A blog post leads to doubts about Cornell’s food laboratory | Hacker News Reader

A blog post leads to doubts about Cornell’s food laboratory

chronicle.com

77 pointsnkurz15 comments

This is what Andrew German has been covering as "Pizzagate" http://andrewgelman.com/2017/02/03/pizzagate-curious-inciden...

I had never heard of Wansink but it sounds like he's dominated the food science field, and has been the main purveyor of "common sense" concepts such as how serving food on a small plate will trick you into feeling fuller, and all of these findings are suddenly under attack [0]. Even worse, these attacks arise from blatantly obvious errors or anomalies that are apparent without even looking at the data -- which implies severe weaknesses in the peer review process. Given the scope and severity of the errors, it's really hard to imagine Wansink or the Food Lab surviving this.

I wonder if one of the main takeaways for other academics is to never blog? All of this only came about because Wansink seemed to sincerely want to give advice on how to succeed in academia. Though the blog post's tone of "just work really hard for free!" was already boiling into a scandal in itself, before people started actually reading what his hard working grad student actually produced.

[0] https://mobile.twitter.com/StuartJRitchie/status/84686385484...

A researcher with hundreds of publications hadn't heard of p-hacking or the reproducibility crisis until a month ago? Lay people who occasionally read science news have been hearing about those for years now. This is the "I'm too stupid to be a fraud" defense, apparently.

> It’s also led to an undercurrent of anxiety among scientists who fear that their labs and their publication records might come under attack from a feisty cadre of freelance critics.

Formerly known as "peer review".

> Formerly known as "peer review".

An in-depth review of a paper in advance of publication is common (although that system isn't as reliable as had one been assumed). Such an exhaustive review after publication is rare, for multiple reasons.

One serious problem with post-publication review is that the original data are rarely made available to reviewers. For example, academic critics of the work of Diederik Stapel couldn't justify revoking his Ph.D. because the original data that might have proved his fraud no longer existed. He eventually voluntarily gave up his degree[1], possibly to try to avoid a deeper investigation into his behavior.

1. http://www.science20.com/cool-links/dutch_psychologist_diede...

I love how you believe that most reviewers have time to conduct a thorough review for a journal, most of the time.

As opposed to something much closer to "LGTM!" With some "but be sure to cite my vaguely-relevant work in Foo et al, 1952".

> I love how you believe that most reviewers have time to conduct a thorough review for a journal, most of the time.

Your objection is to a position I never took -- in fact I suggested the opposite. The depth of a review is gauged by its length, not its accuracy or objectivity.

Ah, OK. I think we are on the same page. My mistake.

Bluntly, no. Laypeople are not the intended audience for scientific papers and usually lack the background to provide an informed review (hell, even most researchers in the field do). They do lack certain biases and incentives that researchers familiar with the topic usually have, but that doesn't compensate for what they don't know (for example, whether the claimed phenomenon has been tested many other times and not shown up; whether the claimed mechanism is bunk; whether the supposedly novel idea has already been published many other times).

That being said, laypeople can (and should) absolutely call out research malpractice like what's described in the article.

I disagree pretty strongly.

You've got worldwide universities churning out masters and pdh-level educated persons, who mostly go to the private sector for employment. Many of those would be perfectly able to perform peer review. And with this p-hacking crisis, mostly due to abject statistical knowledge, there are literally millions of technically able non-scientists able to review this nonsense. Any physicist, econometricist, statistician (and >100 more technical fields) could kindly take a shot at the social science methods in dispute here.

You're focusing on exactly one aspect of peer review, determining whether the statistical methods used were valid, which I agree can be done by pretty much anyone with a background in statistics (given that these days it's pretty easy to get in touch with the authors to ask for clarification). But peer review is about more than just determining whether the paper is internally coherent (or directly lifts text from other papers by the same author, which can be determined by a machine). I shouldn't have to say this, but a former physics Ph.D. is emphatically not qualified to perform peer review on papers on cell biology solely because physics and cell biology are both science. Papers don't exist in a vacuum, and by engaging in practices like selective citation (and others I mentioned in the previous post), it's possible to convince intelligent and reasonable people of a point that is completely wrong. In fact, it's so easy it even happens somewhat routinely within narrow subspecialties, where there may only be a handful of experts on a phenomenon worldwide. People who aren't up to date with the field already usually don't have time to spend weeks, months, or years trudging through the literature to gain enough background knowledge to make those kinds of decisions.

So, peer review doesn't scale, and some alternative is needed.

Sounds about right. How about we crowdsource the task? Since all the qualified specialists are busy writing grants against a 15% payline, we can farm it out to people without those pressures.

Oh... wait...

You're right (and we probably agree). I did focus on the number 1 problem in social science / pop psychology today -- sloppy statistics.

But you do make a convincing point. The 'stink' that is uncovered nowadays is mostly in those sciences where the layman / statistically schooled can make a difference. We hear a lot less about the replication crisis in those sciences where one indeed needs to have a lot of background knowledge before being able to do peer review. That's a problem.

There is no reason to assume that the same problems are not paramount in cell biology and other fields. We actually have enough reason to believe that many trials in medicine are flawed. Given that researchers in all fields face the same incentives (publish or perish) we are probably looking at at least 20 years of false, peer-reviewed and published results in all fields. That's a huge problem for mankind.

> His insights have also attracted interest — and, more importantly, money — from policy makers. In 2014 the U.S. Department of Agriculture forked over a $5.5-million grant to support Wansink’s Smarter Lunchrooms program, which has since been put into practice in more than 30,000 schools across the country.

there's something painfully appropriate about that quote appearing on the same page with a "trending" article titled: "Trump Proposal to Cut Indirect Research Payments Would Hit State Universities Hardest"

Is this another example of why journals should publish negative results? Is there a way to review papers where you are just evaluating the soundness of the methodology regardless of results?

"Do good science" is easier to set up reliable cultural practices around than "discover something new".

Preregistration, as demanded for clinical trials, is essentially the only way to ensure that researcher degrees of freedom are avoided.

Buy the ticket, take the ride. Needless to say, it's a scary ride, so most groups won't do it.

Reply on news.ycombinator.com