He does mathematical modeling and is asking for career advice: wants to move from biology to social science

Rick Desper writes:

I face some tough career choices.

I have a background in mathematical modeling (got my Ph.D. in math from Rutgers back in the late ’90s) and spent several years working in the field of bioinformatics/computational biology (its name varies from place to place). I’ve worked on problems in modeling cancer progression and also on (mathematically similar) problems in phylogeny estimation. But after several years of doing that, I decided that I don’t really have the passion for biology to spend the rest of my life as a full-time researcher there.

Basically I found myself more and more interested in issues of public policy, economics, and the like. But most of my contacts from computational biology have little idea about what’s going on in the social sciences.

I was wondering if you had any advice regarding resources or people to talk to regarding a potential mid-career switch. I’m currently in the DC area – I moved here several years ago for a job at the NIH, though currently I have a different job for a different HHS agency. I would prefer to stay local, but that’s not an absolute – it’s just that I figure somebody in this city must be doing this kind of work.

I’m not sure what to say here. I do think there’s a recognized need for mathematical modelers in social sciences. Yes, there are a lot of economists and even some political scientists and sociologists floating around, but each of these groups tends to have its own data-analytic tools they focus on, so I’m pretty sure that someone with experience in mathematical modeling in biology could have something useful to add.

My guess is that Rick’s way into the policy world will be by way of his existing biology expertise. There must be groups and organizations working on biology-related policy, this could be a pharmaceutical company or a government regulatory agency, for example. Health care policy is huge, and I’m guessing it’s dominated by economists, but there could be more technical areas, for example allocating resources within cancer research or performing cost-benefit analysis of some biology projects, where a model of the biology itself could be relevant. Then, once you get involved in policy modeling in a specific context, maybe you can move on.

But maybe I’m missing some other ideas. Do you readers have any suggestions?

30 thoughts on “He does mathematical modeling and is asking for career advice: wants to move from biology to social science

  1. Areas like Natural Resource Economics are often heavy on bio-mathematics and might be a natural fit. A lot of co-operation between economists and natural scientiests here as well – unusual for economists.

  2. Some thoughts:

    * Why leave such a cool field?

    * JHU has a pretty good masters program in DC.

    * Sargent and Stachurski have a good collection of macroeconomic modeling lectures online, together with Python and Julia code. Macro and lots of finance are now basically applied stochastic processes.
    http://quant-econ.net/

    * Policy is harder than you might guess. It’s hard enough to figure out what’s going on, much less why and how to change it.

    Email me if you have questions.

    • the field is a rubbish bin of weak quantitative research and sloppy science
      nobody can write good code
      p-values may be abused in psychology, but in bioinformatics they’ve decided it’s a good idea to make plots out of thousands of p-values
      there are thousands of papers and reviews on normalization methods for the same technology
      everybody creates their own pipeline scripts. no two pipelines produce the same result

      • I’m literally experiencing this right now. I off and on do consulting with biologists, especially my in-house biologist (my wife). She came to me with a question recently. She has a list of differentially expressed genes, and wants to know something about a subset of genes… and so I produce this graph, and I put her list on the graph, and literally NONE of the genes she has in her list from some expensive canned software that her university has an expensive license for overlap with the list that my simple graphical visualization method picked out.

        sigh

      • “Bionformatics” encompasses a wide variety of different topics, some of which are more rigorous than others, and some of which are more scientifically sound than others. Some of its topics don’t interest me in the slightest: for example, sequence alignment (although I have to confess I haven’t kept up to date with that field since shotgun sequence rendered more mathematically sophisticated approaches uninteresting).

        I would certainly disagree with the complaint that “nobody can write good code”. Certainly if one tried to do phylogeny research with poor code, the limitations would become quickly apparent.

        There probably is the issue of “computer scientists trying to deal with statistics, and doing so poorly”.

        But for me the biggest issue was “I honestly don’t care at all how the basal nodes of the Tree of Life were resolved.” Even questions like “how should the clade of apes be drawn?” are less interesting to me than questions in other walks of life.

  3. Experimental or clinical psychology or neuroscience in which most of the problems are still based around lab research would be less of a leap than political science or economics where observational studies are going to be more common. Behavioral pharmacology has many policy implications in terms of drug abuse, and that would be a comparatively small change.

  4. I spent 1994-2000 as a part-time visiting fellow at the Bureau of Justice Statistics, working on graphical and geographical methods for analyzing their data. I enjoyed my time there, and still retain connections. One of their deputy directors has a PhD in physics, and I found that my own background (PhD in EE) was useful in terms of looking at new ways of dealing with their data, of which they have plenty. I don’t know if they have any positions open at the moment.

  5. Perhaps a modeling unit within a government department or a policy consultancy? I worked with a couple of hard-science (physics, math) guys who entered the tax policy scene on the back of their modeling skills.

