bayessays

The DC area does have some respectable programs like George Mason and GWU, but yeah, with your profile I think the smarter move would be to just go to a program that has research that interests you and is solidly ranked.

The federal hiring process is so different than academia/industry (and also location-limited) that I'm not sure any department has a huge number of people go into government positions, but programs in the DC area are probably good to find those connections. University of Michigan has the survey research center, so there are opportunities there for biostat/stat PhD students to do related research. CMU has the stats+public policy concentration. But agreed with above, that your research area will matter somewhat. For instance, I know some people who do statistics+privacy research, and that is an in-demand skill for the Census Bureau now.

Even if the topics are similar to your undergraduate class, I'm sure Wharton's PhD mathematical statistics class will cover them at a much deeper level. That will be the most useful for your future (will make the PhD classes+exams much easier for you and help your application by showing you can succeed in these classes). The machine learning and analysis classes are also reasonable to take if you really do not want to take the mathematical statistics class. I don't see Harmonic Analysis being as useful.

I think even 40-60 might be pretty big reaches with these math grades from a non-elite school and no research experience.

Other people may have different opinions on this, but I would also be worried about it being judged negatively, and the people on admissions committees are just humans like us, so I would leave it off. It would be another story if you didn't have a long history of improvement. But lots of people screw up at the beginning of college. Do you have some research experience and some close relationships with professors to write you letters? These will also help. I had Bs in calculus and a B- in linear algebra first year of college and it didn't present huge obstacles for me in stats PhD admissions.

No, grad school will care significantly less about this than a job would. I took a few years off and took acting classes - nobody cared or asked.

Do you think you can get your GRE Q above 160? I feel like that's really the only flaw in your profile.

As @Stat Assistant Professor alluded to, the biggest issue is that these schools aren't really much less competitive than Harvard/Chicago/Duke. I'm just a sample of 1, but I was waitlisted at Chicago twice and rejected multiple times from Columbia, NCSU, Purdue, and Duke. Schools like UIUC/UF/OSU/TAMU are definitely less competitive but very high-quality programs and I think that's a good range to focus on, with a few below and a few above.

Great advice here. Sometimes I realize I'm not specific enough in what I'm saying. I don't think UF/UIUC are "safe" schools, and certainly don't think UW/NCSU are close to safeties for OP. But I think some schools like FSU/Colorado State/Rice are pretty good bets, and OP should apply to more schools in that range as safer options, rather than going much further down in the rankings. There's a pretty big drop-off in program quality after 50 (besides a few standouts like SC) in my opinion, so I think OP would be better served by sending out more applications within the top 50 than applying to programs outside the top 50. I think the wildcard here is the research, which seems like it could be more substantial "real statistics" research that a lot of applicants don't get. If this research+letter of rec is good, that could stand out even to top 10-15 programs. But yes, apply as widely as your budget allows.

Stanford says in their FAQ you can submit an extra letter if you want. I think this may be one of those rare cases where it makes sense to have both. I think the Stanford connection can only help you - nothing says "this person can handle the program" like someone who has been through it themselves.

I'd be somewhat surprised if you got rejected from UF/UIUC, and very surprised if you got rejected from FSU. I think you could reasonably apply to any program besides Stanford, and I don't think you need to apply to any programs outside the top 50.

People leave their PhD programs with an MS all the time. I would be utterly shocked if your program did not allow this. Talk to your advisor/the grad program chair and figure out how you can stay and leave with an MS. Your first year of courses is probably the same, so even in the worst case that they cut off your assistantship, you'd still save money compared to starting over somewhere else. You can re-apply and I'm sure you'll have a lot of success, but really ask if that's the best path for you. I'd do anything I can to try to make it work at your current program to leave with an MS, which will save you lots of time and money. And you go to a top program, so it doesn't make sense to transfer down.

You went to an Ivy League school and have one bad math grade, and a great GRE score. I don't think those reaches are big reaches. Drexel and Temple are more than safe enough. I would recommend just adding more schools like Brown, Minnesota, Penn.

There is not a ton of info available on admissions to these lower-ranked programs, but I think you would get into some of those on your list. I'm not as optimistic as @StatsG0d and I think Iowa/Pitt/FSU are already "reach" schools for you and you shouldn't apply much higher than that. For an international applicant especially, a 3.4 graduate GPA is very low, so I think you are right to focus on more applied programs. Since you have 5 years of experience working in pharma, I don't think anyone will question that you want to go to a more applied program.

Hey, this is mostly a statistics forum, so I will forewarn you that this is a *very* uninformed opinion. But some people here over the years have applied to both statistics and operations research programs, so I've seen a little bit of these admissions results. I'm guessing when you say you go to a liberal arts college that it's a pretty good one? Top 20 or top 50 at least? Regardless, your profile looks very strong both in terms of coursework and research and I'd recommend you apply to any of those programs that interest you and I suspect you'll have good results. It seems to me, from an outsider perspective, that once you get past MIT/Princeton, the level of competition drops dramatically and these programs tend to not be as competitive as statistics PhDs, where I would recommend you apply to top programs. I have seen some applicants here who didn't get into top 40 statistics programs but got into OR programs like Berkeley, so I imagine you will have a lot of good choices.

@Bayeasy do you mean 80-90 in Asian rankings, or 80-90 in the world rankings? Also, did you mix up your GRE scores? I'm guessing the 168 is your math/quantitative score?

