PreprintPDF Available

Retracted Journal Articles and New Research Opportunities to Change Eating Behavior

Authors:
Preprints and early-stage research may not have been peer reviewed yet.

Abstract

Brian Wansink 2019 - 3 In 2017-19, about 18 of my research articles were retracted. These retractions offer some useful lessons to scholars, and they also offer some useful next steps to those who want to publish in the social sciences. Two of these steps include 1) Choose a publishable topic, and 2) have a rough mental roadmap of what the finished paper might look. That is, what’s the positioning, the study, and the possible contribution. The topics I’ve described here offer one set of roadmaps that could be useful. First, they were of interest to journals in medicine, behavioral economics, marketing, nutrition, psychology, health, and consumer behavior. Second, they each show what a finished paper might look like. They show the positioning, relevant background research, methodological tips, and key implications. I find all of these topics super interesting and of practical importance. This document provides a two-page template for each one that shows 1) An overview why it was done, 2) the abstract (or a summary if there was no abstract), 3) the reason it was retracted, 4) how it could be done differently, and 5) promising new research opportunities on the topic.
RETRACTED JOURNAL ARTICLES
AND NEW RESEARCH OPPORTUNITIES
TO CHANGE EATING BEHAVIOR
Brian Wansink, Ph.D.
December 2019
TABLE OF CONTENTS
Introduction and Overview 3
Table 1. Which Research Questions Might Be Most Useful to Answer? 4
RESEARCH TOPICS
Do Healthy Ingredients Make You Hate the Food? 5
Can Taste Profiles Predict Food Preferences? 7
Do Large Serving Bowls Make You to Eat More? 9
Why Do Overweight People Underestimate How Much They Eat? 11
Can You Confirm the Sweet Tooth Hypothesis? 13
Do “Clean Plate” Kids Turn into Overeating Adults? 15
Have Classic Recipes Increased in Calories Over Time? 17
Will You Hate the Food You Eat During Bad Experiences? 19
Can Brand Logos Encourage Kids to Eat Heathy Foods? 21
Do Short-Term Fasts Lead to Long-term Weight Loss? 23
Will Cool Names Lead Kids to Choose Healthy Foods? 25
Do Hungry Shoppers Buy More or Just Buy Worse? 27
Does Preordering Lead to Healthier Lunches? 29
Do Different TV Shows Influence How You Eat? 31
Do High Menu Prices Make You Regretfully Overeat? 33
Does Traumatic Violence Change Judgement and Choice? 35
APPENDICES
Appendix A. JAMA Request to Cornell for Research Validation 37
Appendix B. Peer Review Form for Research Integrity Investigation Reports 40
Appendix C. Cornell’s Investigation into Possible Errors in Six JAMA Papers 42
Appendix D. What are Some Useful Eating Behavior Questions to Answer? 46
Brian Wansink 2019 - 3
In 2017-19, about 18 of my research articles were retracted. These
retractions offer some useful lessons to scholars, and they also offer some
useful next steps to those who want to publish in the social sciences. Two of
these steps include 1) Choose a publishable topic, and 2) have a rough mental
roadmap of what the finished paper might look. That is, what’s the positioning,
the study, and the possible contribution.
The topics I’ve described here offer one set of roadmaps that could
be useful. First, they were of interest to journals in medicine, behavioral
economics, marketing, nutrition, psychology, health, and consumer behavior.
Second, they each show what a finished paper might look like. They show
the positioning, relevant background research, methodological tips, and key
implications.
I find all of these topics super interesting and of practical importance.
This document provides a two-page template for each one that shows 1) An
overview why it was done, 2) the abstract (or a summary if there was no
abstract), 3) the reason it was retracted, 4) how it could be done differently,
and 5) promising new research opportunities on the topic.
Table 1 and Appendix D lay out an estimate of how much effort it might
take to do studies on these topics, and Appendix B lays out other issues related
to how these specific papers were investigated. I’ve also estimated what I think
the practical impact each research project might have. These are my own
subjective estimates, but you might find them a useful starting point if you’re
looking for a tie-breaker between two different topics.
I would strongly encourage anyone who’s interested in publishing in
these areas to closely follow the principles of open science. You can start by
preregistering hypotheses and planned analyses, and following the other steps
along the road to publication. Making specific hypotheses and testing them
followed by open science principles will be the best next way forward on these
topics.
Academia can be a tremendously rewarding career both you and for the
people who benefit from you research. Best wishes in moving topics like these
forward, and best wishes on a great career.
Sincerely,
Introduction and Overview
1 A useful description of these principles can be found at Klein, O., Hardwicke, T. E.,
Aust, F., Breuer, J., Danielsson, H., Hofelich Mohr, Al, …. Frank, M. C. (2018). A
Practical guide for transparency in psychological science. Collabra: Psychology,
4 (1), 20.
Brian Wansink 2019 - 4
Research Question Original
Publication Year Potential Practical
Usefulness1
(1=Lower; 5=Higher)
Potential Effort
Required1
(1=Easier; 5-Harder)
Page
Number
Do Healthy Ingredients
Make You Hate the Food?
Journal of
Sensory Studies 2002 3 1 5
Can Taste Profiles Predict
Food Preferences? Appetite 2003 3 1 7
Do Large Serving Bowls
Make You to Eat More? JAMA 2005 4 1 9
Why Do Overweight
People Under-Estimate
How Much They Eat?
Annals of Internal
Medicine 2006 5 3 11
Can You Confirm the Sweet
Tooth Hypothesis? Appetite 2006 2 1 13
Do “Clean Plate” Kids Turn
into Overeating Adults?
Archives of Ped &
Adolescent Medicine 2008 3 1 15
Have Classic Recipes Increased
in Calories Over Time?
Annals of Internal
Medicine 2009 2 1 17
Will You Hate the Food You
Eat During Bad Experiences? Appetite 2009 2 2 19
Can Brand Logos Encourage
Kids to Eat Heathy Foods?
Archives of Ped &
Adolescent Medicine 2012 3 2 21
Do Short-Term Fasts Lead to
Long-term Weight Loss?
Archives of Internal
Medicine 2012 3 2 23
Will Cool Names Lead Kids
to Choose Healthy Foods? Preventive Medicine 2012 4 2 25
Do Hungry Shoppers Buy
More or Just Buy Worse?
JAMA Internal
Medicine 2013 4 1 27
Does Preordering Lead to
Healthier Lunches? JAMA Pediatrics 2013 5 4 29
Do Different TV Shows
Influence How You Eat?
JAMA Internal
Medicine 2014 2 3 31
Do High Menu Prices Make
Your Regretfully Overeat? BMC Nutrition 2015 3 3 33
Does Traumatic Violence Change
Judgement and Choice?
Frontiers in
Psychology 2016 2 4 35
Table 1
Which Research Questions Might Be Most Useful to Answer?
1 Estimates of potential usefulness and effort are subjective.
Brian Wansink 2019 - 5
DO HEALTHY INGREDIENTS
MAKE YOU HATE THE FOOD?
Have you tried the meatless “Impossible” Whopper at Burger King?
If you hear that a snack has a healthy ingredient in it, does your mouth stop
watering? Lots of products have tried to promote themselves as having a
healthy ingredient. Yet this might be a big turn-off for a lot of people.
Let’s see what would happen if we tell people a product has a healthy ingredient
even though it doesn’t. Are they going to hate it because of the fake ingredient
that’s not even there? If they do, there’s important precautions to take.
The Original Findings
A similar version of this paper with the same data had been published
in an agriculture journal in 2000, and this version was published in
the Journal of Sensory Studies in 2002.1 Here is the abstract:
The paper was retracted due to major overlap with a previously published
article: Wansink B, Park SB, Sonka S, Morganosky M (2000) How soy labeling
influences preference and taste. The International Food and Agribusiness
Management Review 3: 85—94. doi: 10.1016/S1096-7508(00)00031-8.2
1 Wansink, Brian; Park, Se-Bum (November 2002). "Sensory Suggestiveness and Labeling: Do Soy Labels
Bias Taste?". Journal of Sensory Studies. 17 (5): 483—491. doi:10.1111/j.1745-459X.2002.tb00360.x.
2 https://onlinelibrary.wiley.com/doi/10.1111/joss.12259
Other Ways to Answer this Question
1. Hypotheses and Extensions: Many sensory studies involving new or different
ingredients have added a new ingredient (or ingredient replacement) to a
product and then tested to determine if people can detect a difference between
the old product and the reformulated one. An approach that seems to not be
used before is to tell people there’s an ingredient in a food even though it isn’t
Brian Wansink 2019 - 6
People ate the food in a room modified
to look and feel like a kitchen.
The product they ate was a chocolate
PowerBar with a modified label.
there and then run a taste test to see if they can “taste” this
missing or phantom ingredient. This examines whether
the suggestiveness of just thinking this ingredient is in the
product is enough to make them think they taste it.
Using this phantom ingredient approach worked well when this
study was run back in 1998 because back then soy had a bad
taste perception among most consumers. The same phantom
ingredient test could be used to examine if there are unarticulated
biases against other new product ingredients — such as meatless
meat and Impossible Whoppers. There are dozens of new
ingredients that could be influenced by this Phantom Bias.
