Question: Transitioning into another research field

[StatGuy2000]

Hi everyone. This question is directed at those who either have completed their PhD in physics or is well on their way to completing their PhD.

Given what I have read and heard about the lack of opportunities in physics in the US (both in academia and in independent research labs), I was wondering if any of you have either transitioned (or is seriously considering transitioning) to another research field, such as engineering, computer science, statistics, or operations research.

After all, I already know of a number of physics PhD graduates (including a number of those who post regularly here at Physics Forums) who are currently working outside of the physics field in finance or insurance, so pursuing another research field (whether in academia or in a similar research setting) may not be so difficult to fathom.

Also, when I was a graduate student in statistics, I've known of a number of faculty members and fellow grad students who had started their careers in physics or otherwise had a physics background. And historically, a number of prominent statisticians in history started out in physics (W. Edwards Deming, Walter Shewhart, P.C. Mahalanobis, Karl Pearson, among others).

 

[StatGuy2000]

Well, the thread I started had 317 views, but not a single reply! Curious, as I assume that there are people with physics PhDs who have transitioned to research outside of physics, and would like to hear more about that experience.

 

[chiro]

Well, the thread I started had 317 views, but not a single reply! Curious, as I assume that there are people with physics PhDs who have transitioned to research outside of physics, and would like to hear more about that experience.

Maybe it's a selection and audience bias.

I don't know sample wise, if it's going to net you many examples in this kind of thing: finishing a PhD in physics is a feat in itself, but then doing a 180 and becoming a statistician is another thing.

Say it takes 10 years to get a PhD from the start of undergraduate entry. This is a huge investment to then go to statistics: my guess is most people will go the path of 'lesser' (not necessarily least though) resistance than do a PhD for physics and fall into another field. Your examples do point out that there are cases, but again the sample size is rather small.

Having said that, I do think that people that do this kind of thing have the potential to bring a whole lot to a new field based on the fact that they bring with them, a whole set of thinking paradigms, ideas, and approaches that would probably be very foreign to people only in the target field and this is always, in the right circumstances, going to be beneficial for the target field itself.

 

[StatGuy2000]

Maybe it's a selection and audience bias.

I don't know sample wise, if it's going to net you many examples in this kind of thing: finishing a PhD in physics is a feat in itself, but then doing a 180 and becoming a statistician is another thing.

Say it takes 10 years to get a PhD from the start of undergraduate entry. This is a huge investment to then go to statistics: my guess is most people will go the path of 'lesser' (not necessarily least though) resistance than do a PhD for physics and fall into another field. Your examples do point out that there are cases, but again the sample size is rather small.

Having said that, I do think that people that do this kind of thing have the potential to bring a whole lot to a new field based on the fact that they bring with them, a whole set of thinking paradigms, ideas, and approaches that would probably be very foreign to people only in the target field and this is always, in the right circumstances, going to be beneficial for the target field itself.

I concede that I am dealing with a small sample size and selection bias here. The reason I bring up this thread is that there are many people who have already devoted 8-10 years of study to earn a PhD (starting from undergraduate study) in physics, but then are unable to find secure an academic job in that field. Quite a number of these people (including several who have posted here in Physics Forums) have found careers in finance, insurance, consulting or programming, using the skills they have gained during their physics PhDs to be gainfully employed in (these admittedly unrelated) fields.

Now if these people who have earned PhDs are capable of working in the above business settings, then it is not inconceivable that they could also transition to other research fields, opening up careers in academia outside of physics, such as statistics (as an example). And indeed, these people could provide fresh perspectives in their respective new fields (as just one example, Jerry Friedman, a world-renowned statistics professor at Stanford University whose research specialty is in machine learning, originally started out his academic career in particle physics).

 

[gbeagle]

I looked into this briefly when thinking about what to do after my PhD, but it seemed to me that I really had no chance of doing this successfully. I did experimental high energy physics for my PhD. The only really concrete transferable skills I gained out of it (in my own estimation) were programming and some minor skills with statistics.

Programming skills are not synonymous to computer science research skills at all. Two of my very good friends are CS PhDs, one of which will be an assistant professor starting in the fall, and based on looking at their work there is no way my current skills would allow me to be a productive researcher in CS within any reasonable time frame. I'd be starting from scratch. Their assessment is basically the same. My skills could get me a job programming, which they did, but not doing CS research.

In terms of my "skills" at statistics it was more or less the same. I was able to take a number of observed events and estimated expected number of events with associated errors, and plug these numbers into the standard stats packages used in my field. I use this to extract limits on the parameters of various flavors of supersymmetry. Not what I would call skills... Observing the other graduate students on my experiment, this seem to be a pretty typical level of knowledge. It makes sense to some degree. Statistics are just a tool being used they aren't the end in themselves, so no more effort is exerted than the minimum necessary*. There definitely are HEP people who really grok statistics, and they are the ones that drive the standard techniques and packages that the rest use. Those individuals could probably transition to research in statistics, I would guess. People that good though are usually the ones that don't have trouble finding jobs to stay in the field.

*Note: this same tendency seems to extend to knowledge of quantum field theory as well... There is a large fraction of HEP-ex PhD students who have little more than the most rudimentary understanding of QFT.

One other aspect is that was not clear to me was how to even go about making the transition. I knew how to find post-doc openings in my own field, but not in others. Also even if I was able to find positions to apply for how could one make an argument that you are a good fit for a position that you have never done research in? In the end I think a lot of the problem is that there is just poor overlap of the core skills from high energy experiment and other fields. It is probably not as bad in other areas of physics where there is more overlap with other fields.