Friday links: predictions need explanations, scientific band names, and more

From Jeremy:

E. D. Hirsch Jr. on how most educational research is “cargo cult science”. There are lots of rigorously conducted randomized experiments with statistically-significant results–but yet little if any generalizable policy guidance. Hirsch argues that this is because the research is atheoretical and phenomenological. A “just the facts” approach aiming for purely statistical description and prediction of what works and what doesn’t leads to what’s effectively a stamp collection of case studies that can’t be extrapolated to even seemingly-similar contexts. Hirsch argues that there’s no prediction without explanation–you can’t discover or predict what works without the guidance of good theory about why it works. Go read, argue if it’s right, discuss relevance to ecology (it’s a great read, not too long, and not technical). As part of your discussion, try to think of counterexamples–successful, predictive sciences based on phenomenological experiments that treat the study objects as “black boxes”. Some chunks of medicine, maybe? (indirect ht Denim and Tweed)

Science blogosphere makes contact with outside world. Very sharp and funny satire, with more than a little truth to it. “It is believed that 80% of science blogs are actually about other blogs.” Maybe it’s for the best that there’s no ecology blogosphere! And before someone points it out, yes, I am aware of the irony of blogging about this piece. (ht Terry McGlynn, via Twitter)

Ten reasons why scientists should read non-scientific literature. I totally agree with the general point, although I’d quibble with the details. Somewhat contra the post, I don’t think there’s much correspondence between the type of literature you read and what you gain from it, except for history and biography. And I don’t know that scientists necessarily are more likely to appreciate, or get more out of, literature with scientific themes as opposed to any other sort of literature. But while the recommendations do skew towards literature with scientific themes, they span a really wide range of time periods, forms, genres, and styles. (ht Small Pond Science, via Twitter)

Data on changes since 1990 in the number of people employed in various roles in US higher education, both overall and on a per-student basis. Basically, expanding student numbers have been dealt with by hiring more part-time and time-contract faculty. And also by hiring more administrators, the justification for which isn’t always clear. (ht Economist’s View)

The wolf population on Isle Royale may be on the verge of dying out, which would terminate one of the longest-running and most famous studies of natural predator-prey dynamics in history. I have a bit of personal interest in this as I did my undergraduate honors thesis on Isle Royale (on rockpool microplankton). I have fond memories of all the moose I saw–I was there in the summer of 1994, just before a big crash in the moose population. Deciding what, if anything, to do about the possibly-imminent wolf extinction isn’t an easy call, either on scientific grounds, or conservation or legal grounds.

My friend and Seattle resident Greg Crowther notes that the crowd size at the Seattle Seahawks Superbowl victory parade was badly overestimated by the media–and that people in Seattle do not like it when you try to point this out. Another blowout victory–for motivated reasoning over facts.

A Twitter robot that automatically generates fake New Scientist headlines. “Tomorrow’s technology today: inflatable pregnancy tests.” “Isolated moose deaths hint that time itself could be reversed.”πŸ™‚ (ht Scholarly Kitchen, via Twitter)

xkcd pokes fun at Meg! Well, sort of.πŸ™‚

And finally, phrases from scientific papers that would make good band names.Β My addition to the list would be Chaos Chaos*. Chaos chaos is the Latin binomial of a giant amoeboid protist (actually, the more accepted name is Chaos carolinensis, but that would be a terrible name for a band). I’ve long dreamed of somehow rigging up culture conditions that would cause this organism to exhibit chaotic population dynamics, just so I could write a paper titled “Chaos in Chaos chaos“.πŸ™‚

*Turns out there is a band called chaoschaos! No idea what sort of music they play. Or if they realize they’re named after a giant amoeba.πŸ™‚

18 thoughts on “Friday links: predictions need explanations, scientific band names, and more

  1. Thanks for the link to the Hirsch essay! I’ll let actual ecologists judge the essay’s relevance to ecology, but it’s interesting to ponder as many scientists (like me) start venturing into science education research. Given that we are generally teaching undergrads, we are likely to do our studies in the classroom (e.g., give section A this assignment and section B this alternative assignment…) rather than the lab. Can those studies still be useful? Do we need to team up with more cognitive scientists, etc.?