2. New Methodology Ideas: Many sensory studies involve a small
piece of a product in a labeled baggie, or in little labeled cups.
Presenting foods in this way might be seen by some consumers
as artificial, especially if they taste these in tasting cubicles. An
environment like this could reduce the impact of this labeling.
One way this could be eliminated is by having the person
taste the product in a naturalistic environment, such as
kitchen or kitchenette with other snacks that are sitting
out. Additionally, another way to make the product
seem more realistic is to print colored labels and to
package it as close to a real product as possible.
Many taste tests use a standard set of validated descriptors.
If you are analyzing different ingredients that you fear might
have a unique sensory property (such as gritty or fatty) or some
unique perceived side-effect, try to include these adjectives
as descriptors when asking for their evaluation. This may
not be important for publishing the article, but it might be
of particular interest to those companies, health agencies, or
people who are thinking of using the results of the article.
Conclusion
This research is quite fun to do. It’s easy to implement, and any extra effort to do it in a realistic and
compelling context and with a realistically presented product will make it a lot more useful. For instance,
we’ve done taste tests at Halloween parties, progressive dinners, and at an Academy Award party.
Many of these sensory articles are often conducted and targeted to sensory journals. Yet these
issues are of huge interest to public health, parents, psychology, marketing, and food technology.
Redirecting your research toward one of these new areas could be tremendously rewarding.
Brian Wansink 2019 - 7
CAN TASTE PROFILES PREDICT
FOOD PREFERENCES?
You know your new special friend loves red wine and hates sprouts. If you order the
anchovy pizza, will this be your last date, or will you hear start to hear wedding bells?
Being able to predict what new foods people will love based what foods they
currently love would be useful. You wouldn’t make a new recipe no one else eats,
and you might be able to better predict whether you’d like a restaurant menu
item before you order it. Being able to make these predictions would also be
useful to food companies because they would know who might be most likely to
try their new meatless meat product, or their pumpkin chutney canned soup.
The Original Findings
This was based on survey research and published 2003.1 The abstract describes the findings:
The paper was retracted at the request of the Editors on the grounds
of unreliable data in Table 1 and duplication of text in the results and
discussion sections from an earlier publication (B. Wansink and J. Cheong,
Taste profiles that correlate with soy consumption in developing countries,
Pakistan Journal of Nutrition 1: 276; 2002; DOI: 10.3923/pjn.2002.276.278).2
1 Wansink, Brian; Westgren, Randall (December 2003). "Profiling taste-motivated segments". Appetite. 41
(3): 323–327. doi:10.1016/S0195-6663(03)00120-X.
2 https://www.sciencedirect.com/science/article/pii/S019566630300120X
Other Ways to Answer this Question
1. Hypotheses and Extensions: This isn’t a hypothesis type of project as much as it is
an empirical test of concept. In this particular case, let’s take people who really
like a particular product (such as meatless meat, tofu, anchovies, or whatever
our target food is), and see what other clusters of food these people might also
tend to love. Large scale data bases can make this easier to accomplish.
Brian Wansink 2019 - 8
This process would also be beneficial for public health and nutrition. Say that people can be
categorized into four quadrants: 1) People who love vegetables the most, 2) people who love
fruit the most, 3) people who love them both, and 4) people who dislike both of them. Suppose
you know what non-vegetable foods vegetable lovers (group 1) love more than fruit lovers. This
can help one know what types of foods to direct toward vegetable lovers vs. fruit lovers so that
everyone eats more produce, and everyone is healthier. If you knew your child was more of a
vegetable lover than a fruit lover, you’d know what new foods you could direct them toward.
Being able to have a basic understanding of food profiling would also be useful to a parent who
is cooking for a picky child or spouse. That is, knowing what taste profile they might belong to
would also give you an idea of what other foods tend to be eaten by similar people in their profile.
In this way their taste preferences can slowly be expanded past their current limited set of foods.
2. New Methodology Ideas: This is basically a statistical exercise, but it needs the right data to be
able to categorize a person as a being a taste champion of the particular target food. The more
fanatical of a fan a person is about the target food, the more they valuable their taste profile
will be in helping generate insights. Because of this, a basic Likert scale for food liking may not
be as discriminating as using multiple scales and combining it with food frequency questions,
and selecting people based on the frequency they eat the target food you are focused on.
Conclusion
There’s a lot of practical promise in being able to develop a common set of 6-8 different
taste profiles for fanatics of a specific target food. It will give lots of insights about cross-
promotions, recipe ideas, potential partners, and gateway paths to adoption.
And by-the-way, if you order the anchovy pizza for your special friend who
loves red wine but hates sprouts, start listening for wedding bells.
This study was a part of a larger mail survey. If people completed the 12-page survey,
they could cash the $3 honor check that
was included.
Brian Wansink 2019 - 9
DO LARGE SERVING BOWLS
MAKE YOU TO EAT MORE?
Suppose you’re at a Super Bowl party and you are surrounded by an endless
supply of snacks. Will you serve and eat more if the snacks are in large bowls or
would you eat more if the same volume of snacks were in twice as many bowls
half that size? This has implications for dieters as well as for health conscious
and thrifty hosts who don’t want to encourage too much festive overeating.
The Original Findings
The original study was based on a field study involving MBA students at a
Super Bowl party in a sports bar in Champaign, IL in 2000. It was published
as a two-page research letter in JAMA.1 Here’s what was found:
The paper was retracted because JAMA asked Cornell to provide an independent
evaluation of this and five other articles to determine whether the results are valid.
In their retraction notice, JAMA wrote, “[Cornell’s] response states: ‘We regret that,
because we do not have access to the original data [original coding sheets or surveys],
we cannot assure you that the results of the studies are valid.’ Therefore, the 6
articles reporting the results of these studies that were published in JAMA Pediatrics,
JAMA, and JAMA Internal Medicine, are hereby retracted”2 (Appendices A-C).
1 Wansink, B; Cheney, MM (13 April 2005). “Super Bowls: serving bowl size and food consumption”. JAMA.
293 (14): 1727–8. doi:10.1001/jama.293.14.1727. PMID 15827310.
2 https://jamanetwork.com/journals/jamapediatrics/fullarticle/2703492
Other Ways to Answer this Question
1. Hypotheses and Extensions: Let’s say that people do eat more from bigger bowls. Do
they know they are doing so? One extension of this would be to intercept people
after they party was over and ask whether they believe the size of the serving
bowl had any impact on how much they served and ate. Causal conversations
with people after studies like this surprisingly seem to suggest they don’t
think the size of a bowl could influence know much they ate, and even when
it’s pointed out, they have alternative rationalizations why they might have
eaten more than average (“I was hungry,” “I didn’t eat lunch,” and so forth).
Forty MBA students at a Super Bowl party in a bar were randomly led to serving
tables of a snack mix that was either presented in 2 large serving bowls (4-liters
total capacity) or 4 medium servings (also with 4-liters total capacity). Those
serving from larger bowls unknowingly served and consumed about 53% more
snack mix, and this was primarily driven my males.
Brian Wansink 2019 - 10
A second interesting extension of follow-up to this would be whether bowl size influences them
more if they are in a bad mood or in a good mood. Major sporting events offer an opportunity to
do this. Knowing which team a person is cheering for can be used to see if happy winners celebrate
more when given big bowls, or whether unhappy losers drown their sorrows in big buckets.
2. New Methodology Ideas: This particular study was conducted in a noisy sports bar under
realistic conditions. Other than being randomly assigned to a serving table and
inconspicuously led to that table, everything thing else was natural. Another approach
would have been to more tightly test this as a lab study than as a field study in a bar.
As a rough guideline, most of these field studies indicate that people serve and eat around 20%
more from larger containers and plates. Seldom more than 30% and seldom less than 10%.
But scholars have also hypothesized that bowl and plate size effects are less strong (or even
nonsignificant) when conducted in lab settings, and systematic meta studies have also
shown that this effect is much stronger in the field than the lab. Yet what has been missing
to date is a very explicit test of a field study versus lab study comparison. An excellent study
of this would be useful in resolving some of the effect size differences in these studies.
Conclusion
Bowl sizes and plate sizes have been a fertile ground for lots of useful studies that have led to
new dinnerware lines, changes in hotel and restaurant chain buffet plates, and eating behavior
changes among dieters. Things are now at a stage when it would be useful to learn what are
the limitations and boundary conditions around using dinnerware to perceptually change
how much is served. Knowing the point at which smaller and smaller dinnerware backfires
or the circumstances when it does and doesn’t work will provide a new level of impact.
Additionally, there might be very practical situations where dishware sizes clashes with a
perception of quality or value. It would be important to identify these because they are a
different type of boundary condition. For instance, serving a 10-oz steak on a 10-inch plate
might make it seem huge compared to when it is served on a 12-inch plate. But is this something
a restaurant should do? That is, does it make the steak look like a better value, or does it
make it look cheap? Answering these questions would have immediate implications.
We followed up this “Super Bowls” study with
other ones at this same sports bar.
