Also this week: sleep vs. you, Tony Ives vs. statistical machismo, tips for gender-balancing your seminar series, the origin of deanlets, a rare retraction in ecology, why ecologists and evolutionary biologists give good talks, and more. Lots of good stuff this week!
Improve the gender balance at your conference (or in your department’s seminar series) using these four simple, straightforward tips. One thing they suggest is that it can be helpful to have a list of names of women in a particular field. I have parked the domain ecoevowomen.wordpress.com for this purpose, but haven’t done anything with it. (I got the idea to park that domain based on Anne’s List, which highlights women neuroscientists.)
I am so glad to read that I’m not the only person who gets anxious when receiving vague email requests to meet. There is no surer way to get my anxiety up than to send an email saying, “Can we chat some time tomorrow?” with no indication of what the email is about. And I don’t just worry if it’s from a boss-like figure. It’s also true when I get a vague email from a collaborator, colleague, or lab member.
I also enjoyed this post by PsycGirl, who has the helpful reminder that people will hate you. This is something that is hard for me (and, based on the twitter discussion, apparently for many others, too), but that I’ve been working on. One thing that I remind myself of is that other people will disagree fundamentally with some of my values (just as I will disagree with theirs). If we didn’t disagree, something would be wrong.
Tim Poisot says reviewers shouldn’t enforce R as a standard (or even enforce open source software, he might have added). Tim’s right. And as Ethan White notes in the comments, if you disagree then what you’re really saying is that R shouldn’t exist because everyone should’ve just stuck with
adding machines whatever everybody used before adding machines SAS.
A rare retraction in ecology: a high-profile paper in Global Change Biology, which found that plants migrate to lower elevations in response to global warming, has been retracted because of a coding error. The authors were careful about double-checking their results but nevertheless missed the error, which happens. And as soon as the error was discovered, the authors did the right thing and retracted, for which they deserve kudos. So I’m not sure why the reviewer who discovered the error sounds snarky about it. Anybody can make an honest mistake, and it’s bad for both individual scientists and science as a whole to pretend otherwise. More on this from Brian in a forthcoming post.
Tony Ives shows that it’s fine to just log-transform your count data and use least-squares linear models for null hypothesis tests. You only sacrifice a little bit of power vs. a properly-specified generalized linear model or generalized linear mixed model, and your inferences about the null hypothesis will be much more robust to model mis-specification, preventing inflation of the type I error rate. Tony 1, statistical machismo 0. I’m curious if Tony was prompted to write this paper because he ran into reviewers insisting on generalized linear models in a context where transforming the data and then doing least squares was just fine. (ht Meg)
The NSF Biological Sciences directorate has a blog now. (ht Terry McGlynn, via Twitter)
A hypothesis on why your university has so many vice-deans and other administrative layers, and so much red tape. I’ll add my own hypothesis (based on no evidence or even anecdata): faculty don’t like having to do administrative tasks, so ask for more administrative support. Which results in more administrators getting hired. Which then generates more admin work for everyone. And so the cycle repeats. This isn’t a criticism of individual administrators, the large majority of whom are competent and hardworking. It’s a hypothesis about the dynamics of the whole system.
Speaking of why universities are the way they are…How come any list of the top universities in most any western country has about the same rank ordering as it had a century ago? Whereas most of the biggest corporations from a century ago no longer exist? And what are the implications for university management and national higher education policy? Interesting discussion that I’m still mulling over.
Simply Statistics on the
curse blessing of dimensionality. That is, why it can actually be useful to have many variables that were measured on only a few subjects.
I’ve always had the impression that EEB folks give pretty good talks on average, because EEB folks know to foreground the big questions and general concepts. So I was interested to read that system biologist Arjun Raj thinks that too many cell/molecular/biochemical talks lack big questions and general concepts.
This is old but I missed it at the time: Here’s a video of Sarah Hird (of Nothing in Biology Makes Sense! fame) giving a hilarious BAHFest talk on why mammals sleep. It’s only 7 minutes, click through!🙂
Aww, isn’t that a cute bunny rab…OH GOD RUN FOR YOUR LIFE!!!11!🙂 (ht Marginal Revolution)
And finally: the most 1970s thing you’ll read this week. I give you the US Forest Service on how to make a cocktail. Diagrams and all! Yes, really.🙂