Research report

Im watching you. Awareness that food consumption is being

monitored is a demand characteristic in eating-behaviour
experiments
Eric Robinson
a,
*, Inge Kersbergen
a
, Jeffrey M. Brunstrom
b
, Matt Field
a
a
Department of Psychological Sciences and UK Centre for Tobacco and Alcohol Studies (UKCTAS), Eleanor Rathbone Building, University of Liverpool,
Liverpool L69 7ZA, UK
b
Nutrition and Behaviour Unit, School of Experimental Psychology, University of Bristol, UK
A R T I C L E I N F O
Article history:
Received 2 June 2014
Received in revised form 25 July 2014
Accepted 28 July 2014
Available online 30 July 2014
Keywords:
Demand characteristics
Experimenter effects
Laboratory methods
Eating behaviour
Awareness
A B S T R A C T
Eating behaviour is often studied in the laboratory under controlled conditions. Yet people care about
the impressions others form about them so may behave differently if they feel that their eating behaviour
is being monitored. Here we examined whether participants are likely to change their eating behaviour
if they feel that food intake is being monitored during a laboratory study. In Study 1 participants were
provided with vignettes of typical eating behaviour experiments and were asked if, and how, they would
behave differently if they felt their eating behaviour was being monitored during that experiment. Study
2 tested the effect of experimentally manipulating participants beliefs about their eating behaviour being
monitored on their food consumption in the lab. In Study 1, participants thought they would change their
behaviour if they believed their eating was being monitored and, if monitored, that they would reduce
their food consumption. In Study 2 participants ate signicantly less food after being led to believe that
their food consumption was being recorded. Together, these studies demonstrate that if participants believe
that the amount of food they eat during a study is being monitored then they are likely to suppress their
food intake. This may impact the conclusions that are drawn from food intake studies.
2014 Elsevier Ltd. All rights reserved.
Introduction
Eating behaviour is often studied under laboratory conditions.
In this context, participants eat in a controlled environment and the
dependent variable of interest is often meal size the amount that
people consume when offered ad libitum access to a food. For
example, researchers have used laboratory methods to study cog-
nitive (Higgs, 2002), social (Conger, Conger, Costanzo, Wright, &
Matter, 1980) and environmental inuences on food consumption
(Rolls, Roe, Halverson, & Meengs, 2007). Meiselman (1992) has sug-
gested that the laboratory creates an articial setting that tells us
about eating in an unnatural context, and that greater emphasis
should be placed on studying human behaviour in realistic situa-
tions. de Castro (2000) expressed similar concerns and suggested
that the articial nature of the laboratory environment may result
in researchers reaching invalid conclusions about human eating
behaviour on the basis of lab studies (see de Castro, 2000).
The prospect that demand and/or experimenter effects can bias
participant behaviour has been discussed extensively by social psy-
chologists (Laney et al., 2008; Orne, 1962; Orne, Whitehouse, &
Kazdin, 2000). However, in relation to studies of eating behaviour,
less is known about whether participants change their eating
behaviour or meal size in response to awareness that food con-
sumption is being monitored by an experimenter. Previously, it has
been suggested that the amount or way in which a person eats can
act as a powerful self-presentation tool. This is because we form
judgements about other people based on their eating behaviour and
are aware that others may do the same about us (Vartanian, Herman,
& Polivy, 2007). For example, people eat smaller portions when in
the company of strangers (Salvy, Jarrin, Paluch, Irfan, & Pliner, 2007a)
and women may eat smaller meals to portray femininity (Mori,
Chaiken, & Pliner, 1987; Pliner & Chaiken, 1990). Moreover, if others
are watching, then we may make strategic food choices that can in-
uence the impression that is formed by our observers (Berger &
Rand, 2008; Guendelman, Cheryan, & Monin, 2011).