We did about eight of these studies in Jillian’s
Sports Bar in Champaign, IL.
Brian Wansink 2019 - 11
WHY DO OVERWEIGHT
PEOPLE UNDERESTIMATE
HOW MUCH THEY EAT?
This is one topic that can help improve the bedside manner of some of the doctors
and dieticians who work with overweight and obese patients. It can also help with
one of the difficulties obese people face when being counseled about their weight.
To help people lose weight to eat better, doctors and dieticians often rely on
food diaries to diagnose problem areas and make recommendations. Whereas
most people are inaccurate about estimating the calories and portion sizes
they eat, the heavier a person is, the more they underestimate what they eat.
Some health professionals have alleged that these inaccuracies are due to
heavy people being uninformed, in denial, lying, or other hurtful reasons.
There may be a better and more empowering explanation. As things get larger,
we tend to underestimate them. Heavy people may simply underestimate
how much they eat because the quantities are large. Skinny people might be
equally inaccurate if they ate huge amounts (like on Thanksgiving). If this is
true, it’s meal size, and not body size that explains these inaccuracies.
The Original Findings
This 2006 paper involved both a lab study and a field study involving fast food
restaurants.1 Similar findings had been robust in a number of pilot studies, and here is
the abstract which shows the results of the two studies that were finally published:
The paper was retracted because . . . “Annals contacted the authors to inquire
whether they had concerns about its validity. In his response, the coauthor,
Pierre Chandon, reported that age was not a variable collected during the study.
Yet, the article reports mean age for male and female participants. In addition,
the editors identified no age variable in the data files nor on the sample paper
Brian Wansink 2019 - 12
1 Wansink, Brian; Chandon, Pierre (2006-09-05). “Meal Size, Not Body Size, Explains Errors in Estimating
the Calorie Content of Meals”. Annals of Internal Medicine. 145 (5): 326–32. doi:10.7326/0003-4819-145-
5-200609050-00005. ISSN 0003-4819. PMID 16954358.
2 https://annals.org/aim/fullarticle/2717783/notice-retraction-meal-size-body-size-explains-errors-
estimating-calorie
Other Ways to Answer this Question
1. Hypotheses and Extensions: This estimation bias has been shown
to be very robust in our prestudies and pilot studies. A
powerful extension to this would be to show how this bias
exists across a large range of populations and expertise.
We might think that “education” would be one way to reduce this
bias, but a couple pilot studies showed that it wasn’t that effective
and that even dieticians significantly underestimated calories
once the foods or portions got outside of a narrow range. One
way to more convincingly show health professionals that “more
education” is not the answer to this is to compellingly conduct
a study that shows how even they – as educated experts – are
biased in their estimates of large portions of food and calories.
We also don’t know how this news, or a new counseling approach
will influence heavy patients. Suppose a health professional tells an overweight
person that everyone has these estimation biases, and they are not to be blamed.
It might give them more resolve to eat better (and not give up), or it might deflect
their sense of responsibility. Following-up will provide useful course correction.
2. Outreach Suggestions: What would make these findings even more useful would
be if there were a reliable way to come up with some easy guidelines that could
help a health professional or consumer self-correct their estimate. This would
be an alternative to the seemingly ineffective reliance on “more education.” If
these guidelines could be calibrated and tested, they could be very useful in a
wide range of weight loss contexts – both with health practitioners and dieters.
Conclusion
As mentioned earlier, the findings of this research could help improve
the clinic or “bedside manner” of some of the doctors and dieticians who
work with overweight and obese patients. It can also help with one of the
difficulties obese people face when being counseled about their weight.
The insights of how to improve weight counseling could be put into use immediately.
data forms that Dr. Chandon provided in response to our query. In light of the
reporting of a variable (age) that seems not to have been collected, the editors
cannot be confident in the integrity of the work reported in this article.”2
If you run a study in a food court, you
can get exact calories of what people
ordered because it is usually posted.
Brian Wansink 2019 - 13
CAN YOU CONFIRM
THE SWEET TOOTH
HYPOTHESIS?
Being able to predict what a person likes to eat would be practical. A parent could
guide kids to unfamiliar healthy foods they might like, and companies could
advertise products to the most interested shoppers. One way to explore this is to
see if we can predict whether people are more likely to prefer new fruit dishes more
than vegetable dishes based on whether they prefer sweet snacks or salty snacks.
Although some people love all foods, knowing whether a child or spouse has a
natural leaning to one versus the other can prevent you from fighting an uphill battle
to get them to eat broccoli if that’s not how their taste preferences are wired.
The Original Findings
This paper included two studies. One study involved a large USDA database (similar
to NHANES data), and the other study involved a mail survey we conducted at the
University of Illinois at Urbana-Champaign. The resulting paper was published
as a four-page research note.1 Here is the abstract of what was found:
1 Wansink, Brian; Bascoul, Ganaël; Chen, Gary T. (2006-07-01). “ The sweet tooth hypothesis: How fruit
consumption relates to snack consumption”. Appetite. 47 (1): 107–110. doi:10.1016/j.appet.2005.07.011.
ISSN 0195-6663. PMID 16574275.
2 https://www.sciencedirect.com/science/article/pii/S0195666306000250
Other Ways to Answer this Question
1. Hypotheses and Extensions: Given the practical value of this question,
the most useful study that is needed is one that uses large-scale data
bases to determine if there are reliable clusters of taste preferences.
The first step in doing so would be to come up with some general food
categorizations that make sense and are easy to operationalize or code.
The paper was retracted “at the request of the editors and the authors
due to substantive errors in the reported methods.”2
Brian Wansink 2019 - 14
Sweet versus salty preferences would be one categorization,
but it’s not the only categorization method in the original
study. Although some people like both, a good percentage
had leanings more toward one or the other. Other efforts
to categorize food and look for cluster commonalities is by
their bitterness, their calorie content, or their hardness.
Another approach to this would be to make it purely
empirically exploratory. Conduct statistical cluster analyses
on consumption frequencies of certain categories of foods
and see how they correspond with consumption frequencies
of seemingly unrelated categories that actually do have an
underlying explanatory relationship (fresh fruit consumption
clusters and sweet snack clusters, for example). This could
be followed up with a shorter questionnaire that explicitly
asked people about the linkages between these different
foods clusters by using attitude or preference scales.
Something that would also be worth exploring is how these
cluster preferences different between cultures. That is,
what are the unexplained clusters of food preferences of
Europeans vs. Asians vs. Middle Easterners, and so forth.
2. New Methodology Ideas: Using large-scale data bases from the USDA is a great
way to start this. Following up any promising findings by using a more
confirmatory survey could take this research to the next level of contribution.
In our pilot studies, a lot of general tastes appear more fixed than fluid by the
time a person is a young adult. Because of this, if you decide to do a confirmatory
study that explicitly asks people about the linkages between the food clusters
they like, it would perfectly legitimate to do this with college students.
One methodological caution is to try to make sure the analyses are done
within homogeneous groups. If you are not explicitly making the comparisons
between demographic subgroups (mentioned above), they’ll need to be
controlled for in how they are sampled (or statistically analyzed).
This is a long questionnaire. We also
did this follow-up, where we used their
real snack selection during the survey to
validate their profile.
Conclusion
A compelling discovery in this area would have a very useful ripple effect across academia,
practice, and everyday life. It would help academics control for noisy variance when
analyzing taste (by asking control questions in the survey). It would help practitioners
better determine what segments of people might most like a new product. And it would
help us be savvier in how we feed our kids and how we explore new taste adventures.
Brian Wansink 2019 - 15
DO “CLEAN PLATE” KIDS
TURN INTO OVEREATING ADULTS?
Kids can be really smart. That’s why some of our best ideas as
parents back-fire. Take the Clean Plate Club, for example.
Some parent insist they kids clean their plate. Other parents are more relaxed
about it. If a parent regularly insists their child clean their plate, will it alter the
amount of food a child decides to serve themselves? Maybe they serve less of new
foods because they don’t want to have to eat them if they don’t like them. Or maybe
they serve themselves a lot more of the unhealthy and indulgent foods they love
because they know that once they get on their plate, they’ll be able to eat them all.
The Original Findings
This was published in 2008 as a two-page letter in the Pediatric Forum of
what is now JAMA Pediatrics.1 It was based on a lab study with preschoolers.
There’s no abstract to the paper, but here is what we found.
1 Wansink, B; Payne, C; Werle, C (October 2008). “Consequences of belonging to the “clean plate club””.
Archives of Pediatrics & Adolescent Medicine. 162 (10): 994–5. doi:10.1001/archpedi.162.10.994. PMID
18838655.
2 https://jamanetwork.com/journals/jamapediatrics/fullarticle/2703492
Other Ways to Answer this Question
1. Hypotheses and Extensions: Although a lot of people think they are members of
the Clean Plate Club, when this article was first published there wasn’t a lot
of research on it. One set of questions that would be promising to explore
The paper was retracted because JAMA asked Cornell to provide an independent
evaluation of this and five other articles to determine whether the results are valid.
