Wednesday, October 1, 2014

Walkabout Wednesday: replication, wolves and student evaluations.

Replicating scientific research is important. The ability to replicate a study's results is what sets science apart from other ways of knowing. And sometimes, failure to replicate can overturn an entire paradigm. It turns out that Angus Bateman's classic study on fruit fly sex had a fatal flaw, exposed when Patricia Gowaty* and colleagues attempted to replicate the work after 64 years. And so goes the notion that women are passive objects of male-male competition. Finally. Probably not. Overturning paradigms is hard. 

And here's another example of replication leading to self-correction from physics.

It's been a banner week for replication! Here's a nice article on how Bayesian statistics is helping many disciplines, arguing that Bayesian methods allow checking conventional Frequentist conclusions and therefore improving replicability. Most interesting to me are the references to Andrew Gelman's work on detecting spurious results. However, he takes issue with way those ideas were presented in the article. While I agree there's a "crisis of replication" in science, I'm not sure Bayesian methods are the cure-all the article makes out.  (ht: Jeff Thompson)

I love wolves, and they've provided plenty of fodder for people interested in the intersection of science and policy (see here and here, for example). Now a federal judge has put management of Wyoming's wolves back on the USFWS, calling the handover to Wyoming's state agency "arbitrary and capricious". I gave Wyoming a D- on their AM strategy, so maybe that played a role?

The public regards Scientists as competent but not warm. And they can tell when we're playing "stealth issue advocate" so we should always seek to be "honest brokers". I think stealth issue advocacy is responsible for much of the crisis of replication mentioned above. 

The federal government is preparing to rate colleges and universities to improve students and families ability to make good decisions, and ultimately reduce costs. Here's one op-ed that suggests it won't work, at least at public institutions like mine. "At public colleges, then, the explanation for rising tuition prices isn’t spiraling costs. The costs are the same, but the burden of paying those costs has shifted from state taxpayers to students." (HT: Jeremy Fox @ Dynamic Ecology)

Pre and post testing of physics students at MIT indicates that students taking a MOOC do better than a traditional format.  But: "Although approximately 17,000 people signed-up for 8.MReV, most dropped out with no sign of commitment to the course; only 1500 students were “passing” or on-track to earn a certificate after the second assignment." That's the problem with MOOCs, if it is a problem. The paper is open-access, and uses alot of interesting ideas about how to measure student performance. (HT: Jeremy Fox @ Dynamic  Ecology)

And student evaluations suck. 

This is quite a good talk about Gluten; I hadn't heard of Rodney Ford before but he's a pretty good speaker. I eliminated gluten from my diet completely in early 2014. I will write a post about how I've been feeling since then soon. But here's a teaser: better.

*I was lucky to meet Patty when I was an MSc student at Simon Fraser University in the early 1990's. Really cool work then and now! 

No comments:

Post a Comment