Cornell and the First Law of Fooddynamics

If you’ve been following the Cornell Food and Brand Lab story, you may have seen the recent Retraction Watch interview of the lab’s big cheese, Brian Wansink.

Just like Wansink’s bottomless bowls, this is the story that just keeps on giving. There is so much to digest in the interview, it’s really a buffet of options.

I’m not even going to talk about how Retraction Watch asked some soft serve questions.

Or talk about how this is the second helping by Retraction Watch and they still haven’t tapped any of my coauthors for comment.

Or bring up that Cornell stated investigators can decide if they will share data “in the absence of sponsor or publisher data sharing requirements”, but BioMed Central has had an open data policy since 2011, so Cornell’s response was nonsensical.

No, I won’t talk about any of that. Because I want to talk about the First Law of Fooddynamics.

Retraction Watch linked to my Medium post and asked Wansink if he could explain why some numbers don’t seem to add up. He decided to go with what Neuroskeptic calls the “food fights and stolen apples” excuse:

The data in question are presented in this table:

The “number eaten” and “number uneaten” don’t add up to “number taken”.

In this study the “number eaten” was calculated by subtracting the “number uneaten” from the “number taken”.

If a food theft did occur, or food was dropped, or food was smuggled in pockets, or an all-out food fight did occur, that would have just resulted in less food remaining on the plates, and more food would have been counted as “eaten”.

That brings us to the First Law of Fooddynamics:

Number taken = number uneaten + number eaten + number stolen + number smuggled + number dropped + casualties of war

The researchers only measured “number taken” and “number uneaten”, and defined “number eaten” as the difference. As a result, for them:

number eaten = number eaten + number stolen + number smuggled + number dropped + casualties of war

The researchers essentially grouped all possible fates of food as “number eaten”. The fact that the food had destinies other than a digestive tract cannot explain why their “number eaten” and “number uneaten” do not sum to “number taken”.

If the researchers accounted for stolen food, smuggled food, dropped food, or food fight casualties, they would have had to explicitly subtract those numbers from “number taken”, which could account for why “number taken” does not equal “number eaten” + “number uneaten”.

But it still would not be able to account for their specific values. Their “number eaten” + “number uneaten” is larger than “number taken”. If they accounted for the other possible fates of food, it would have just decreased their value for “eaten”, not increased it.

As a result, no matter how you slice it and dice it, flip it or reverse it, they violated the First Law of Fooddynamics.

Perhaps to cover all his bases, Wansink provided a second excuse for why “number eaten” + “number uneaten” does not equal “number taken”. He says the values were

“based on the well-cited quarter plate data collection method referenced in that paper”

“That paper” is “Attractive names sustain increased vegetable intake in schools”, which was the subject of my blog post and the only study involving elementary schools I’ve discussed. It was published in 2012. The quarter plate method Wansink refers to was published by his group in 2013. Although half of the references in the 2012 paper were self-citations, the quarter plate method was not referenced as he claims.

The quarter plate method was brought up as a possible excuse for the discrepancies because the method relies upon visual inspection, which has a certain amount of inherent error. However, it is clearly stated in the paper:

the weight of any remaining carrots was subtracted from their starting weight to determine the actual amount eaten.

If this statement in the methods is taken seriously, then it is clear a visual method to determine carrot consumption was not used, and inaccurate measurements cannot explain the discrepancies, unless of course the scale used was extremely inaccurate.

Addendum 20170612:

Apparently Wansink’s bogus explanation even threw me off for a second. I was focused here on the fact that his interview blatantly contradicted what was written in the paper, but it actually doesn’t matter whether the measurements are accurate or not, they could be randomly picked out of a hat and there would still need to be conservation of carrots. The only logical explanation I have for why the numbers don’t add up is they have some missing values, for example they have more values for “taken” than “uneaten”.

Why does any of this matter?

Who really cares how many carrots were actually eaten?

As Wansink himself states, “Social science isn’t definitive like chemistry.”

Well, Wansink’s work is seen as important enough and accurate enough for him to speak in front of Congress about his work. And on the day the Retraction Watch article was posted no less.

Just as you can’t blame a cow for spoiled milk, you can’t blame stolen apples for violations of the First Law of Fooddynamics.