In their retraction notice, JAMA wrote, “[Cornell’s] response states: ‘We regret that,
because we do not have access to the original data [original coding sheets or surveys],
we cannot assure you that the results of the studies are valid.’ Therefore, the 6
articles reporting the results of these studies that were published in JAMA Pediatrics,
JAMA, and JAMA Internal Medicine, are hereby retracted”2 (Appendices A-C).
Sixty-three preschool children were asked to indicate how much of a sugared
cereal they wanted scooped into their bowl (either 16- or 32-oz) for their
morning snack. Children with larger bowls requested nearly twice as much
cereal and the volume they took was correlated with their parent’s answer
to the scaled question (1-9) such as “I tell my child to clean their plate.”
Brian Wansink 2019 - 16
are those which would examine the long-term consequences
forcing kids to clean their plate. That is, maybe they learn to
take smaller portions of healthier foods and larger portions
of desserts. Maybe they grown up to be a heavier adult.
Maybe they grown up to be less adventurous eaters because
they are afraid to try new foods for fear that they would
have to finish all of them (just like they did as a child).
2. New Methodology Ideas: Many of the basic questions asked
above could be at least preliminarily examined by simply
using surveys. It’s not always that compelling, but in an area
as under-researched as this, it will give some toeholds for
subsequent researchers who want to examine it more causally.
To this end, there can be causal experiments done with children,
and the one here represents a gateway into doing so. The idea
would be to look for the behaviors that we think kids from
Clean Your Plate households would demonstrate compared
to those in normal households. After being able to determine
what household a child was from, the study would examine how
much new foods or how much of a favored food they served
themselves and ate when their parents weren’t around. A good
place to do this research would be in a daycare setting.
Conclusion
The Clean Plate Club is something everyone knows about. Doing more research
in this area would have a lot of appeal a lot of immediate applications. Looking
at some of the long-term consequences would be great, but in the meantime,
there’s a lot of useful insights that could be examined immediately.
An effective way to do food studies
with preschoolers is to separate them
from their friends but keep their friends
within view.
Brian Wansink 2019 - 17
HAVE CLASSIC RECIPES
INCREASED IN CALORIES
OVER TIME?
Fast food and restaurants have widely been blamed for the obesity problem: Increasing
portion sizes and more calories. We wanted to know if the same thing was going on in
homes. Are portion sizes of classic recipes getting larger and more caloric over time?
One way to investigate this would be to examine how the same recipes have
changed over time. Have the recommended serving sizes of chili gotten bigger and
do they have more meat or butter? If this is so, the obesity problem might not just
be a problem with restaurants, but it might also be what we do in our homes.
The Original Findings
This was published in 2009 as a one-page letter to the editor in the Annals
of Internal Medicine.1 It was based on secondary analysis. There is no
published abstract, and here is a summary of what was found:
1 Wansink, Brian; Payne, Collin R. (2009-02-17). “The Joy of Cooking Too Much: 70 Years of Calorie
Increases in Classic Recipes”. Annals of Internal Medicine. 150 (4): 291–2. doi:10.7326/0003-4819-150-4-
200902170-00028. ISSN 0003-4819. PMID 19221391.
2 https://annals.org/aim/fullarticle/2717784/notice-retraction-joy-cooking-too-much-70-years-calorie-
increases
Other Ways to Answer this Question
1. New Methodology Ideas: This is an interesting question
with immediate implications for people who cook at home.
There’s a number of other ways it can be tackled.
This paper was retracted because...“Annals contacted the authors to inquire
whether they had concerns about its validity. The contact information we had for
Dr. Payne was no longer current, and we were unable to locate current contact
information. In response, Dr. Wansink reported that he was also unable to contact
Dr. Payne, but Dr. Wansink provided a reanalysis of the data and reported that, “The
files we reran gave the same conclusions, but different numbers in the table.” In
fact, almost every number was different from those in the published article, many
substantially so. In light of the inability to reproduce the published results, the
editors cannot be confident in the integrity of the work reported in this article.”2
To make a historical comparison of the same recipes over time, the
sample frame was all of the recipes in the Joy of Cooking that were in all
seven editions and which did not change their name. The total calorie
content was determined for each recipe, along with the calorie content per
recommended serving size. The results showed the calorie density in these
recipes increased by 35%, and that this was driven by the inclusion of more
caloric ingredients (butter, nuts, sugar) and by larger serving sizes.
Brian Wansink 2019 - 18
a. Analyze a wider range of cookbooks. In this particular study, we focused on one cookbook, the
Joy of Cooking, because it had been around for a long time, had a large number of editions,
and all of the editions were available through interlibrary loan. A longitudinal comparison
with other cookbooks (such as Better Homes and Gardens Cookbook) could broaden the
generalization of this (although there are fewer editions over a shorter time period).
b. Analyze all of a cookbook’s recipes in a given year compared to other years. The published study
only focused on the recipes that were in every single edition of the Joy of Cooking and
which did not undergo a name change. Another way to conduct this research would be
to analyze all of the recipes in a cookbook and compare it with all of the recipes in later
editions. This wouldn’t need to be done for every edition, because the original analysis
showed there were big inflection points (such as after WWII and in the mid-60s) that created
much of the change. Looking at three or four editions would probably be sufficient.
c. Focus on food-type trends. The increasing trend in portion sizes and calorie levels over the years
has only been examined across all foods. It may be that these trends are more dramatic in
some food groups than others. For instance, they may not have changed much in vegetable
side dishes or in salads. However, they may have changed more in entrees or desserts
where either recommended serving sizes have increased or where the addition of more
or different ingredients has happened, such as more butter, sugar, nuts, and so forth.
d. Ask people. The first three new method suggestions still used secondary analysis – serving sizes
and calorie levels of articles published in cookbooks. Another way would be to interview or
survey a population of people who cook at home. Their cooking behavior could then be assessed
longitudinally by asking a series of questions as to how their cooking has changed. Their cooking
behavior could also be compared cross-sectionally by comparing the cooking habits of older
people (which may have been habitually unchanged for some time) with those of younger people.
Topics to compare would be ingredient use, quantities, subjective serving sizes, and so on.
2. Carefully Coding and Estimating Calories: Three of the four method variations are
focused on each recipe’s calories per serving. While serving sizes are usually stated in
the recipes (such as “serves 6”), calorie counts per serving are not stated (except in some
recent cookbook editions). For this reason, it will be important to hire two well-trained
and credible people to do this and cross-reference their work. We used a dietetics intern
and a nutrition major but using two Registered Dieticians would have been even better.
Conclusion
This is a fun article and it has useful implications to a home cook: Using an
older cookbook or serving half of the dish and freezing the other half would be
two easy changes a food loving cook might want to try out for size.
Choosing the
sample frame
and calculating
calories per
serving are the
two biggest
parts of this
research.
Brian Wansink 2019 - 19
WILL YOU HATE THE
FOOD YOU EAT DURING
BAD EXPERIENCES?
You and the love-of-your-life used to eat Thai food every weekend . . . and then she
dumped you. Do you still like Thai food? Perhaps a great experience or a horrible
experience that is going on when you are first exposed to a food, can engineer
whether you will like the food in the long-term. If so, giving your kids carrot
sticks at Disney Land and not cotton candy could be doing them a much longer-
term favor if they associated vegetables with the “Happiest Place on Earth.”
This started out as “hobby research.” When I was growing up, there were
a number of WWII veterans (US) in our neighborhood, and I always found
it interesting to talk with them. In one set of conversations, I found it
curious that some Pacific vets hated rice and others loved it. In contrast,
European vets seemed to be basically indifferent about it. Hmmm . . .
The Original Findings
This study was originally published as a three-page research note.
It was based on a large-scale mail survey we did of WWII veterans
(US) in August of 2000.1 Here’s the abstract of what we found:
Other Ways to Answer this Question
1. New Methodology Ideas: A useful context to look at this would be in the context
of examining foods that people really seem to hate. Doing a qualitative set of in-
depth laddering interviews would help see if there are some of these reasons that
might be connected to childhood experiences. The key in doing this research is you
have to have them focus on something that they hate that most other people like.
The paper was retracted “at the request of the Editors on the grounds
of serious errors in the description of the method and duplication
and errors in the raw data as identified by the authors.”2
1 Wansink, Brian; Van Ittersum, Koert; Werle, Carolina (2009-06-01). " How negative experiences shape
long-term food preferences. Fifty years from the World War II combat front". Appetite. 52 (3): 750–752.
doi:10.1016/j.appet.2009.01.001. ISSN 0195-6663.
2 https://www.sciencedirect.com/journal/appetite/vol/131/suppl/C
Brian Wansink 2019 - 20
If you were to take this qualitative and narrative approach, the research
could be really interesting and filled with great examples. It would probably
necessitate you partnering with a qualitative researcher (like a cultural
anthropologist) if you wanted it to make sure the details were nailed down.
A second way to tackle this question would be to find a common event that
people could very clearly code as either a good experience or a bad experience
and where they are experiencing a new food for the first time. WWII and rice
were one example, and something similar could be done with other veterans
in other theatres (such as the Mideast). Still another context would be with
vacationers who either had a great time or a terrible time on their vacation when
they were being exposed to a different food. For instance, two different people
on a cruise ship to the Caribbean or South American might have very different
feelings about plantains depending on whether were seasick or weren’t.
