Via Rachel and Meowser, I learned of a study, “Eating Behavior and Obesity at Chinese Buffets,” published by the academic journal Obesity. The nice folks at Cornell University’s Food and Brand Lab emailed me a copy of the study.
In a nutshell, they sent “trained observers” to various Chinese buffet restaurants. 213 patrons were spied upon for behaviors such as sitting in a booth or at a table, using chopsticks or forks, placement of napkin, how much was left on the plate, how many times each bite was chewed, etc.. The study authors theorized that these behaviors caused higher weights, but admitted that they couldn’t establish the direction of causality.
Personally, I don’t mind creepy spy studies like this one; a restaurant is a public place, and I like my science creepy. (If the researchers could indulge me with an occasional “bah hah hah!” laugh and gigantic goggles, so much the better.) But the methodology and thinking behind this particular study are jaw-droppingly awful.
The post is going to be very long, so I’ll first list some highlights:
1) There’s no reason to think visual assessments of weight by secret observers are accurate — and the citations the study used to support this methodology are, when you look them up, inapplicable.
2) Nothing in the study protects against the observations being tainted by bias and stereotypes about fat people.
3) Their sample of “normal” weight people included underweight, and even severely underweight, people.
4) A theory they describe on page one — that fat people go to the buffet more often — is not only not tested for by the study, the study is designed to exclude buffet trips from the study’s results.
5) Some causal connections speculated about in this study incorporate ridiculous anti-fat stereotypes. For instance, the study assumes that chopstick users “probably always eat with smaller utensils” when at home, and that people who don’t put their napkins on their laps lack “table manners” and therefore lack “careful consumption monitoring.”
Details after the fold.
1. “That guy across the room looks 214 pounds to me.”
Their measurement of weight and height consisted of “trained observers” making their best guess. To justify this, the authors write that “training has been shown to improve these estimates with the worst estimates still resulting in very high correlations.” But the two studies they cite to support this claim — here and here — show that even trained medical professionals, who have years of experience asking hundreds of people their weights, are “significantly inaccurate for individual observers.” Furthermore, doctors and nurses in an E.R. closely examine their patients — unlike observers seated “in unobtrustive locations in the restaurants.”
The authors don’t establish a credible basis for assuming their observers estimate height and weight accurately. They should first have conducted a study to find out if non-medical observers can be trained to accurately access weight and height of strangers in street clothes in secret from across a room.1
In my experience, perceived weight has a lot to do with class; someone in cheap, ill-fitting clothes may seem fatter to observers than someone of the same weight but wearing better fabrics and cuts. This is important, since many of the behaviors this study focused on may be related to class background (use of chopsticks and placement of napkins, for example).
2. Observer Bias
It’s well-known that observer bias can effect both how subjects are classified, and how their behavior is measured. (For example, social scientists have shown that interviewers for the highly-respected National Longitudinal Survey of Youth are less likely to classify the same subject as racially “white” if he’s in jail, and more likely to classify him as “white” if he’s not in jail.)
According to this study’s results, “low BMI” subjects chewed their food an average of 15 times per bite, versus 14 times per bite for overweight subjects and 12 times per bite for obese subjects.2 A small difference like that could easily be accounted for by observer bias, if observers believe that fat people chew their food less. Similarly, how reliable is the observation that low bmi eaters leave 11% of their food on their plates, compared to 7% or 6% for heavier eaters? Will a thin person who walks five or six steps before putting food on her plate be coded as “browsing the buffet” before serving, while a fat person with identical behavior is coded as diving right in?
Observer bias could also operate in the other direction. An observer, observing a subject chewing only 10 times or leaving only 3% leftover or taking food without “browsing,” could be influenced to decide that the subject’s weight is “actually” 180, rather than 150.
There are established methods of reducing uncertainty — such as having multiple observers independently observe and report on the same subjects and actions. That wasn’t done in this study.
3. Underweight and severely underweight patrons included in “normal” weight group.
Quoting from the study:
As Table 1 indicates, the 213 patrons were grouped into categories that represented the bottom third (n = 71), middle third (n = 70), and top third (n = 72) of their estimated BMI. These generally corresponded to the common descriptors given to normal weight, overweight, and obese individuals.
Yet according to Table 1, their classification of “normal” weight included patrons with BMIs as low as 13.5, a BMI that is medically classified as “severely underweight.”3 “Underweight” people in general have a greater risk of premature death than “overweight” (pdf link). If you take the weight/health link seriously (and if you don’t, what on earth is the point of this study?),4 then it’s absurd and irresponsible to suggest that more people should take on the eating habits of a group that includes underweight people.
When I asked study co-author Collin Payne about the inclusion of severely underweight people in his sample, he argued that there were very few underweight people included. But if there were too few underweight subjects to matter, why not exclude them altogether? Or report results for underweight people separately, as most BMI studies do? Even a couple of outliers can easily skew the statistics of a larger group.