    • Here at Calgary there are several instructors in Biosci who are collaborating with folks from the Faculty of Education on pedagogical research. I think that sort of collaboration is the way to go for scientists who want to get into this area.

  2. A more sobering comment related to the Hirsch piece on atheoretical/phenomenological science: Did you read the Yet Another P-Value opinion in Nature yesterday? The article isn’t bad overall (although it doesn’t explain false discovery rate very well) but highlighted a young academic in psychology who’s fame evaporated when a replication produced a high P-value: “Motyl and his adviser, Brian Nosek, decided to replicate the study. With extra data, the P value came out as 0.59 β€” not even close to the conventional level of significance, 0.05. The effect had disappeared, and with it, Motyl’s dreams of youthful fame.”

    Huh? Do scientists really think a P-value will bring fame? And I thought fame came from cleverness, as in a clever theory or a clever experiment. Or maybe fame just comes from hard work and doing the theory or experiment that takes forever, so no one does it. I think only in an atheoretical/phenomenological field, P-values could produce fame.

  3. “Turns out there is a band called chaoschaos! No idea what sort of music they play.”
    I would hope they would cover that old Youngbloods classic “Darkness, Darkness”!

    Haven’t read the Hirsch piece, but if he really believes that good prediction can only come from good system understanding, then he pretty clearly has screws loose somewhere.

    • Well, read the piece and see what you think. His comments are grounded in educational research. As I said in the post, one natural question to ask is about the circumstances under which purely phenomenological statistical prediction is likely to work, and the circumstances under which it isn’t likely to work. This is an issue we’ve touched on in various old posts (e.g.,, and I keep coming back to it because I think it’s an interesting, challenging, and important issue.

      • OK, I read it. I can definitely sympathize with his principal point, which seems to be: experimental studies in educational practice need to be based on an understanding of human cognition, or similar types of fundamental-level understandings of human psychology (and/or biology?), in order to make real progress. It seems to be an application of the general idea that science progresses fastest when theory is developed/derived from understandings of more fundamental, lower level processes, a viewpoint with which I agree pretty strongly.

        This is not to say however, that doing controlled experimentation on classroom size, like the STAR study, is therefore *necessarily* a losing proposition, which I think he seems to sort of imply. I can’t imagine that they’d have conducted the experiment if they didn’t have some underlying reason for thinking that factor might be important, even if it does sort of represent a black box with no theoretical underpinnings. On the other hand, it undoubtedly cost a lot of money, both the study itself and for the school districts who made changes based on it, and you can’t conduct major experiments like that on every idea under the sun. You have to be efficient.

        I did think a couple of his examples were pretty weak or hard to follow however. Like the bit about gender determination in young chickens…not terribly relevant.

  4. There was an Australian band called Exploding White Mice, which would be a good name for a scientific paper. The band’s name was taken from a scene in the 1979 film Rock ‘n’ Roll High School, where a laboratory mouse spontaneously explodes upon exposure to music by The Ramones.

    One paper I found: “As the worm turns.” Bull Environ Contam Toxicol *2004) 72:219-224

  5. I had been thinking about the issues raised in the Hirsch article from the opposite direction, can advances in ecology inform the application of educational research to classrooms? I am neither an ecologist or ed researcher but from what I do know about both fields have theoretical bases from which to start but are often beset with strong context dependent signals. I am thinking of the work by Oswald Schmitz, trying to understand context dependency. A better understanding of how teachers, schools and students (like players in a food web), create context may help education avoid some of the pitfalls Hirsch discusses. Not being fully familiar with ecology literature I would be interested in other analogous approaches.

Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s