Working between disciplines is exciting, and I wouldn’t want to have it any other way. Psychology has become much more interdisciplinary in recent years, and there was a discussion today on social media whether psychologists should ‘know math’. This reminds me of the statement Plato had famously engraved at the door of his Academy in Athens: “Let no one ignorant of geometry enter”1.
Few people have made this pointed more clearly than Meehl (1990): “I think that PhDs in psychology should be required to learn a little undergraduate mathematics different from cookbook statistics. Inability to think mathematically among psychologists except in certain special areas is sometimes so gross as to be embarrassing to one familiar with the quantitative sophistication in other sciences.”
In principle, knowing more is always better, and knowing math is likely much more insightful than knowing many other things. I like the idea of teaching it to psychologists, and we have a preprint arguing that lack of formalized theories in psychology may posit fundamental barriers to scientific progress. Math helps.
On the other hand, the idea that psychologists should know math made me want to compile a list of the things that people commonly expect me — a 35 year old postdoc working between clinical psych and methodology — to know and do. Here we go:
- Good understanding of basic statistics
- Good understanding of more elaborate models: basic SEM such as EFA, CFA, or ESEM; fancier SEM such as latent change score models or latent class analysis; basic IRT
- Network models, frequentist and Bayesian, for cross sectional and time-series data; extensions such as continuous time-series models
- At least a basic idea of Bayesian statistics
- Pick up cool new methodologies such as machine learning & neural nets quickly
- Know R well; use Mplus for things R cannot do (e.g. DSEM in Mplus8); use some other programs such as Onyx to visualize models, know JASP well enough to teach students
- Ability to write at least basic R-packages
- Write papers in LaTeX, and write R code in Markdown
- Substantive knowledge in my fields (measurement and modeling of mental illness, and substantive research on mental illness such as RCTs; easily 500 papers per month in decent journals). It feels like this should be more than one point, because it takes up a huge chunk of time…
- Have a least a minimal background in philosophy of science
- Keep up with stats developments (e.g. regsem package)
- Know or at least learn one proper programming language (e.g. Python)
- Run a blog to get visibility and increase chances to get tenure
- Embrace & support open science practices, prepare all your syntax so it can be uploaded, upload your data when possible, write high quality reviews you feel comfortable signing, publish your reviews on publons, etc.
- Engage in debates on social media
- Market your research on researchgate, academia, psyarxiv, etc.
- Oh yeah, write tons of papers. Collaborate. Set up cool projects. Plan ahead. Be creative, have great ideas and follow them up. Science!
- Be editor for 1 or 2 journals to increase chances for tenure
- Review several papers a month (in most months, I receive at least 10 invitations; fewer in November, more in January)
- Right, teaching …
- Supervise students
- Write a few major grant applications per year
- Apply for positions
- Be involved in a few large international collaborative grant applications
- Travel a lot, organize and give talks at conferences
- Move every 2 years to another country (for me it’s been 6 countries since 2005); learn the local language to have chances of getting tenure
- Stand out as calm and collected member of the university, be helpful, unstressed, have an open ear for problems of students and co-workers ;)
And now we add knowing math. I’m wondering if we’re starting to ask a bit much of psychologists. Again, I personally think math is more important than a lot of the things I list above, and I would have loved to learn more math in my university curriculum, or found time during my Postdoc to get a better grasp of math. And maybe2, just maybe3, I should have studied math at the cost of other points in the list. As Meehl (1990) wrote, “the common rationalization of mathematically ignorant psychologists (“Well, I understand the logic of factor analysis even though I don’t understand the math”) should not be tolerated in intellectually polite circles!”. I’m definitely one of these rationalizers.
But these days, I only sleep every second Wednesday, and if I have to give that up for learning matrix algebra, I’m not sure I could pretend to keep things together much longer.
Rogier Kievit just shared a new paper on Twitter entitled “What faculty hiring committees want”, with a graph arguably supporting my question whether we’re sometimes asking too much …
Beautiful visualization by Susan Wardell:
Did I miss it or did you forget the most important one “designing and carrying out reproducible experiments”? :-)
Pingback: A summary of my academic year 2017 – Eiko Fried
I don’t know if I’m relieved because you pointed out these requirements are ridiculous, or stressed out because I hadn’t thought of several of them..
That’s actually quite a useful list for orienting myself. In your opinion, are those competencies that everyone who starts a postdoc position should have? Because if yes, I really need to learn Mplus…
No, it’s absolutely unreasonable to expect these from any single human ;) … personally, I moved away from Mplus for numerous reasons in the last years, but might return now due to the DSEM innovations in Mplus 8. It’s really neat and does things that R really can’t do yet.