2. Publishing and Outreach Suggestions: Being able to quantitatively show
there are better situational memories associated with some favored foods and
that there are worse situational memories associated with some unfavored
foods could be easy if you only use a questionnaire However, this will be a
whole lot more interesting if you can find a specific event that is interpreted
either positively or negatively by the same people at the same time (e.g.,
military service, a seasick cruise, etc.). Also, getting lots of examples of where
and how this is played out will give the article a lot more interest, or will at
least give you useful examples when sharing your implications with others.
Conclusion
There’s interesting news-you-can-use here that may initially seem nonobvious.
Suppose there is a tendency to get your feelings for food tangled up with
how much you’re enjoying the moment when you eat them. If this is true,
you might be able to orchestrate healthier food preferences for both you
and your kids -- like making their annual birthday celebration be all-you-
can-eat watermelon party instead of an ice cream sundae buffet.
Most of the
research team
on this WWII
survey project
were second
year MBA
students.
Brian Wansink 2019 - 21
CAN BRAND LOGOS
ENCOURAGE KIDS TO
EAT HEATHY FOODS?
Brand names and logos are used to sell cookies and candy. Can they also be
used to sell more fruit by making fruit seem more hip, interesting, or tasty? If
so, instead of banning branded products or logos in school cafeterias, it might
be better to redirect the branding and logos to the healthier products.
The Original Findings
This 2008 study was published as a two-page research note in what is now
JAMA Pediatrics in 2012.1 It involved a week-long study with Head Start
preschoolers. There’s no published abstract, but here is what was found:
1 Wansink, Brian; Just, David R.; Payne, Collin R. (1 October 2012). “Can Branding Improve School Lunches?”.
Archives of Pediatrics & Adolescent Medicine. 166 (10): 1–2. doi:10.1001/archpediatrics.2012.999. PMID
22911396.
2 https://jamanetwork.com/journals/jamapediatrics/fullarticle/2659568
This paper was retracted because “Following the notice of Retraction and
Replacement, the funder of this study informed us of another important
error. We had erroneously reported the age group as children ranging
from 8 to 11 years old; however, the children were 3 to 5 years old…
“Given this additional substantial error in reporting the correct ages of the children
and the inadequate oversight of the data collection and pervasive errors in the analyses
and reporting, the editors have asked that we retract this article. We regret any
confusion or inconvenience this has caused the readers and editors of the journal.”2
The study involved 208 preschool children in Head Start afterschool
programs (most of which met in elementary schools). Over the course of
a week, children were presented the choice between apples or cookies
that were either unbranded or branded (with an Elmo sticker) in different
combinations on different days of the week. Having a branded Elmo
sticker on an apple greatly increased the likelihood children would
select it, but the same sticker had no impact on cookie selection.
Other Ways to Answer this Question
1. Hypotheses and Extensions: One of the reasons that branding helps increase
fruit selection so much more than cookie selection is that most kids
naturally love cookies – even without a brand. Therefore, there’s not much
higher their likelihood of selection can go. It’s reached a ceiling.
From a nutrition or public health standpoint, one immediate set of studies
that could be conducted would be to examine this with different ages of
students (toddler, preschool, and elementary students) to see if this is
Brian Wansink 2019 - 22
differentially effective at some ages than others. Also, it could
be examined whether different types of stickers or logos
(familiar vs. unfamiliar; colorful vs. less colorful) are more
effective with some ages or genders than with others.
From a psychology standpoint, what would most interesting would
be to better understand why we might expect results such as these.
Seeing a brand – such as an Elmo logo – on an apple might make
a child take it simply because it looks different or curious. But it
might make someone take the apple because they think it might
taste better than an unbranded apple. If taste expectations can
bias real taste experiences, it might even end up being that seeing
a brand sticker on a piece of fruit, not only leads more people
to take the fruit, but it also makes them think it tastes better.
2. New Methodology Ideas: This study used a within-subject design
and although within-subject designs can control for a lot of
factors, they also come with another host of problems such as
reactivity. This can be especially concerning if the experiment
seems too artificial or fake. An opposite approach to this would
be to use a between-subject design and to rotate the four different
conditions (apple x cookie; branded x unbranded) across these
schools. Yet this seems like it would be way too much overkill to
answer a fairly simple question. In addition, it potentially suffers from the noise
of a bandwagon effect. A child may be more likely to take the same item his friend
ahead of him took, regardless of what the food or branded condition was.
An alternative to either might be to rotate conditions within one school and to have
children make their selection between the apple and the cookie alone as they came
out of the lunch line (or during a break). On one day each week, the combination
of choices could be rotated, and the spacing out would probably nullify reactance,
but the context would still be very real. Setting up the study in this way would
also allow to ask the child a couple quick questions after they selected the item.
Conclusion
Over the last few years there have been some promising steps in this direction of trying
to brand fruits. McDonald’s use of Cuties Mandarin Oranges is a one example of the
promise that smart branding can have for fruit.
In order for this to become more widespread, we can try and imagine what type of
research would be most useful in helping inform this trend:
• What ages and gender of kids are most influenced by branding?
• Do colorful but unfamiliar brands or images work just as well as familiar ones?
• Does branding make kids believe the branded food tastes better?
Some of these questions are the ones already noted above. We need to be mindful that
the more realistically our studies are, the more they are likely to be compelling to the
people making these decisions to brand healthy foods.
This prestudy showed that even putting
stickers in front of foods, such as in a
buffet line or on a platter led kids to
choose healthier foods.
Brian Wansink 2019 - 23
DO SHORT-TERM FASTS
LEAD TO LONG-TERM
WEIGHT LOSS?
Short-term food fasts are the popular weight loss rage. But what
happens when they’re over – do you reset your diet down to a more
mindful one, or do you binge like a 12-year-old on Halloween?
One of the interesting twists on this might not be the obvious issue about calorie
replacement. Instead, a cool question is what types of foods is someone going to be
most drawn to when they are coming off of a fast – sweets, meats, carbs, or treats?
The Original Findings
This was published as a three-page research note in 2012, and there is no
abstract.1 It was based on a lab study that observed what people ate at a
buffet after they had been fasting for 18 hours. Here is what was found:
1 Wansink, B; Tal, A; Shimizu, M (25 June 2012). “First foods most: after 18-hour fast, people drawn
to starches first and vegetables last”. Archives of Internal Medicine. 172 (12): 961–3. doi:10.1001/
archinternmed.2012.1278. PMID 22732752.
2 https://jamanetwork.com/journals/jamapediatrics/fullarticle/2703492
The paper was retracted because JAMA asked Cornell to provide an independent
evaluation of this and five other articles to determine whether the results are valid.
In their retraction notice, JAMA wrote, “[Cornell’s]response states: ‘We regret that,
because we do not have access to the original data [original coding sheets or surveys],
we cannot assure you that the results of the studies are valid.’ Therefore, the 6
articles reporting the results of these studies that were published in JAMA Pediatrics,
JAMA, and JAMA Internal Medicine, are hereby retracted”2 (Appendices A-C).
128 staff and students were recruited for a free lunch. Half were asked
to not eat for 18 hours prior to that lunch. Upon arriving, they were
offered two different starches, two vegetables, and two proteins. Those
who had been fasting were more likely to start their meal with the
starches. Moreover, whatever food a person ate first, the more they
were likely to consume over 40% more of it relative to other foods.
Other Ways to Answer this Question
1. New Methodology Ideas: Using a limited buffet context worked really well
for controlling what people could eat. Because we used only two types of
vegetables, starches, and proteins (six total items), hidden scales could
be put beneath each of the serving bowls (under the tablecloth), and how
much each person takes can be unobtrusively recorded. Another pair of
foods that could be included would be desserts and maybe even salads.
How much people eat of a particular item can be determined by subtracting
how much food is left on their plate from the initial amount they took.
Brian Wansink 2019 - 24
Using a between-subjects buffet study has disadvantages because
it doesn’t control for lots of factors. There are lots of potentially
interfering stimuli (such as how much the person ahead of
you took), but it was a realistic intervention to see if anything
notable would be worthy of further future investigation. Similar
findings could be followed up using within-subject lab studies.
Additionally, although measures of each diner’s preferences of
the different foods was taken, these were not used as covariates
in any analysis. That too would be useful to do in future studies.
2. Measurement and Compliance Suggestions. Asking someone to fast
for 18-hours has compliance challenges. That is, people don’t
like to do it. We had done some prestudies with this, and one
key learning was that you need to be really thoughtful in the
instructions you give to those people in the two conditions. You
also have to be careful, so they don’t behave too unnaturally
when they start eating. The concern would be that they are overly
conscious of what they take and how much because they believe
they are being closely observed. One way to decrease this potential
reactivity is to give both groups an unrelated task to complete.
In one case, we tried giving them a concentration test before
lunch so they would focus more on the concentration test and
not feel they were being as closely scrutinized during lunch.