As a reader, I shouldn’t have to be told by the study’s co-author, over the phone, not to worry about the inclusion of underweight people in the “normal weight” sample, because in his opinion (without having the numbers in front of him) it didn’t matter. That’s not the right way to do a study.
4. The study’s stated theory was never tested for in the study.
This is the study’s first sustained discussion of a theoretical connection between BMI and behavior, and was printed on page one:
Observers coded whether patrons sat at a booth vs. a table and whether patrons faced the buffet while eating vs. their side or back. Increased effort has been shown to decrease consumption (9). Because booth seating (vs. table seating) may make it more difficult to return to the buffet (i.e., eating companions may need to move, heavier patrons may not comfortably fit) it would be expected that this type of habitual seating would be related to lower BMIs. Increased salience of food has also shown to increase consumption (10). Patrons facing the buffet are expected to have higher BMIs than those whose side or back are toward the buffet. The continual salience of food that one is facing might keep it too temptingly top-of-mind.
So they found that heavier patrons were more likely to sit at tables, facing the buffet. They speculated that this behavior encourages heavier patrons to “increase consumption” by “return[ing] to the buffet.”
But here’s the problem: The study notes that their analysis “controlled for the random effect of observer bias and the fixed effects of sex, perceived age, and number of return trips to the buffet.”5 This means that the theory written about on page one of the study — that fatter patrons return to the buffet more often because of their seating behavior — is never tested for by this study.
To test their theory, they would have had to run statistical tests to see if seating patterns are actually associated with number of trips to the buffet or not. (Or, for that matter, to see if number of trips to the buffet is actually associated with BMI.) They could have run these tests. But they didn’t. Collin Payne told me that there wasn’t room — but how can there be room to explain a theory at length, but no room for even a single sentence regarding if their results support their theory or not?
In fact, by showing that higher BMI patrons are more likely to sit at tables facing the buffet even after controlling for return trips to the buffet, they’ve proven a relationship between weight and seating that can’t be explained by return trips to the buffet. That’s what “controlled for…. return trips to the buffet” means.6
So — aside from the study’s implication that fat people are ravenous pigs — why is it that fat people are more likely to sit at tables, and facing the buffet? Because booths are often too small for fat people’s comfort. (This explains why obese people avoided booths but merely “overweight” people didn’t, according to Table 2.) And, in restaurants where tables are adjacent to the buffet, fat diners might sit opposite the buffet to avoid blocking foot traffic.
5) Ridiculous variables which incorporate anti-fat stereotypes: chopsticks and napkins.
From the study:
Those using these smaller utensils would be expected to have lower BMIs than those who use larger utensils. This is because it is assumed that patrons’ utensil use is not unique to the single eating occasion noticed in this study—patrons who eat with smaller utensils probably always eat with smaller utensils. The placement of a napkin on a patron’s lap may also be related to lower BMIs. This may be because napkin wearing is reflective of table manners, which has been associated with more careful consumption monitoring.
In other words, fat people are slobs with no manners.
There’s an obvious reason fat people are less likely to put the napkin on their lap — it doesn’t do any good there. As “Buttercup” writes in comments at The F-Word, “I can put a napkin on my lap six ways to tuesday and it won’t do a damn bit of good. my boobs or belly catch any wayward food long before the napkin would.”
The study’s lead author, Brian Wansink, knows that fat people may not put napkins in our laps because they’re useless there — he’s quoted saying as much in a LiveScience article charmingly entitled “the science of pigging out” (no bigotry there!).7 Yet he doesn’t mention this possibility in the study, instead focusing on insulting stereotypes about fat people’s bad “table manners.”
To compound the silliness of measuring chopstick usage, the study didn’t control for race or class. But Chinese-Americans have lower average BMIs, and — at least in the restaurants I go to — are much more likely to use chopsticks than forks, and are a significant portion of chopstick users.
Nor did the study control for arthritis or class — factors probably related to both body shape (especially perceived body shape, since people with less money may wear lower-quality clothing) and chopstick usage.
Conclusion: Rachael at “The F Word” says that this lead author has published credible work in the past, and I don’t doubt her. But this study is nonsense. It is poorly conceived and badly executed, and if it didn’t align neatly with anti-fat stereotypes, it probably wouldn’t have gotten through peer review.
Please, scientists, please: If you can’t study fat people intelligently and without bias, just leave us alone.
- I admit, this might be hard to accomplish. But “it would be hard to do it right!” is no excuse for doing it wrong. [↩]
- Yes, there are people who sit in restaurants counting how many times you chew per bite. How creepy will that thought be the next time you eat out? [↩]
- To picture a 13.5 BMI, imagine someone five and a half feet tall who weighs 84 pounds. [↩]
- To be clear, I think the health/weight link is seriously overstated. But I doubt the authors of this study would agree with me. [↩]
- I also confirmed this on the phone with Collin Payne. [↩]
- Full disclosure: When I suggested this criticism to study co-author Collin Payne, he argued that I’ve misunderstood the study. Could be. [↩]
- Obviously, I don’t blame the study authors for the LiveScience headline. [↩]