Hey, just once, the Buzzfeed-style hype is appropriate. We have 3 amazing postdoc opportunities here, and you need to apply NOW.
Here’s the deal: we’re working on some amazing projects. You know about Stan and associated exciting projects in computational statistics. There’s the virtual database query, which is the way I like to describe our generic MRP and deep-interactions research, which we’re applying to so many different problems, ranging from voting and turnout to scores on medical board exams and surveys of at-risk births. We’re working on differential equation models in pharmacometrics and soil science. Physics and astronomy. Ornithology, for chrissake! You name it. We’re doing some of the most exciting stuff around in causal inference, with links to some amazing causal-inference researchers at Columbia and elsewhere. Treatment interactions, anyone? These models are fundamental to a modern view of applied statistics and policy analysis. Weakly informative priors. Important in themselves, also we have some ideas of how they can help resolve the replication crisis. That’s right, I’m not just whining about replication on the blog, I’m doing research on how to do better. I think I’m forgetting about 80 other projects we’re working on. Oh yeah, we’re evaluating the Millennium Village project and we’re working with some economists on general ideas for hierarchical models for multiple outcomes. That’s right, we want to tackle the Edlin factor. And penumbras. And, did I mention that we’re working on the hot hand? We’re aiming big here, we’re doing the most advanced computation, working on every possible problem you can think of, with an amazing interdisciplinary team. And we’re not too proud to get help from our friends like Jennifer Hill, Guido Imbens, Dave Blei, Havard Rue, and so on. Developing the best tools to do the best work, that’s what we’re all about. And you—that’s right, you—can be part of it. You just have to apply.
2-year postdoctoral research position on statistics and education research
This postdoc is part of an Institute for Education Sciences training grant, operated jointly by Andrew Gelman (Columbia), Jennifer Hill (NYU), and Marc Scott (NYU). We are working on a range of different research problems, including causal inference, computation for hierarchical modeling, and various applications in education research. The candidate should have strength in statistics, interest in education research, and ideally be a strong programmer as well. This work will take place in an interdisciplinary research environment, and the postdoc will have the opportunity to collaborate on multiple projects in statistics and applied research. The candidate must be a U.S. citizen or permanent resident. If you are interested, please email me your CV, a letter of application, some of your papers, and three letters of recommendation. We’ve already started collecting applications, so apply right away.
Also, our colleague Sophia Rabe-Hesketh at the UC Berkeley school of education has a 15-month postdoctoral opening to work on quantitative education research. We’re working together on Stan, so if you’re interested in contributing to Stan in a useful, fun, and exciting way, either from NYC or Berkeley, this could be a great opportunity for you.
2-year postdoctoral research position on informative priors for Bayesian inference
This postdoc, supervised by Andrew Gelman, is funded by the Office of Naval Research to perform research on the use of informative priors that add a small amount of information to stabilize a statistical analysis without overwhelming the information in data. Much of the work will be done in the context of particular applications in the social, behavioral, and natural sciences, and the ideal candidate will have a deep understanding of Bayesian modeling, an interest in applied statistics, and excellent computation skills. This work will take place in an interdisciplinary research environment, and the postdoc will have the opportunity to collaborate on multiple projects in statistics and applied research. The candidate is not required to be a U.S. citizen or permanent resident. If you are interested, please email me your CV, a letter of application, some of your papers, and three letters of recommendation. We’ve already started collecting applications, so apply right away.
2-year Earth Institute postdoctoral fellowship
The Earth Institute at Columbia brings in several postdocs each year–it’s a two-year gig—and some of them have been statisticians (recently, Kenny Shirley, Leontine Alkema, and Shira Mitchell). We’re particularly interested in statisticians who have research interests in development and public health. It’s fine—not just fine, but ideal—if you are interested in statistical methods also. The EI postdoc can be a place to do interesting work and begin a research career. Details here. If you’re a statistician who’s interested in this fellowship, feel free to contact me—you have to apply to the Earth Institute directly (see link above), but I’m happy to give you advice about whether your goals fit into our program. It’s important to me, and to others in the EI, to have statisticians involved in our research. Deadline for applications is 30 Oct, so it’s time to prepare your application NOW!