It’s important to get more accurate assessments of how both
groups behaved in the prior 18 hours, including what they
ate or drank and whether they exercised. There are a number
of different ways to assess whether someone has eaten in
the past 18-hours (or whether they followed the directions
they were given). It’s good to include at least two converging
measures to better ensure that the fasters actually fasted.
In one of our prestudies we had people fast for 24 hours, but that was too long. We had too
low of a compliance rate, and we aren’t convinced that someone who’s 6 hours hungrier is
going to act dramatically different than someone who’s fasted for a more doable 18 hours.
Last, some people believe a fast should involve no liquids, but we decided it was
unnecessary for this research question. You can keep a fast limited to only water, but in
another study, we’ve done, we found that allowing people to drink non-caloric drinks
(such as coffee or diet soft drinks) seemed to increase compliance and happiness.
Conclusion
This issue of fasting is hot. While much fasting research focuses on the calorie compensation,
there are so many more rich and interesting questions that can be asked.
Because of the prevalence of fasting, using a realistic research context and a useful
research question would be of most value. From an immediate impact standpoint,
this type of research would be widely welcomed by many people.
To help prevent the fasters from being
too self-conscious about what and how
much they ate, we have scales under the
table cloth. They are read off the three
monitors on the floor.
Brian Wansink 2019 - 25
WILL COOL NAMES
LEAD KIDS TO CHOOSE
HEALTHY FOODS?
Are kids more likely to choose and eat healthy foods when they’re
named X-Ray Vision Carrots or Silly Dilly Green Beans? Even if they
do, will the name make them like the food more and will it make them
continue to select it and eat it after the name changes back?
Adding new words to restaurant menu items makes adults more likely to
change their food order. If something like this could work with kids in cafeteria
lines, it would be an easy and healthy change for schools to make.
The Original Findings
These two studies were published as a three-page Short Communication
in Preventive Medicine In 2012.1 It is based on a lab study and a 1-month
field study in elementary schools. Here is what was found:
1 Wansink, Brian; Just, David R.; Payne, Collin R.; Klinger, Matthew Z. (October 2012). Attractive names
sustain increased vegetable intake in schools”. Preventive Medicine. 55 (4): 330–332. doi:10.1016/j.
ypmed.2012.07.012. PMID 22846502.
2 https://www.sciencedirect.com/science/article/pii/S0091743512003222
The paper was retracted “at the request of the Editor and with the authors’ agreement
because additional corrections regarding funding attribution were brought to the
journal’s attention after it published a Corrigendum. The need for further amendments
to an article whose contents are no longer a valid description of the methodology
and findings of the original research record would have been detrimental to the
opportunity for knowledge translation that the original 2012 article was intended to
provide. The authors have been offered the opportunity to incorporate all necessary
amendments and disclosures into the manuscript and resubmit it for consideration
for eventual publication in Preventive Medicine, subject to peer review.”2
Other Ways to Answer this Question
1. Hypotheses and Extensions: Two of the bigger questions related to
how this works with young children are these: 1) What ages and
gender are most influenced by attractively named foods, and 2)
Does the effectiveness of attractive names begin to wear off?
Brian Wansink 2019 - 26
There seems to be early evidence that this works even with preliterate kids (when
the labels are read to them), but it’s not clear who these attractive names influence
the most. Since many elementary schools have their children go through the
lunch lines by grade, this would be an easy question to ask in field studies.
The issue as to whether the effectiveness wears off is an interesting one
that’s been raised a number of times in the past. Our general observations
are that the effectiveness of an intervention like this drops off as much as 30-
40% in 2-3 months. The good news, however, is that it seems to reset itself
after summer break (and even after winter break to a lesser extent).
2. Methodology and Outreach Suggestions: Although this seems like it would be an easy
intervention, there is still resistance by school cafeterias who believe it would be too
difficult. An effective research project in this area could be one that not only show
the promise of naming, but which also show that it can be easily implemented.
One way to do so would be to conduct this study by having a high schooler
do it (as an Eagle Scout project, or as a school project). By having them in
charge of implementing and tracking the study, it would show that “it’s so
simply a high schooler could do it.” If that person does a good job, they might
also become an author on the paper (like the fourth author on this paper).
Conclusion
Using attractive or cool names is also worth experimenting with at home with your
own kids. Instead of plopping the vegetables in front of them and expecting them to
clean the bowl, take 5 seconds to come up with an interesting or silly name for what
you’re serving, or a description of where it came from, or how it was made. You can
creatively do anything that builds anticipation, sensory suggestiveness, or engagement.
At best, they’ll eat it and like it a little better. At worst, your teenager will
roll their eyes and think to themselves, “At least my goofy dad tries.”
Almost any type
of catchy name
increases a kid’s
interest in healthy
foods, according to
pilot studies and
follow-ups.
Brian Wansink 2019 - 27
DO HUNGRY SHOPPERS
BUY MORE OR JUST
BUY WORSE?
The hungrier you are the more food you buy, right? Maybe not. Being hungry
might lead you to buy ready-to-eat foods that you can quickly and conveniently
eat – like in the car on your way home. However, it might not lead you to buy
more total food (such as foods that can’t be eating quickly, like vegetables and
meat). If true, the advice to dieters and fasters is not to avoid shopping when
hungry so you’ll buy less. Instead, it’s to avoid shopping when hungry if you
can’t discipline yourself to buy better foods (the non-ready-to-eat foods).
The Original Findings1
This was published in 2013 as a three-page research letter, and there is no abstract.
It was based on a lab study and a field survey conducted with shoppers after they
completed their grocery store check-out. Here’s a summary of the findings:
1 Tal, A; Wansink, B (24 June 2013). “Fattening fasting: hungry grocery shoppers buy more calories, not more
food”. JAMA Internal Medicine. 173 (12): 1146–8. doi:10.1001/jamainternmed.2013.650. PMID 23649173.
2 https://jamanetwork.com/journals/jamapediatrics/fullarticle/2703492
The paper was retracted because JAMA asked Cornell to provide an independent
evaluation of this and five other articles to determine whether the results are valid.
In their retraction notice, JAMA wrote, “[Cornell’s]response states: ‘We regret that,
because we do not have access to the original data [original coding sheets or surveys],
we cannot assure you that the results of the studies are valid.’ Therefore, the 6
articles reporting the results of these studies that were published in JAMA Pediatrics,
JAMA, and JAMA Internal Medicine, are hereby retracted”2 (Appendices A-C).
There was both a lab study and a shopping study. In the lab study, people who
had been instructed to not eat 5 hours before the study chose more higher
calorie snacks, but no more of the healthier snacks (than those in the control
condition). Consistent with this, a shopping survey showed people shopping
late in the afternoon (4-5 hours after last eating a meal) tended to buy a
total basket that was less healthy than those shopping right after lunch.
Other Ways to Answer this Question
1. Hypotheses and Extensions: This notion that hungry shoppers want to buy the
tastiest calories they can quickly buy and eat is compelling. It doesn’t seem like
such a person would leisurely shop the aisles and price-compare frozen meat.
At this point, I think there are two big extensions that can be made. One is
to combine the hunger and time element. If people shop differently when
hungry, then people shopping just before lunch (1100) should shop differently
than those shopping just after lunch (100), and those shopping mid-
Brian Wansink 2019 - 28
afternoons should shop better than those in the late afternoon. Now there
are all sorts of other covariates to measure but using time as a surrogate for
hunger will give a better process link and it has much broader implications
for retailers and for informing health-minded consumers when to shop.
The second extension would be to predict the specific types of foods that are most
prone to be purchased by a hungry shopper. Is it cookies, chips, and breakfast
cereal, or is prepared foods, or is it candy at the checkout? These results may not
be important for theorizing, but they are important as implications for dieters.
2. New Methodology Ideas: We thought it was cool to have a lab study that showed
that people didn’t eat more of everything when they were hungry, they just
ate more of what was easiest to eat – carbohydrate-packed snacks.
The best way to tackle this compellingly might be to forego any lab study and
do a really great scanner data study in grocery stores. Taking multiple stores
and analyzing shopping baskets content by time (1100ish vs. 100ish or 200ish
vs. 400ish) would be best. Then a field survey of shoppers could be done in
one or two grocery stores as a manipulation check to confirm that their hunger
corresponded to those time periods. In addition, some self-report process
questions can help confirm whether they shopped differently than usual.
3. Publishing and Outreach Suggestions: The results of this are of great interest to
shoppers who want to eat healthier, but they are also of interest to retailers.
A scanner data study (combined with a small survey of exiting shoppers)
would make this a useful public health article or marketing article. Adding
the real-world advice of what categories are most important to avoid when
hungry would make this a useful article to lots of different people.
Conclusion
Having some scanner data analysis prowess would make this a relatively easy
hypothesis to examine by using shopping time as a surrogate for hunger. A
short in-store survey for a second group of shoppers would take this from a
useful effects article to a very useful and memorable recommendation.
We also ran a
food sampling
study with hungry
shoppers, but it
made this paper
too long and didn’t
add anything new.
Brian Wansink 2019 - 29
DOES PREORDERING LEAD
TO HEALTHIER LUNCHES?