What are the published papers like? Did you do some data analysis for a professor, or did you do methodological research and submit a paper to JASA? Huge difference. Assuming the research experience you have is average/typical of an undergrad, you'll be competitive for top 20 schools, but I don't think it's safe to only apply that high. Add some schools in the 20-50 range (even schools ranked #50 include UC-Irvine and UT-Austin, which I think are as good as some top 25 programs). I think you should definitely apply to any biostatistics program that interests you. For stats, my guess is that schools in the Wisconsin/NCSU/Minnesota/TAMU range are good "targets", with schools like Michigan/Washington/CMU/Duke as reasonable-but-not-guaranteed, and then schools like UIUC/OSU/Colorado State on the lower end of where you should be targeting. The more money you have to spend on apps, the more you can send to higher ranked places -- I don't think anywhere besides Stanford would be a waste of money with your profile.

Not sure about economics, but for statistics, you can overcome this, but your grades need to improve. Realistically, you want to get your GPA above 3.0 to meet the minimum requirements for most graduate schools. So if you could get a 3.5 or above for your last two years, you could probably get into a lower-ranked master's program (lots of people screw up early in college). You will want to take some more math and stats classes to make up for the low grades in early ones, too. I know people who got an online/mostly-online MS in data analytics and got some good jobs afterwards. These programs are all getting very competitive though, even non-prestigious ones, so you need to get the grades up.

Most beginning PhD students do not have this, so don't worry. Read the first couple paragraphs on Wikipedia. There is no reason for you to need to understand open problems in areas you will not be studying.

You're reading the Michigan website incorrectly, they don't even require a single probability class and have one of the lightest course requirements now.

UCSB sort of sticks out as a school that's not in the same tier as the others and probably wouldn't serve your goals to be an R1 professor very well. Is NYU data science because of Jennifer Hill? I think the interdisciplinary programs are a little riskier than getting just the stats/biostats PhD but some of those programs are pretty cool too, so it's a tough choice there. What about other programs that are ranked similarly to CMU, like Duke and Michigan? I think there is some interesting social science research done in both those places.

I don't think there's too much risk in the next decade of you not being able to find a job in DS. Also, if you and your wife are both data scientists at big tech companies, you could be millionaires in 5 years and have the freedom to do whatever you want, really. The one thing I would not underestimate is the prestige and opportunities that having a job at one of these companies will open for you. I have friends with PhDs in biostats who struggled to find data scientist opportunities, and I could barely get interviews before the FAANG job, but once you get in, getting any other interview is absurdly easy. Also, jobs that involve data are pretty common in a variety of industries, so it would be pretty easy to pivot to a different type of role in a totally different industry. Switching to pharma after a decade might be harder, but I don't think you'd have problems if you decide you didn't like tech in 2-3 years and move to pharma -- this isn't much different than people who get an academic postdoc and then decide to get pharma jobs, which I've seen. If you're undecided, I personally feel like the potential benefits of accepting the tech job outweigh the risks, which is just being a couple years behind in the pharma world. Heck, pharma companies like Eli Lilly hire data scientists too so you could always switch into something like that.

I think not only is the career path more clear/linear for pharma, but the job itself is more clear. I've never worked in pharma, but because of the regulatory aspects and the goals of clinical trials, I doubt there is much ambiguity. Being a data scientist requires a certain level of comfort with ambiguity as there is often a lot of freedom in the job and it's not totally clear what you should be doing. Some people might love this, and it might drive some people crazy. I was more on the latter side, so I totally understand where you're coming from in your assessment of the pharma advantages. If you have an offer at a FAANG-type company, I would heavily consider it though. You're not going to make that type of money in pharma. Maybe the starting salaries are similar, but once you start acquiring more stock after a few years, I don't think the compensation will be comparable - you can save close to 6 figures a year at these jobs when you're starting out. Work a couple years at one of the companies with a promotion or two, and you will be set for a manager position somewhere. After one year at a FAANG company, I was getting final-stage interviews to be heads of data science at known companies. I see people on LinkedIn all the time who get jobs at average companies, get a promotion or two over the course of 2-3 years, and then are a manager and can then get manager jobs anywhere. I think there's a lot of opportunity to quickly move up to an extremely well-paying career. If you're unsure about which you'd like to do, another advantage of tech is that the options are pretty broad and job-hopping is common. You can try out a new job every year or so until you find what you like. I'm not sure if pharma is like this.

Yes, that score will be sufficient.

You will probably get into quite a few top 10 programs anyways, but getting that GRE Q up a couple points would help. I would say apply to all the top schools where you find interesting people except for UNC/Washington, which are probably more theoretical than you would prefer. The best way to get familiar with programs is to look at the faculty web pages (and if available, student pages) and look at the types of papers they are writing and seeing if it interests you. Look at the US News statistics rankings and look at the webpages for all biostat programs in the top 50. For biostatistics, generally speaking, the programs get more applied as you go down in the rankings. However, I think you would be doing yourself a disservice by going to an extremely low-ranked program. There are some lower-ranked programs that are explicitly epidemiology and biostatistics departments (like UIC), though often the PhD degrees are still separate. But again, I think you can find stuff that interests you in top biostat programs. I know someone who got into a top PhD in epidemiology program without an MPH, but I'm not sure how rare that is. You'll have a lot broader range of options with the biostatistics PhD and a higher salary, in all likelihood. I would look at UPenn's biostat PhD. They are a top 10 program, have some causal inference people, and they do lab rotations their first year or 2, so you get a lot of applied experience right away (and some of these are epidemiology-related). A lower-ranked program I might look at is Vanderbilt, which also has a lot of applied opportunities through a similar program.