These observations highlight the possibility that eating behaviour
can be modied by awareness that food intake is being moni-
tored. Consistent with this proposition, in some studies overweight
and obese individuals (who may be particularly concerned about
how others perceive their eating) ate less than their lean counter-
parts (Salvy, Coelho, Kieffer, & Epstein, 2007b; Shah et al., 2014),
which is also compatible with ndings that the overweight and obese
are more likely to under-report dietary intake (see Mela & Aaron,
1997). A study by Polivy, Herman, Hackett, and Kuleshnyk (1986)
* Corresponding author.
E-mail address: eric.robinson@liv.ac.uk (E. Robinson).
http://dx.doi.org/10.1016/j.appet.2014.07.029
0195-6663/ 2014 Elsevier Ltd. All rights reserved.
Appetite 83 (2014) 1925
Contents lists available at ScienceDirect
Appetite
j our nal homepage: www. el sevi er. com/ l ocat e/ appet
also suggests that awareness during eating may be of importance;
in one condition of this laboratory study participants were made
to feel more conscious of their eating behaviour and this resulted
in participants reducing their food intake. Likewise, Roth, Herman,
Polivy, and Pliner (2001) found that a reduction in food intake can
occur merely due to the physical presence of an experimenter during
a test meal. Although we presume it would be rare for a re-
searcher to be present (although this actually does occur in some
studies, e.g., Andrade, Kresge, Teixera, Baptista, & Melanson, 2012),
it could be the case that mere awareness that eating behaviour is
being recorded also affects meal size. Thus, although little work has
specically examined whether participants modify their food intake,
if they believe that their food consumption is being monitored (i.e.,
the researcher will later record how much has been eaten), exist-
ing studies suggest this may be the case.
There are two reasons why this type of demand characteristic
could be problematic for the interpretation of ndings from labo-
ratory studies. First, different research groups may use different
methods, making it dicult to evaluate ndings across studies. Spe-
cically, some conceal the fact that food consumption is recorded
(e.g., Hermans, Larsen, Herman, & Engels, 2010) whilst others reveal
this information to their participants (e.g., Yip, Wiessing, Budgett,
& Poppitt, 2013). Second, if participants are eating very little due
to heightened demand awareness during a study, this may create
an articial oor effect on food intake. In other words, if partici-
pants experience external pressure to consume a small meal this
would make it more dicult to detect an additional meaningful de-
crease in food consumption that might occur as a result of
experimental manipulations. For example, consider a study testing
whether an experimental manipulation reduces food intake. If food
intake is signicantly suppressed then this may limit the opportu-
nity to observe further reductions caused by the experimental
manipulation. The aimof the present studies was to assess the extent
to which people adjust their food intake when they are aware that
their meal size is being monitored. In Study 1 participants were pro-
vided with vignettes of typical eating-behaviour experiments and
were asked if, and how, they would behave differently if they felt
their eating behaviour was being monitored. In Study 2 we ex-
plored the effect of telling participants that their intake would be
monitored on actual food intake.
Study 1
Overview
Study 1 was an internet survey in which we provided partici-
pants with a number of vignettes describing typical laboratory
eating-behaviour experiments. In the rst set of vignettes partici-
pants were asked if and how awareness that their eating behaviour
was being monitored would inuence their food consumption. We
reasoned that this awareness might also be associated with suspi-
cions of specic experimental hypotheses being tested. Accordingly,
we included a second set of vignettes in which participants were
provided with the study aim before being asked whether their food
intake would be inuenced by awareness of monitoring of their
intake. We hypothesised that participants would report that aware-
ness of monitoring would reduce their food consumption.
Study 1: Method
Participants
We aimed to recruit one hundred participants, but allowed for
a slightly larger sample to allow for cases where participants failed
to complete all of our questions. One hundred and eight partici-
pants (mean age = 20.9, SD = 3.6) completed the study. All were
recruited via a text advertisement on online notice boards at the
University of Liverpool, UK. Adverts were accessible to only under-
graduate and postgraduate students and the study was described
as an investigation of eating behaviour. Ninety four participants were
female and 14 were male. All were entered into a small cash-prize
draw. The study received ethical approval from the Research Ethics
Committee at the University of Liverpool.
