Content uploaded by James E Painter
Author content
All content in this area was uploaded by James E Painter on Jan 19, 2015
Content may be subject to copyright.
SHORT COMMUNICATION
The office candy dish: proximity’s influence on
estimated and actual consumption
B Wansink
1
, JE Painter
2
and Y-K Lee
3
1
Cornell University, Ithaca, NY, USA;
2
Eastern Illinois University, Charleston, IL, USA and
3
Kyungpook National University,
South Korea
Objective and purpose: Although there is increasing interest in how environmental factors influence food intake, there are
mixed results and misunderstandings of how proximity and visibility influence consumption volume and contribute to obesity.
The objective of this paper is to examine two questions: first, how does the proximity and salience of a food influence
consumption volume? Second, are proximate foods consumed more frequently because they are proximate, or are they
consumed more frequently because people lose track of how much they eat?
Research methods and procedures: The 4-week study involved the chocolate candy consumption of 40 adult secretaries. The
study utilized a 2 2 within-subject design where candy proximity was crossed with visibility. Proximity was manipulated by
placing the chocolates on the desk of the participant or 2 m from the desk. Visibility was manipulated by placing the chocolates
in covered bowls that were either clear or opaque. Chocolates were replenished each evening, and placement conditions were
rotated every Monday. Daily consumption was noted and follow-up questionnaires were distributed and analyzed.
Results: There were main effects for both proximity and visibility. People ate an average of 2.2 more candies each day when they
were visible, and 1.8 candies more when they were proximately placed on their desk vs 2 m away. It is important to note,
however, that there was a significant tendency for participants to consistently underestimate their daily consumption of
proximately placed candies (0.9) and overestimate their daily consumption of less proximately placed candies ( þ0.5).
Discussion: These results show that the proximity and visibility of a food can consistently increase an adult’s consumption of it.
In addition, these results suggest that people may be biased to overestimate the consumption of foods that are less proximate,
and to underestimate those that are more proximate. Knowing about these deviation tendencies is important for those
attempting effectively monitor their consumption of fat and sugar.
International Journal of Obesity (2006) 30, 871–875. doi:10.1038/sj.ijo.0803217; published online 17 January 2006
Keywords: consumption; proximity; convenience visibility; candy; estimation; salience; food intake
Introduction
Clinicians and health professionals want to understand the
determinants of consumption volume as a foundation for
effective nutrition education and counseling.
1
Although
taste,
2,3
mood,
4,5
stress,
6
social context,
7
and role models
8
have been shown to influence consumption volume, it is not
clear how a food’s proximity or visibility influences how
much one consumes and how much one believes he/she has
consumed. Existing findings related to proximity and
visibility are largely inconsistent.
For instance, when comparing food storage habits in
homes of obese and non-obese families, Terry and Beck’s
9
first study showed food to be more visible in the homes of
obese families, but their second study showed the opposite.
A more recent experiment
10
indicated that accessible food
is often eaten more frequently, but there was no systematic
evidence whether this was related to poor consumption
monitoring or simply due to the convenience of the product.
Some indirect evidence exists that highly salient, stock-
piled and accessible foods are eaten more frequently than
less salient foods.
11
Yet this was only found to be the case
with convenient-to-consume foods such as chips, granola
and juice and was not with less convenient-to-consume
foods such as popcorn, soup and refrigerated cookie dough.
Although such results suggest that consumption conveni-
ence can lead to increases in consumption frequency, it may
be that proximate foods can be consumed with an increased
frequency because they are more quickly consumed and
more easily forgotten. Developmental psychology has shown
that the more effort or time invested in a unique activity, the
Received 22 September 2004; revised 23 March 2005; accepted 6 June 2005;
published online 17 January 2006
Correspondence: Dr B Wansink, Cornell Food and Brand Laboratory, Cornell
University, 110 Warren Hall, Ithaca, NY, 14853, USA.
E-mail: wansink@cornell.edu
International Journal of Obesity (2006) 30, 871 –875
&
2006 Nature Publishing Group All rights reserved 0307-0565/06
$
30.00
www.nature.com/ijo
more it is likely to be recalled.