You might heroically plan on eating a healthy salad for lunch, but when noontime
rolls around, the French fries will smell too good to pass up. If you had to pre-
order your lunch when you first got to work, would you eat better? If so, work
cafeterias and school cafeterias could offer a preordering option. This way they
could help their employees or students eat healthier and less indulgent lunches.
The Original Findings1
The original field research was conducted in a public-school
district in the Finger Lakes area of New York. It was published as a
two-page research letter, and here’s a summary of the results:
1 Hanks, AS; Just, DR; Wansink, B (July 2013). “Preordering school lunch encourages better food choices by
children”. JAMA Pediatrics. 167 (7): 673–4. doi:10.1001/jamapediatrics.2013.82. PMID 23645188.
2 https://jamanetwork.com/journals/jamapediatrics/fullarticle/2703492
The paper was retracted because JAMA asked Cornell to provide an independent
evaluation of this and five other articles to determine whether the results are valid.
In their retraction notice, JAMA wrote, “[Cornell’s] response states: ‘We regret that,
because we do not have access to the original data [original coding sheets or surveys],
we cannot assure you that the results of the studies are valid.’ Therefore, the 6
articles reporting the results of these studies that were published in JAMA Pediatrics,
JAMA, and JAMA Internal Medicine, are hereby retracted”2 (Appendices A-C).
In the first two weeks of this four-week study, students ordered lunch entrees
as they usually did. In the third week, they pre-ordered their lunch entree
using a paper order form. Longitudinal sales data and intake measures
(inferred through plate waste) showed healthier foods were selected about
twice as often (29% vs. 15%) when students had to preorder their entrée.
Other Ways to Answer this Question
1. Hypotheses and Extensions: This was a small pilot study that has sizable promise.
Two useful extensions would be to a) generalize it to other populations (such as
employees in cafeterias), and b) determine if this only works in the short run (like
for the first couple weeks) or if it can be sustained past the first three months.
Some of our research with other interventions has shown a decay rate of up to
40% over a three-month period unless small variations are made to keep it fresh.
Brian Wansink 2019 - 30
2. New Methodology Ideas: Using a before-after within-subject
study would be one approach that eliminates some individual
variation. However, it would also need a large control group
to not run the risk that something else could influence the
results (weather, midterm exams, other menu changes, and
so on). One way to solve this problem this would be to split
the group in two and reverse the order of the conditions in
each group. That is, one group be a control-treatment group
(no preordering during month1 but preordering during
month2), and the other group be the treatment-control group
(pre-ordering in month1 but no preordering in month2).
It would be great to show how preordering influences how many
calories kids eat, and how it influences whether these calories
are starch calories. This can be done on an individual level by
using the Quarter-plate Method of measuring. Alternatively, if
connecting a student’s plate waste with his student ID number is
too difficult, this can be recorded in the aggregate. At this stage,
knowing if preordering leads to healthier meals is the primary
message that would need to communicate to health-minded
cafeterias. Answering the follow-up issue of who it influences
most can be done with more precision in a follow-up study.
3. Publishing and Outreach Suggestions: A wide range of journals would
find different aspects of this interesting in different ways. Here’s two approaches:
A) Publish a shorter “Effects” or “Outcome” article in a public health, nutrition,
or medical journal, or B) Publish a longer “Process” paper – perhaps with a
preceding lab study, and a follow-up study – in a consumer behavior, economics,
psychology, or marketing journal. If this is as effective as these earlier studies
suggest, I think publishing a shorter piece would get the word out and start
getting these changes made in schools and cafeterias sooner rather than later.
Conclusion
This is a great research question and if the study’s done well, it will have directly
relevant implications for whatever is found. There are two keys to making this
an influential paper. The first key is to do it in a real cafeteria that is really trying
to help people eat healthier. Schools and company cafeterias are two examples,
and a hospital cafeteria would also be great. The second key is to set up a pre-
ordering intervention that is simple and scalable and not overly complicated
or artificial. If simple pre-ordering system is shown to be effective – even if
it’s not 100% perfect – it is likely to make a much more compelling point.
The photos in this photo-based order
form kept deviating too much from what
was actually served, so we used a text-
based order form instead.
Brian Wansink 2019 - 31
DO DIFFERENT TV SHOWS
INFLUENCE HOW YOU EAT?
Eating while watching TV isn’t highly recommended because it’s believed to
cause you to eat poorly. If this is indeed true, it could either be because something
like TV is distracting or because the pacing and stimulation of it speeds up our
eating. For instance, exciting shows with lots of cut scenes or noise might cause
us to eat more because it’s really stimulating, or it might cause us to eat less
than a boring news show because we are more engrossed and distracted.
If a dieter or food-loving person absolutely believes they must,
must, must eat while they watch TV, they might like to know which
types of TV shows don’t lead to regretful overeating.
The Original Findings1
This research was originally published in 2014 as a two-page research
letter in JAMA Internal Medicine. It was based on a lab study conducted
with undergraduates in Ithaca, NY. Here’s a summary of the findings:
1 Tal, A; Zuckerman, S; Wansink, B (November 2014). “Watch what you eat: action-related television content
increases food intake”. JAMA Internal Medicine. 174 (11): 1842-3. doi:10.1001/jamainternmed.2014.4098.
PMID 25179157.
2 https://jamanetwork.com/journals/jamapediatrics/fullarticle/2703492
The paper was retracted because JAMA asked Cornell to provide an independent
evaluation of this and five other articles to determine whether the results are valid.
In their retraction notice, JAMA wrote, “[Cornell’s] response states: ‘We regret that,
because we do not have access to the original data [original coding sheets or surveys],
we cannot assure you that the results of the studies are valid.’ Therefore, the 6
articles reporting the results of these studies that were published in JAMA Pediatrics,
JAMA, and JAMA Internal Medicine, are hereby retracted”2 (Appendices A-C).
Ninety-four undergraduates were shown one of three types of programming:
1) An action movie (The Island), 2) a talk show (Charlie Rose), or 3) the
same action movie with the sound turned off. People watching the action
movie ate more calories than watching the talk show or the action movie
with no volume. This difference was particularly dramatic with males.
Other Ways to Answer this Question
1. Hypotheses and Extensions: There are lots of directions to explore how far this
could be generalized and what types of foods are most susceptible to being
overeaten. As an initial exploration of this, we did this study with small groups
of people rather individually, and this raises a number of key extensions.
These people had a number of snacks sitting in front of them, and there’s a
wide range of ways this could be varied. First, the size and gender composition
of the groups could be varied, but it’s not clear what would happen:
Brian Wansink 2019 - 32
• Larger groups may lead people to eat less because
they are self-conscious, or they might lead
people to eat more if they feel anonymous.
• A mixed gender group might lead women to eat less because
they don’t want to be seen as piggish, but it might lead
guys to overeat to show they are insatiably macho.
• The study can also be conducted within subjects where people
watch their own programming, and the programming
can then be coded and categorizing based on scene
cuts and volume fluctuations. The figure shows how
this varies across a wide range of programming:
Second, a researcher could examine how the distance of the food influences how much is eaten.
Although the general belief would be that food within arm’s length will be eaten more frequently, we
noticed in pilot studies that the farther a person had to reach for food, the more of it they took each
time they served themselves. Also, food placed in front of where they are sitting might also be eaten
more or less often than that on the side since it is more obvious to others that you are taking it.
Furthermore, a useful twist has to do the types of snacks
offered. If watching certain types of TV programming leads
people to not pay much attention to what or how much
they eat, this might be a great way to encourage people to
mindlessly eat the boring healthy foods they don’t typically
eat – like raw vegetables and fruit. This could be easily tested.
2. New Methodology Ideas: Many of the extensions noted above
have different implications for who you recruit, and how
you set the viewing environment up. To seem most natural,
we arranged the furniture in a manner that was typical for
fraternity and sorority TV rooms. This adds realism, but noise.
Another way to set them up is to give everyone their own chair.
Conclusion
Distracting dining is becoming the norm for many people. Preaching snacking abstinence probably
won’t work. Instead figuring out how to minimize the damage would be useful. An even better
idea is to see if this can be used to turn around snacking in a way that encourages more people
to eat healthier snacks instead. If people don’t pay any attention to what they eat as they watch
TV, see if anybody notices when you switch a bowl of baby carrots for their bowl of Cheetos.
A segment from the movie, the Island,
was compared to an interview on the
Charlie Rose show
Brian Wansink 2019 - 33
DO HIGH MENU PRICES MAKE
YOU REGRETFULLY OVEREAT?
Do you eat to get your money’s worth? A gazillion-dollar restaurant industry would
love to know how its menu pricing might change what you think of its food and how
much you eat. This would also be very useful knowledge for savvy dieters who want to
make sure they enjoy their food the most without overeating to the point of regret.
Yet take buffets, for instance. A high-priced buffet might cause you to
overeat in order to “get your money’s worth.” But it might also make you
believe the food is higher quality “because it’s more expensive.” In such a
case, you could overeat, but not regret doing so. Especially when compared
to eating the same amount of the same food at a lower-priced buffet.