Procedure
After accessing the online study site, participants were told they
would be provided with hypothetical scenarios and were asked to
answer honestly about how they would behave. In this rst section
participants were asked You are participating in a psychology study
and are provided with a bowl of cookies during the study, which
you are asked to make taste ratings about. If you thought that the
researcher would later measure how many cookies you had eaten
(as opposed to you believing they werent measuring this), do you
think it would inuence how much you would eat? (Monitoring of
snack food intake question) and answered by selecting Yes, No or
Unsure. On the same page participants were asked In the above
scenario, in what way would or wouldnt your behaviour change?
and given options I would eat the same amount of cookies, I would
eat more cookies, I would eat fewer cookies or Unsure. Next, par-
ticipants were asked You are participating in a research study taking
place at lunchtime and during a task the researcher leaves you with
a lunch buffet. If you thought the researcher would be keeping track
of howmuch youd eaten of each food (as opposed to themnot mea-
suring howmuch youd eaten), do you think it would inuence how
much you would eat? (Monitoring of lunch food intake question) Par-
ticipants were then asked in what way they would or would not
change their behaviour using the same response formats as de-
scribed above.
In the next section participants were given two hypothetical sce-
narios about participating in a between-subjects experiment. In a
study you are asked to watch TV and the researcher leaves a se-
lection of snacks and drinks. You notice there are food adverts during
the TV programme and think the study might be examining whether
food adverts increase how much food you eat (TV advert hypothe-
sis awareness question). Participants were asked two questions: Do
you think knowing the study aims would inuence how much you
would eat (Yes, No, Unsure)? and In what way would or wouldnt
your behaviour change? (Id probably eat the same/more/less food
than if I didnt know the aims, or Unsure). The next vignette was
You are taking part in a research study and the researcher happens
to leave nutritional information about a food, which indicates that
the food product is high in calories. You are later served the food
in question and you believe that the study is probably testing
whether calorie labelling reduces how much you eat (Food label-
ling hypothesis awareness question). Participants were asked two
questions: Do you think knowing the study aims would inuence
how much you would eat? and In what way would or wouldnt
your behaviour change? The same response formats were used as
in the TV advert hypothesis awareness question.
In the nal section participants were given two hypothetical sce-
narios about participating in a repeated-measures experiment.
Participants were rst told You take part in a study with multiple
visits to a laboratory. During these visits you rate hunger before and
after being provided with a meal. You are asked to eat at a normal
speed on one day, very fast on another day and very slowon another
day. You think that the study is probably testing whether how fast
you eat affects how much you eat. Knowing this, do you think it
would inuence howmuch you eat during any of the sessions? Par-
ticipants were also asked to indicate whether this would result in
themeating more, less, the same amount of food (or unsure) during
the slow and fast eating days individually (Eating rate hypothesis
20 E. Robinson et al./Appetite 83 (2014) 1925
awareness question). The nal vignette was You are participating
in a research study with multiple laboratory visits. On both visits
you eat a lunch of pizza, although in one of the sessions you are
asked to smell and taste some of the pizza prior to lunch, whilst
having the amount of saliva you are producing measured and in the
other session you have the amount of saliva you are producing mea-
sured, but are not told to smell or taste the pizza rst. You think
the study might be looking at whether the smell and anticipation
of food inuences how much you eat. Knowing this, do you think
it would inuence how much you eat during any of the sessions
(Yes, No, Unsure)? Participants were also asked to indicate whether
this would result in them eating more, less, the same amount of
food (or unsure) during the two visits individually (Food-cue reac-
tivity hypothesis awareness question), using the same response
format as in the eating rate hypothesis awareness question. Par-
ticipants then provided demographic information and were
debriefed.
Analysis
We did not plan a formal statistical analysis strategy, as our
main aim was to explore the frequency by which participants
would report that awareness of a researcher monitoring their
food consumption would inuence their eating behaviour and
reduce food consumption in a variety of hypothetical laboratory
scenarios.