12
As an analogue, it may be
that a person who has to walk to another table each time
he/she wants a snack may be more likely to remember how
many snacks they have eaten compared to a person who
only has to reach for a snack that is in a bowl in front of
them.
This raises two questions that are relevant for clinicians,
dieters and nutritionally conscious individuals. First, do
people eat more when a food is within sight, or when it
is within reach? Second, are increases in proximity-driven
consumption associated with systematic biases in how much
people believe they have eaten?
Methods
The participants were 40 female staff members (42.2711.3
years) from across six different departments at the University
of Illinois at Urbana-Champaign. Staff members were
recruited by an e-mail which asked them whether they
would be involved in a study related to candy. In exchange,
they were told they would be given a free candy dish filled
with chocolates and a $10 gift certificate to a local restaurant
(the Spice Box).
Those who claimed to typically consume three or more
pieces of candy each week were personally contacted and
told they would be periodically asked about their candy
preferences over a 4-week period. They were also told that
as a thank you, they would be given individually wrapped
chocolates (candy ‘kisses’) over that time period with the
only stimulation being that because of ‘cost constraints’ they
could not share them with others or take them home. They
were told that there were a number of people involved in the
study and that their candy dishes would be rotated with
others at the end of every week and that it might even
change places in their office. They were asked to leave the
bowl wherever it might be relocated.
The study utilized a 2 2 within-subject design with
repeated measures where visibility (visible vs non-visible)
was crossed with the proximity (proximate vs less proximate)
of the food’s location relative to each participant. Visibility
was manipulated by placing the chocolates in covered bowls
that were either clear or opaque. Proximity was manipulated
by placing the chocolates on the desk of the participant or
2 m from the desk of the participant at roughly the same
level as desk level. A distance of 2 m was chosen because it
was just out of reach of the participants and necessitated
them to stand up to get the candy.
10
It should be noted that
making an opaque bowl more proximate, might also make
the candy more salient when in the clear bowl. In general,
this form of salience is more one of salience than of visibility
at small distances of 0.5 m vs 2 m.
In the first week of the study, participants were divided
into four chocolate–placement conditions of 10 people each:
(1) proximate and visible, (2) proximate and non-visible, (3)
less proximate and visible and (4) less proximate and non-
visible. In each condition, the participants were given a
lidded container holding 30 chocolates (candy ‘kisses’).
During each day of the 4 weeks of this study, 30 chocolates
were placed in 20 clear containers and 20 opaque containers
and delivered to the 40 participants. The containers were
replenished every afternoon after 1900 hours. They were kept
in the same location for four consecutive business days
(Monday to Thursday). Because of irregular Friday schedules
(owing to flex-time), no intake data were collected on Friday.
On the next test day, the Monday of the following week, the
containers were rotated for each participant. The procedure
was repeated at the end of the first, second and third week,
and the study was completed on the Friday of the fourth week.
Researchers kept a daily record of the number of chocolates
consumed from each container. Comparisons were made of
the data collected from each location during the study
period. The within-subject factor was the proximity and
visibility of the chocolates; the 4 days in each location were
treated as repeated measures within each condition.
At the end of each week, on a Friday morning, each
participant was given a questionnaire which asked them how
much they thought they had eaten over the entire week and
which asked them their perceptions about the chocolates
(‘it was difficult to resist eating them,’ ‘I thought of eating
chocolates often,’ etc.) on 9-point scales (1 ¼strongly
disagree to 9 ¼strongly agree). Over the 4 days for each of
the 4 weeks, 16 consumption observations were obtained for
each of the 40 participants. To insure that the data collected
would be anonymous, the Human Subjects Committee
suggested that measures of height and weight not be
collected. As a result, it was not possible to determine the
body mass index (BMI) of the participants and to use this in
analyses.
The data about actual and perceived chocolate consump-
tion were analyzed across the four conditions using two-way
multiple analysis of variance with repeated measures (SPSS
10.2). As indicated in Table 1, manipulation checks for both
proximity and visibility indicated they were successfully
implemented as experimental conditions.