The Original Findings1
This article was based on a multi-week lunchtime field study
during the Fall of 2007 at Aiello’s Italian Restaurant (in Whitney
Point, New York). Here’s the abstract of what was found:
1 Siğirci, Özge; Wansink, Brian (19 November 2015). “ Low prices and high regret: how pricing influences
regret at all-you-can-eat buffets”. BMC Nutrition. 1 (1). doi:10.1186/s40795-015-0030-x.
2 van der Zee T, Anaya J, Brown NJL. Statistical heartburn: an attempt to digest four pizza publications from
the Cornell food and brand lab. BMC Nutrition. 2017;354.
The paper was retracted by this journal because “concerns have been raised
after publication with respect to the analysis of the data reported. The
authors have been offered the opportunity to submit a new manuscript
for peer review. The authors do not agree with this retraction.”2
Other Ways to Answer this Question
1. Hypotheses and Extensions: The rationale is this: A high-priced buffet might
cause you to overeat but also enjoy it more if you suggestively believed the
food to taste better. In such a case, you could overeat, but still regret it less
than if you believed the food to be less expensive and less subjectively tasty.
Brian Wansink 2019 - 34
One of many unanswered questions would be whether this
will work with all people in the same way. You might think
dieters would be less influenced, or that this has no influence
on show-off men eating with women. Similarly, it should have
more of an impact on younger people than on older people.
What’s important to nail down is the process of why this
happens. Although the outcome measures (intake and
regret) are measured, those linkages are within the black box
(measures of “getting my money’s worth” and “taste”).
2. New Methodology Ideas: Conducting this study at an all-you-can-
eat buffet far away from a college campus gives it some real-
world appeal. The fewer other foods the restaurant serves, the
easier it will be to track how much is eaten. Although a plate
waste method could be used (like the Quarter Plate Method), we
found that with pizza people either eat it all or they leave only
the crust. If any other method of plate waste measurement is
too intrusive or impractical, simply noting how many crusts they
leave would be worth piloting for practicality and accuracy.
When deciding how to give the discount, there are two
different ways to do so. It can be given as a promotion 50%
off ($4.00 instead of $8.00), which might be less like to evoke
the taste inferences than if they think the real everyday price
is always $4.00. It will also be important to tinker and test
the pricing. This study was run in 2007, so $8 is unlikely
to evoke the “high quality” inference it did back then.
3. Publishing and Outreach Suggestions: As noted in the intro,
the answer to this question has immediate implications
for both buffet restaurants and for dieters. We published
this in a health journal, but a more detailed or multi-study
version could go to a great economics journal. The notion
of these non-optimal consumption consequences has
tremendous implications in behavioral economics.
Conclusion
One little indicator of a worthwhile research question is whether its answer
will resonate with multiple groups of different people. Really nailing down this
research question of whether pricing influences eating, and regret are interesting
to behavioral economists and buffet owners, as much as it is with us buffet lovers.
Aiello’s Restauarnt
The Restauranteurs:
Charlie and Vinnie Aiello
Brian Wansink 2019 - 35
Traumatic experiences — such as combat, living in a conflict country or wartorn
nation, or experiencing a violent crime or natural disaster — change social
relationships and may also influence a life-time of consumer relationships with
brands and shopping. Our focus on this previously overlooked area is centered
on an analysis of the long-term shopping habits of 355 combat veterans. We show
that those who experienced heavy trauma (e.g., heavy combat) exhibited similar
disconnection from brands as others have experienced in social relationships.
They became more transactional in that they were more open to switching brands,
to trying new products, and buying the least expensive alternative (P < 0.01). In
contrast, those who had experienced a light trauma were more influenced by ads
and more open to buying brands even when they cost more (P < 0.00). Trauma, such
as combat, may change one’ s decision horizon. Functionality and price become
more important, which is consistent with the idea that they are more focused
on the present moment than on building on the past or saving for the future.
How do traumatic experiences — such as combat, living in a war-torn nation, or
experiencing a violent crime or natural disaster — change how a person makes
decisions? If counselors or therapists better understood the answer to this, they might
be more effective in dealing with people suffering from post-traumatic stress. To begin
exploring how trauma might influence routine decisions, risk-aversion, and the need for
variety or stimulation, a nonthreatening way to do so is to ask about shopping behavior.
DOES TRAUMATIC VIOLENCE
CHANGE JUDGEMENT AND CHOICE?
The Original Findings
This research was conducted using an omnibus survey of WWII veterans
that was collected in 2000. It compared the shopping habits of combat
veterans with non-combat veterans. Here’s what it found:
The paper was retracted because “Following publication, concerns were brought to
the attention of the publisher regarding the validity of the article's findings. Adhering
to our complaint’s procedure, Frontiers engaged an expert to assess the raw data for
the study. The conclusion from this assessment, supported by the Specialty and Field
Chief Editor, is that there is no empirical support for the conclusions of the article.”2
1 Siğirci, Ozge; Rockmore, Marc; Wansink, Brian (6 September 2016). "How Traumatic Violence Permanently
Changes Shopping Behavior". Frontiers in Psychology. 7: 1298. doi:10.3389/fpsyg.2016.01298. PMC
5012201. PMID 27656152.
2 https://bmcnutr.biomedcentral.com/articles/10.1186/s40795-017-0195-6
Other Ways to Answer this Question
1. Hypotheses and Extensions: There are different types of trauma, and it’s
not clear that all influence people similarly. A useful first step would be
to develop a more thorough taxonomy of trauma, that would not just
focus on the type (war vs. personal violence vs. etc.) but on the length,
the centrality to one’s identity, the locus of control, and so on.
Brian Wansink 2019 - 36
There’s a lot of important interpersonal questions related to
traumatic events (i.e., trust, forgiveness, projection, and others),
and this area of decision making, and judgment is so fertile that
it will have implications for all sorts of general issues, as well
as for some of interpersonal issues. For instance, a traumatic
incident might have notable influences on whether one wants an
exciting, stimulating life of variety (“Eat drink and be merry for
tomorrow we may die”) or a protected and insolated life. It might
influence whether we have strong loyalties (“My band of brothers”)
or no loyalties (“Every man for himself and God for us all”).
Although it is difficult to hypothesize the direction of these
questions, what is important at this stage is to articulate
different ways trauma might bias a person’s big — and
little — decisions and judgments. The little judgments are
important because they can belie bigger biases that have
gone undetected because of the way the questions are asked
or how an apprehensive respondent might answer them.
2. New Methodology Ideas: There are tons of potential biases
with asking people about retrospective memories. Asking
about these biases directly could either retrigger bad
memories or the apprehensive answers mentioned above.
One way to look for these biases is to do so in an innocuous way or
in an unrelated context. Looking for them in the context of how they
shop is an innocuous way in a seemingly unrelated context. The key
would be to ask them questions the showed how their behavior might
have changed. One way is to ask agreement questions (“Are you more
brand loyal than you were xx years ago?” (1 = strongly disagree; 9 = strongly
agree). These answers can be compared with a demographically similar
sample who did not suffer the same type of traumatic experiences.
Also, the questions need to be asked at different levels of abstraction some
specific and some more abstract. This is because we are not well calibrated to know
which questions are too blunt and which are too overly sensitive or prone to error.
Conclusion
While this might appear that it has implications for economics or consumer
behavior, it is really a topic for psychology or sociology journal. It could have the most
opportunity for impact in a health psychology journal or a counseling-related journal.
This study used archived data from WWII
combat veterans, and it used combat
intensity as a surrogate for traumatic
violence.
Brian Wansink 2019 - 37
Appendix A.
JAMA Request to Cornell for Research Validation
Brian Wansink 2019 - 38
Brian Wansink 2019 - 39
Brian Wansink 2019 - 40
Appendix B.
Peer Review Form for Research Integrity Investigation Reports
Brian Wansink 2019 - 41
Brian Wansink 2019 - 42
Appendix C.
Cornell’s Investigation into Possible Errors in Six JAMA Papers
Brian Wansink 2019 - 43
Brian Wansink 2019 - 44
Brian Wansink 2019 - 45
Appendix D.
What are Some Useful Eating Behavior Questions that can be Easily Answered?
Very
Useful
Moderately
Useful
Dicult to AnswerLess Dicult to Answer
Do Large Serving Bowls
Make You to Eat More?
Do Hungry Shoppers Buy
More or Just Buy Worse?
Will Kids Select
Healthy Foods with
Attractive Names?
Why Do Overweight
People Underestimate
How Much They Eat?
Do “Clean Plate” Kids Turn
into Overeating Adults?
Do Short Term Fasts Lead
to Long-term Weight Loss?
Does Preordering Lead
to Healthier Lunches?
Do Healthy Product Names
Make You Hate the Food?
Can Taste Profiles Predict
Food Preferences
Can Brand Logos
Encourage Kids to
Eat Heathy Foods?
Do Menu Prices Make You
Overeat and Regret It?
Does Traumatic Violence
Change Judgement
and Choice?
Can You Confirm the
Sweet Tooth Hypothesis? Have Classic Recipes
Gotten More Caloric
Over Time?
Will You Hate the
Food You Eat During
Bad Experiences?
Do Different TV Shows
Influence How You Eat?
Brian Wansink 2019 - 46
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.