  6. This may be a bit far afield, but just in case, I am actually an economic modeler in the tax division of the Congressional Budget Office. It is a wonderful and stimulating place to work, and I happen to know of a couple modeling vacancies. The first is actually my spot. I am actually heading to JPMC, induced by access to their ridiculous microdata, but trust me when I say it is a struggle leaving CBO. In my group, they generally seek economists with tax experience, but you never know: https://chp.tbe.taleo.net/chp03/ats/careers/requisition.jsp?org=CONGRESSIONALBUDGET&cws=1&rid=306

    The second relates to mechanism design work in the micro studies division: https://chp.tbe.taleo.net/chp03/ats/careers/requisition.jsp?org=CONGRESSIONALBUDGET&cws=1&rid=279

    I would guess that the latter would be more in line, but I know little about your field. My work now is mostly with our microsimulation model.

    If either of these seem interesting to you, I am happy to speak further about it (though obviously I can cover operational details in tax better than micro). My email is [email protected].

  7. Find a place that does both. For example, EPA does comp tox work and economics policy. It would be easier to come in, work on the former, and then lateral over rather than to just jump in working on something different.

  8. It’d probably be helpful if Rick added a few more details about what he’s looking for. Continuing to work on math and modeling but in the context of social science? Different kinds of social science research? Working on public policy but outside of research? That said, here are a few thoughts:

    *Epidemiology might be a relatively smooth transition. They’re at the intersection of social science and medicine, have lots of government jobs, and have started to play around with agent based modeling. There’s a lot of work here but some places to get you started include:
    **PACER – Homeland Security’s Center of Excellence on Preparedness and Catastrophic Event Response which has a hub at Hopkins. http://www.pacercenter.org/about-us.aspx. They do lots of mathematical modeling.
    **MIDAS – the Models of Infectious Disease Agent Study. http://www.epimodels.org/drupal/?q=node/2. Based out of UPMC, I think, but with a federal presence.
    **ASPR and BARDA at HHS do a lot of work at the intersection of modeling, biology, and social science/policy.
    *A Masters of Public Health might also be a good way to make a transition if you want to go back to school and are interested in health policy.

  9. Rick’s LinkedIn profile mentions some experience in machine learning. There’s definitely a cottage industry of folks trying to apply ML to public policy problems and there might be a place for you to slot in there. Some links to get you started:

    *A recent panel at the annual economics conference w/ the latest on ML and urban governance. https://www.aeaweb.org/aea/2016conference/program/preliminary.php?search_string=predictive+cities&search_type=session&association&jel_class&search=Search
    *A nice Quora discussion of ML v. econometrics. Some of the stuff mentioned might provide some leads. https://www.quora.com/How-will-machine-learning-impact-economics?redirected_qid=6706789#
    *A start up trying to use Data Science for “social impact” http://www.bayesimpact.org/
    *The Data Science for Social Good Center at U Chicago. http://dssg.uchicago.edu/

  10. Water supply and irrigation districts in the West employ modelers. Advancement up to state and regional levels is possible. Water conveyance and, increasingly, water supply (desalination, recycling) cost money, and therefore have social effects. Climate change is creating significant need for analyses.

  11. If you want to work on public policy questions because you want to influence public policy, then a good way to do that might be to use your technical skills directly working as a data scientist for a political campaign. Helping your preferred side get elected might be the most effective way for you to get your preferred policies implemented.

  12. A little off-topic but I’m looking to hire someone with an applied math background – preferably someone with photogrammetry and image processing experience. Experience with machine learning and pattern recognition as applied to image analysis considered a plus.

  13. It would be worth looking at policy/modeling oriented consulting groups active in the DC area, either a boutique firm or a group within a larger firm like Booz with a modeling focus. These groups typically offer a lot of freedom in the domain you work on, once you join, contingent upon sales. I’ve seen people either specialize in particular modeling approaches across domains or have a domain focus but change between e.g. defense, pharma, energy and transport.

  14. Hmm – I’m gonna risk suggesting to take another look at computational biology. I mean just think about how the millions of sequences could change what we can do — what we can really predict, the holy grail of phenotype from genotype and being a part of that discovery. it will need rigorous non-bullshit science and reward for doing that – but thats a problem everywhere. Just a thought.

  15. I don’t think it’s too late for a midcareer switch. When I think about mathematicians who have switched over to social sciences, David Roodman comes to mind. He seems to be doing okay.

  16. Health economics and health technology assessment as somehow suggested by Andrew (“cost-benefit analysis of some biology projects”) might be very interesting to him. There is plenty of health economists from a similar background.

  17. I work in what’s basically a social services think tank – Research and Data Analysis (RDA) division of the WA Dept. of Social and Health Services (DSHS). We employee all kinds of data and statistics PhDs to do complex modeling, predictive analytics and program evaluation, e.g. see https://www.dshs.wa.gov/sesa/research-and-data-analysis

    Our partners in the federal government – such as CMS (Centers for Medicare and Medicaid Services) -have similar capabilities: see https://innovation.cms.gov/ and https://www.cms.gov/Research-Statistics-Data-and-Systems/Research-Statistics-Data-and-Systems.html.

Leave a Reply to Marvin Ward Cancel reply

Your email address will not be published. Required fields are marked *