Results
As a baseline, when the candies were less visible and less
proximate, the average person ate 3.1 candies each day.
When the bowl was more visible and more proximate,
however, intake increased (see Figure 1). People ate an
average of 2.5 more candies each day when they were less
proximate but visible (5.6 vs 3.1; Po0.05), 1.5 more when
the were proximate but non-visible (4.6 vs 3.8; Po0.05) and
4.6 more when less proximate and visible (7.7 vs 3.1;
Po0.05). These results help corroborate the initial hypo-
theses of Terry and Beck
9
regarding the effects of visibility, as
well as the findings of Hearn et al.
13
regarding convenience.
At the same time, they also show the additive effects of the
interaction between visibility and proximity.
The office candy dish
B Wansink et al
872
International Journal of Obesity
It is also important to note that the proximity of a food
biased participant’s estimates of how much they thought
they ate. Participants had tendencies to underestimate how
many candies they ate when the food was proximate and to
overestimate how many they ate when less proximate.
Across both the visible and non-visible conditions, partici-
pants ate an average of 6.1 candies each day when they were
proximate, but they believed they ate an average of 5.2. In
contrast, when the candy was less proximate, participants
ate an average of 4.3 candies, but believed they ate 4.8. This
tendency to underestimate the consumption of proximately
placed candies (0.9) and to overestimate the consumption
of less proximately placed candies ( þ0.5) were both
significant (Po0.05).
The visibility and proximity of candies also influenced
perceptions of chocolate consumption (recall Table 1).
Regardless of whether participants could actually see the
chocolates, chocolates that were sitting on the desk (vs 2 m
from the desk) were rated as being more difficult to resist
(4.5 vs 2.7; Po0.05), as more attention-attracting (4.7 vs 2.8;
Po0.05). Visibility showed similar effects. Compared to
candies in opaque containers, candies in clear containers
were rated as being more difficult to resist (4.1 vs 3.1;
Po0.05) and as more attention-attracting (4.3 vs 3.0;
Po0.05).
One potential concern with within-subject designs is
that of possible demand effects when a person experiences
the different conditions of an experiment. To assess
whether this is a serious concern, the comparison of all
4 weeks of data was compared with that for only the first
week, when each participant was exposed to only one
condition. The basic pattern of the intake results for this
first week was consistent with the aggregate pattern for
the entire 4 weeks. A two-way between-subject analysis of
variance (with repeated measures) indicated a main effect
for both proximity (Po0.5) and a marginal effect for
visibility (Po0.10). This provides confidence that the
influence over the 4 weeks was of a similar nature and
not attributable to demand effects. Furthermore, if there
were demand effects, they would have been more likely to
make the results less significant and not as strong as the
effects found here.
When the candies were 6 feet away from the desk, people
ate more candies but tended to underestimate the number
they had eaten. This raises the issue as to whether this
decrease in consumption occurred because of the distance
itself or because the underestimation that accompanied it.
To examine this, a mediation analysis was conducted using
estimated consumption as a mediator between proximity
and consumption.
14
In conducting this analysis, it was
found that proximity had a direct influence on both
consumption (Po0.05) and on estimation (Po0.05), but
the influence of estimation on intake was not significant
(Po0.20). Furthermore, when consumption estimation was
included in a regression of proximity on to consumption,
the impact of proximity remained significant (Po0.05).
These results indicate that even though the consumption of
a distant object is overestimated, this bias cannot account for
the effect this distance had on reducing intake.
