Statistics Research Areas + Why Statistics

[statistic]

Please forgive me if this post does not fit the focus of this forum - I will move it if there is a better place to ask this. I know this is a long post, but I would be extremely grateful if someone were to read 

 

I am an incoming statistics master's student in a top department. I am a long time lurker and I would frequent this forum during the 2021 application season. I would often read profile evaluation posts and post-admission decision posts in which many folks would discuss their research interests and the institutions' strengths/weaknesses in those areas. As someone who is interested in continuing onto a PhD (perhaps in statistics or an adjacent field) I would like to pursue research during my master's program. However, I am ashamed and feel that I would be severely naive in approaching the faculty to work with them because the department is very highly regarded and has extremely influential statisticians, and I do not even have an elementary understanding of the various research areas in statistics (let alone their subfields and certainly not the open problems within them).

I have a few questions that I hope someone may help with.

1. How did you all gain a working understanding of these research areas, e.g. causal inference, high-dimensional statistics, Bayesian statistics, etc, without already having a solid foundation (graduate-level coursework) in stats?
2. Would someone be able to explain these various areas? Also what resources I can use to get up to speed? Particularly, are there any references that describe these at a high enough level to understand them broadly, but also at a deep enough level to allow me to understand the formulation of open research problems in these areas? Is this even possible without already having a strong graduate-level foundation in stats?
3. Why do/did you all want to study statistics? (I'd love to hear anonymized personal stories of your journeys and how they led to studying statistics!)

I would greatly appreciate all your insight. Thanks!

[Stat Assistant Professor]

If you are in a Masters program and you don't have any prior foundation, then it will probably be tough to do very deep statistics research while taking three classes a semester. It can definitely take awhile to become accustomed to academic jargon, and it will be difficult to juggle trying to do that that while still taking courses. I personally found that even after having taken the graduate-level statistics courses, it was initially very difficult to read/understand statistics papers. It took considerable effort and patience to be able to read academic papers, even with almost full-time effort. But eventually, if you work at it enough, it just sinks in. Your PhD advisor will definitely help you out. As for developing a 'working' understanding of different research areas, it would definitely be helpful to take some Statistics electives like Bayesian statistics, spatial statistics, or high-dimensional data analysis (as these are often useful introductions to specific areas), and to attend departmental seminars. You probably won't understand everything in department research seminars, but you can hopefully pick up bits and pieces here and there and get a flavor about different research areas.  

If you wanted to do some 'light' research, you could get involved in some interdisciplinary research or applied research where you write some code and/or perform some statistical analysis. That would be seen as a definite plus for PhD applications, though it wouldn't carry as much weight as letters of recommendation and grades. For PhD admissions, I think it is more important to have excellent grades and strong letters of recommendation which can attest to your research potential than it is to have "pure" statistics research experience. I would probably focus more on doing well in your classes and getting good letters above all else.

Why study statistics? I suppose the motivation is different for different people. Some students want to get into certain fields (e.g. data science, quantitative finance, etc.), so they pursue advanced degrees in Statistics to help them achieve their goals in industry. Others are motivated by wanting to study specific applications (e.g. environment, demography, public health, etc.), while others are primarily interested in the mathematical foundations of probability/statistics. For me, I personally enjoyed the statistics classes that I took during my Masters program (in Applied Math), especially a class I that I took on Bayesian statistics. After my Masters, I took an industry job but found that I preferred the academic environment, so I pursued a PhD two years later. Over the years, my motivations for continuing to study statistics have also changed a bit. I started out as more of a mathematical statistician during my PhD program (i.e. studying statistics theory rather than applications). But after doing a postdoc, I became a more applied statistician who still does some theory, but it's not always the primary focus of my work. I still enjoy the mathematical challenges of statistics theory, but I find that to stay motivated, it is often useful to have motivating applications for your work -- that is, being motivated by wanting to address real-world scientific problems and wanting to make sense of real data sets (which are often messy and challenging to make sense of!). 

[bayessays]

9 hours ago, statistic said:

I do not even have an elementary understanding of the various research areas in statistics (let alone their subfields and certainly not the open problems within them).

Most beginning PhD students do not have this, so don't worry.

 

9 hours ago, statistic said:

How did you all gain a working understanding of these research areas, e.g. causal inference, high-dimensional statistics, Bayesian statistics, etc, without already having a solid foundation (graduate-level coursework) in stats?

Read the first couple paragraphs on Wikipedia.

 

9 hours ago, statistic said:

Is this even possible without already having a strong graduate-level foundation in stats?

There is no reason for you to need to understand open problems in areas you will not be studying.