Dr. Gelman:
I ran across this () article literally claiming that drinking Sauerkraut juice increases sympathy for Nazi ideology (evidently it is “the first article that demonstrates that consumption of a healthy product also gives one a license to do something immoral”).
The sample size is suspiciously low, there appears to be room for multiple comparison shenanigans, and their difference in means looks like ‘lucky noise’ (small difference with barely significant p-value).
Would a more experienced researcher skim the paper and see if it has any evidence for a real psychological effect?
Ugh, bungled the link:
https://dx.doi.org/10.4236/psych.2015.69112
Tomas:
I took a quick look and I suspect this article is a parody. Or, if not a parody, it is a project along the lines of the sorts of experiments that are done in school biology class where the students divide a set of plants into two groups and play music to one group and not the other, and then compare how high the plants grow. That is, not intended to be taken seriously.
The article is not a parody. I reviewed the paper at a journal higher up the ladder. I recommended rejection, the other reviewer recommended rejection, and the editor added his/her own comments justifying rejection. This was a case of high inter-rater reliability. There are a several problems with the paper, some of them Tomas suggested, and I’m surprised the paper was ultimately published at all.
All of that said, this is a relative win. People often complain of this sort of study getting published in top journals. Here the study is published in one of the lowest tier journals possible. It did not get picked up by the NYT or WSJ or CNN etc. The main promotion mechanism of the paper are the people making fun of the paper (I first saw it after it was tweeted by a parody priming Twitter account), rather than the authors via the mass media. Plus, the journal it’s published in is not read or cited by any serious scholars in fields related to the paper. If we would collectively ignore the paper, it would go away as fast as it appeared.
Andrew:
See? Crap gets published in non-tabloid Journals too! :)
But the open box making people more creative, that one was for real, right? RIGHT?
Apart from all the silliness, this study fails because there was no baseline measure of rightwing-ness (RW). The authors used one measure (basically asking, “are you a Nazi?”) of all three groups, and determined that the difference between them was not statistically significant. They then got them to drink the sauerkraut juice, or Nestea, or nothing, and then applied a different measure (asking a series of questions about people’s attitudes to foreigners, etc). The difference between those groups was statistically significant.
However, as Andrew has pointed out a million times, the difference between significant and not significant is not itself significant. The authors should have performed a condition by time ANOVA. Indeed, if you look at the scores of the groups on the two measures (assuming for a moment that they are commensurate, which is plausible since they both measured some aspect of RW on a scale of 1 to 10), you will see that the sauerkraut group scored higher at baseline. So the authors have to choose: either they consider the two measures of RW to be commensurate, in which case a visual inspection of the descriptives suggests that their condition x time ANOVA will come out with an F somewhat less than 1.0; or, they don’t consider them to be commensurate, in which case they don’t have a baseline, other than “the difference between the groups on RW measure A was not significant, so we assume that they all had identical baseline scores on measure B”.
I had a brief exchange of e-mails with the lead author, who basically accepted the above points.
I also note that the journal in which this article was published is from a publisher that is considered “predatory”. I’m guessing that the study was run to make statistics interesting for the undergraduates who took part. As such, it seems quite fun, but I hope that the department in question doesn’t use this methodology for its serious work.
Nick:
Given that the word “playful” is in the title of the paper, I assume it’s safe to call it a joke, similar to various other jokes such as the salmon in the MRI scanner.
When the Association for Psychological Science or the British Psychology Society publishes such studies, I get more upset because I think these organizations are supposed to be serious.
That was what I thought, but the author seemed quite upset that a group of us had mocked the article on Twitter. At no point did he say “of course, we were joking, why are you taking it so seriously?”.
Of course, humour is hard to translate across national and cultural boundaries. This was a Swiss study, conducted by researchers whom I assume to be Swiss nationals. None of the German citizens I’ve asked about this found it very amusing.