Table 1 Influence of proximity and visibility on processing measures
Proximate (placed on-desk) Less proximate (placed 2 m from desk)
Visible (clear bowl) Non-visible (opaque bowl) Visible (clear bowl) Non-visible (opaque bowl)
It was difficult to resist eating them 4.9672.55
a,b,c
3.9172.39
a,b
3.1772.12
a
2.3071.72
I thought of eating the chocolates often 4.7872.50
a,b,c
3.7472.26
a,b
3.1372.36
a
2.3571.94
I actually forgot they were around 3.1372.47
a,b,c
4.3973.07
a,b
5.4872.66
a
6.4873.06
They kept attracting my attention 5.3572.99
a,b,c
3.9672.38
a,b
3.3072.08
a
2.0471.55
They were very convenient to eat 7.6571.92
a,b,c
5.6172.33
a,b
3.8372.46
a
2.5272.00
They were easy to eat 7.7071.89
a,b,c
6.1371.98
a,b
4.5272.56
a
3.8672.69
They were visible 8.0971.83
a,b,c
2.9672.48
a,b
5.5272.78
a
1.8271.80
It took little effort to get one 8.0471.87
a,b,c
6.7072.10
a,b
4.5772.86
a
3.8272.77
Mean7s.d. Scale 1 ¼strongly disagree to 9 ¼strongly agree.
a
Significantly different from the non-visible and less proximate condition (Po0.05).
b
Significantly
different from the visible and less proximate condition (Po0.05).
c
Significantly different from the non-visible and proximate condition (Po0.05).
7.7
6.5
4.6
3.8
5.6
6.1
3.1
3.5
0
1
2
3
4
5
6
7
8
Proximate &
Visible
Proximate &
Non-Visible
Less Proximate &
Visible
Less Proximate &
Non-Visible
Actual Number of
Candies Consumed
Estimated Number of
Candies Consumed
Number of Candies Consumed
Proximity and Visibility
of the Cand
y
Figure 1 The impact of proximity on actual and estimated candy
consumption.
The office candy dish
B Wansink et al
873
International Journal of Obesity
What then explains how distance influences consump-
tion? Whereas distance and effort would explain why
consumption is decreased,
15
debriefing interviews indicated
an additional explanation. Although some participants
noted the lack of convenience as being a consideration
when they decided whether to eat a chocolate, 26 of 40
(65%) noted that having 2 m between them and the candy
gave them an extra second to pause and reconsider
whether they were really hungry enough to want (or need)
another candy. This notion that extra effort can help
facilitate cognitive control is a finding that was unexpected
and would not have been revealed if this was being
investigated with commonly studied animal models, such
as rats.
16
Interestingly, the corresponding explanation as to why
distance also tended to cause people to overestimate how
much they had consumed was explored in debriefing
interviews. When the candies were sitting on one’s desk,
these interviews indicated that a person typically retrieved
one candy kiss at a time and ate one candy kiss at a time.
When the candies were sitting 2 m away, participants often
reported retrieving two (or more) candies at a time. They
opened the candy jar less often, but they took more because
of the increased effort. As a result, when they estimated how
many candies they ate when the dish was 2 m away, they
typically estimated the number of times they retrieved
candies and multiplied this by two. If combined with
instances when they only retrieved one, this would lead to
an overestimation of how much they had consumed over the
course of the week.
Discussion
In extending other findings,
8,10
these results underscore that
the proximity and visibility of a food can consistently
increase an adult’s consumption of it. Yet although they
ate more, they also tended to underestimate the amount
they had eaten when the candies were proximate and visible.
This is consistent with past work that a basic availability bias
causes people to over-represent or overestimate the inci-
dence or quantity of items that are more salient (or available)
in memory.
17
Such a bias can lead people to overestimate the
consumption of those foods that are less proximate to obtain
(or to prepare), and to underestimate or forget those that
were more proximate to consume.
18
This has important implications for people who are trying
to be accurate in monitoring and controlling their intake of
food. These results underscore that people need to take a
food’s visibility and proximity into account when they try
and estimate their prior consumption of it. In general, a food
that is less proximate to consume – say cookies in the
cupboard vs those on the counter – may be over-consumed
relative to what one might think (or recall).
Furthermore, understanding this bias has implications for
researchers studying consumption data and diary panels.
These findings emphasize that it is important to account for
the visibility and proximity of foods because not doing so
can lead to biased consumption recall studies and biased
diary panel estimates. One way to do this is to ask people
to rate the visibility and proximity of the foods under
investigation. These ratings can then be used as either
covariates or can be used as blocking or segmentation
variables.
