Going out to dinner? Read this before you order dessert.

Photo credit: http://www.caribbeannewsdigital.com

Photo credit: http://www.caribbeannewsdigital.com

"Diners order significantly more food and drink when served by an overweight waiter or waitress." That's the key conclusion of a recent study, as explained in PsyPost, which said people are "four times more likely to order dessert when their server is overweight." 


The article got more coverage, including "Does Weighty Wait Staff Influence the Bill?" from the Bakersfield Californian in an article highlighting the study in the Journal Environment and Behavior.  (The study also was cited in other articles found here and here.) 

This conclusion immediately drew the attention of Everydata.  As a sound consumer of data, here are some questions you should be asking:

First, ask yourself how could one go about collecting reliable data on what a person was eating at a restaurant, their size, and the size of the waiter?   I can barely estimate my own weight reliably, and I get on the scale fairly regularly.  This struck me as a lot of pretty specific data and it wasn't obvious to me how it could be collected.  

Based on the study, I found that teams of research assistants were sent to restaurants, and self-coded the information. The researchers estimated the ethnicity, weight, height, and age of Server and Diner, as well as other key data. Then, to get at BMI, "the body types of servers and diners were determined according to a chart featuring 18 different body types for each gender." How well were these various inputs actually measured? The extent to which this study can be relied upon to draw conclusions hinges critically on this question.

Second, ask yourself about sample selection--how do the specific collection locations potentially effect the results?  From looking at the study, there are at least three specific selection criteria that I would want to think more about.   One, restaurants that were chosen included places like TGI Fridays and Applebees.  Second, nearly half (slightly over 48%) of restaurants were in Ithaca, NY.  Third, restaurants in the sample had to meet these necessary selection criteria: "the availability of soup, salad and alcoholic beverages along with the complementary serving of bread or rolls."   

Third, ask yourself, even assuming this study is determining a statistical relationship between server wait and ordering dessert, is this causal or is it simply a correlation?   Any number of mechanisms could be driving these results, and many don't have anything to do with a server causing you to order dessert.   What if heavier waiters are heavier because they tend to work at restaurants that attract heavier people?  Or serve more foods that are bad for you?  Or are known for their particularly fattening desserts?  

These are just a few of the questions you should ask when you see a study like this, whether it's in a journal or your local paper.