas to pick from and I think that would also be an excellent choice

[dopamine_machine]

Honestly, if UNC is unfunded, I definitely would choose one of the other programs, as the difference between UNC and the other programs is likely negligible.  I would pick the place that is best for your research goals (aka in line with what project you want to work on) and I know that McGill and UBC are both pretty great for worldwide reputation (I'm from SoCal and I really like those two schools).  As per rankings, It does appear that Waterloo is the second best of your choices, which is something to consider (though, who will really care about rankings when you're in year 4 of your PhD).

Hope this helps

[Casorati]

Although I did causal inference research during my master's, I'm also open to other research areas. I have found areas such as functional data analysis and spatial data interesting. McGill's Biostatistics department seems to be pretty small and focused and I prefer larger departments with more research areas.

[dopamine_machine]

Also, I think I wouldn't pay attention to the rankings above that people voted.  I know mine was committed in error before reading your situation and I find myself unable to change my answer with your choices in scope.  I corroborate a lot of the information that others have said on this topic, so I think focus more on the comments 

[bayessays]

Most people probably didn't read your post before voting. Nobody in their right mind would choose an unfunded offer from UNC, so you shouldn't even be thinking about it.

[likewater]

If you're (potentially) interested in causal I don't think UNC seems like a great fit anyway? Off the top of my head only Michael Hudgens really does causal work. Somebody who is more familiar with the actual research done by people at UNC biostat might be able to comment more on this though (my knowledge is relatively limited to what people put under "faculty research interests"). But yeah, unless you are quite wealthy and have money to burn I would advise against taking the unfunded offer especially when you have good alternatives.

Edited April 24, 2020 by likewater

[ZNtheory]

I looked through the website of Department of Statistics in UBC and found that some very strong guys have joined who may focus more on machine learning. I think you are probably right to get into a department with people whose interests are diverse if you are not so stick to Casual inference.

[Casorati]

Thank you for everyone's response. I have no prior experience in machine learning but I am open to research areas I've not yet considered. It appears that ARWU places Waterloo ahead of UBC and McGill while QS statistics ranking places UBC and McGill ahead of Waterloo. Which ranking is a better proxy of the academic reputation?

[Spaghettini Plot]

On 4/24/2020 at 8:21 PM, Casorati said:

Thank you for everyone's response. I have no prior experience in machine learning but I am open to research areas I've not yet considered. It appears that ARWU places Waterloo ahead of UBC and McGill while QS statistics ranking places UBC and McGill ahead of Waterloo. Which ranking is a better proxy of the academic reputation?

Probably hard to say in general. A lot of these rankings seem to be pretty heuristic.  US news just seems to capture it better than most in America but we don't have anything like that in Canada.  In my opinion, based on general reputation and faculty strength etc, I would probably list the top stats programs in Canada as: 

1. U of T

2. UBC 

3. McGill

4. Waterloo 

Personally I think UBC is an excellent choice and very strong overall. However particularly in causal inference, I think that McGill is probably the strongest in Canada, but that's one sub-field. If you decide in your PhD that you would rather work on something like spatial stats or ML, then I think you would have a more options at UBC than McGill

 

[SPIWizard]

I just wanted to say that ARWU is an explicit formula based on publications (although 'top' journals are determined by survey) so I think Waterloo's size plays a factor there. Slightly biased because I just picked UBC over McGill and Waterloo for a masters, but I vote UBC for reasons already mentioned here unless you have a particular advisor in mind at McGill.

[bayessays]

I think you should look at the specific faculty.  I'm not extremely familiar with the anglo-Canadian departments, but UBC and Waterloo stats seem a step above McGill biostat in general, although I saw some McGill people who were pretty solid too (but obviously it's a much more applied department).