19
Whereas the tendency to underestimate one’s consump-
tion of a proximate food may be a general tendency across
people, obese people may be more motivated to deny what
they have eaten than non-obese people. That is, although
there may be a gap in everyone’s estimation of proximate
foods, some populations may be more extreme in their bias
than others. Although measures of BMI were not able to have
been taken in this study, if such a relationship does exist, it
would most likely found in future studies that use within-
subject designs in field situations.
Encouragingly, if visibility and proximity increase the
consumption of chocolate, it may also work for healthier
foods, such as raw fruits or vegetables.
20
What makes the
candy dish nutritionally dangerous, might bring the fruit
bowl back in vogue.
Acknowledgements
Special thanks to Paula for her help in data collection.
References
1 Thomas PR. Improving America’s Diet and Health: From Recom-
mendations to Action. National Academy Press: Washington,
DC, 1991.
2 Glanz K, Basil M, Maibach E, Goldberg J, Snyder D. Why
Americans eat what they do: taste, nutrition, cost, convenience,
and weight control concerns as influences on food consumption.
J Am Diet Assoc 1998; 98: 1118–1126.
3 Sporny LA, Contento IR. Stages of change in dietary fat reduc-
tion social psychological correlates. J Nutr Educ 1995; 27:
191–199.
4 Gould SJ. An interpretive study of purposeful, mood self
regulating consumption: the consumption and mood framework.
Psychol Marketing 1997; 14: 395–426.
5 Patel KA, Schlundt DG. Impact of moods and social context on
eating behavior. Appetite 2001; 36: 111–118.
6 Oliver G, Wardle J, Gibson L. Stress, food choice: a laboratory
study. Psychosomatic Med 2000; 62: 853–865.
7 Berry SL, Beatty WW, Klesges RC. Sensory, social influences on
ice cream consumption by males, females in a laboratory setting.
Appetite 1985; 6: 41–45.
8 Birch LL, Fisher JO. Mother’s child-feeding practices influence
daughters’ eating, weight. Am J Clin Nutr 2000; 71: 1054–1061.
9 Terry K, Beck S. Eating style, food storage habits in the home:
Assessment of obese, non-obese families. Behav Modification 1985;
9: 242–261.
10 Painter JE, Wansink B, Hieggelke JB. How visibility, convenience
influence candy consumption. Appetite 2002; 38: 237–238.
11 Chandon P, Wansink B. Does stockpiling accelerate consumption.
A convenience-salience framework of consumption stockpiling.
J Marketing Res 2002; 39: 321–335.
The office candy dish
B Wansink et al
874
International Journal of Obesity
12 Bauer PJ, Wewerk SS. One- to two-year-old’s recall of events: the more
expressed, the more impressed. J Exp Child Psychol 1995; 59: 475–496.
13 Hearn MD, Baranowski T, Baranowski J, Doyle C, Smith M, Lin LS,
et al. Environmental influences on dietary behavior among
children: availability and accessibility of fruits and vegetables.
J Health Educ 1998; 29: 26–32.
14 Baron RM, Kenny DA. The moderator-mediator variable distinction
in social psychological research: conceptual, strategic, and statis-
tical considerations. J Personality Social Psychol 1986; 51: 1173–1182.
15 Collier G, Hirsch E, Hamlin PH. Ecological determinants of
reinforcement in rats. Physiol Behav 1972; 9: 705–716.
16 Levitsky DA. Putting behavior back into feeding behavior:
a tribute to George Collier. Appetite 2002; 38: 143–148.
17 Lee AY, Sternthal B. The effects of positive mood on memory.
J Consumer Res 1999; 26: 115–127.
18 Wansink B. Environmental factors that increase the food intake
and consumption volume of unknowing consumers. Ann Rev
Nutr 2004; 24: 455–479.
19 Wansink B, Sudman S. Consumer Panels, 2nd edn. American
Marketing Association: Chicago, 2002.
20 Wansink B, Cheney MM. Super bowls: serving bowl size and food
consumption. JAMA 2005; 293: 1727–1728.
The office candy dish
B Wansink et al
875
International Journal of Obesity