Study 1: Results
Participants were asked whether their eating behaviour would
be inuenced by knowledge that a researcher was going to record
a) their consumption of cookies (Monitoring of snack food intake ques-
tion) and b) intake at a lunchtime meal (Monitoring of lunchtime
intake question). In response, 72% of participants for the snack food
question and 59% of participants for the lunchtime meal question
reported their behaviour would be affected. When asked how this
would inuence their behaviour, the most common response was
that it would decrease their food intake. Respectively, in response
to the snack-food question and the lunchtime food question, this
option was selected by 66% and 52% of the participants. See Table 1
for detailed frequencies of responses and questions.
Two questions explored whether awareness of a specic hy-
pothesis might inuence food intake. For the TV hypothesis awareness
question, 60% of participants reported that their behaviour would
be inuenced by this demand awareness. A similar proportion (58%)
responded in the same way to the Food labelling hypothesis aware-
ness question. A sizeable proportion of participants reported that it
would cause them to eat less food (36% and 50% respectively for
the two questions). See Table 1 for detailed frequencies of re-
sponses for the TV hypothesis awareness question and the Food
labelling hypothesis awareness question.
Participants were also asked two questions about whether aware-
ness of a specic research hypothesis might inuence their behaviour.
Generally, a minority reported this to be the case. For the Eating rate
hypothesis awareness question, 28.7% of participants reported that
becoming aware of the aims of the study would inuence their
behaviour, 64.5% reported it would not, and 6.5% reported being
unsure. For the Food cue reactivity hypothesis awareness question,
28.7% of participants reported that becoming aware of the aims of
the study would inuence their behaviour, 62.0% reported it would
not, and 9.3% reported being unsure. See Table 2 for detailed fre-
quencies of how participants reported awareness of the study aims
would inuence their food consumption for the questions.
Study 1: Conclusions
When provided with vignettes of typical eating-behaviour ex-
periments, a high proportion of participants reported that awareness
of a researcher monitoring their food consumption would inu-
ence their behaviour and a proportion of participants reported that
it would be likely to result in them decreasing their food consump-
tion (this varied from11% to 66% of participants across the different
vignettes). There was some variability in the frequency with which
participants reported that awareness would be likely to inuence
their behaviour across the different vignettes, but as we used dif-
ferent hypothetical scenarios and experiments for each of the
vignette types, formal comparisons of these differences (using in-
ferential statistics) were not meaningful. Our main conclusion from
Study 1 is that some participants believe they would be likely to
behave differently if they were aware that their eating was being
monitored in a study. The rationale for Study 2 was to test if these
beliefs translate to actual changes in eating behaviour.
Table 1
Study 1 frequency of responses to awareness questions one, two, three and four.
Inuence how much you would eat? In what way would/wouldnt your behaviour change?
Yes No Unsure Eating
the same
Eating
more
Eating
less
Unsure
Monitoring of snack food consumption Q If you thought
that the researcher would later measure how many cookies
you had eaten, do you think it would inuence how much you
would eat?
72.2% 26.9% 0.9% 29.7% 3.7% 65.7% 0.9%
Monitoring of lunch consumption Q If you thought the
researcher would be keeping track of how much youd eaten
of each food (from the lunch buffet), do you think it would
inuence how much you would eat?
59.3% 37.0% 3.7% 40.7% 4.6% 51.9% 2.8%
TV hypothesis awareness Q You notice there are food
adverts during the TV programme and think the study might
be examining whether food adverts increase how much food
you eat. Do you think knowing the study aims would
inuence how much you would eat?
60.2% 29.6% 10.2% 35.2% 14.8% 36.1% 13.9%
Food labelling hypothesis awareness Q the researcher
happens to leave nutritional information about a food, which
indicates that the food product is high in calories. You are
later served the food in question and you believe that the
study is probably testing whether calorie labelling reduces
how much you eat. Do you think knowing the study aims
would inuence how much you would eat?
58.3% 34.3% 7.4% 37.0% 3.7% 50.0% 9.3%
Values denote percentage of responses from the one hundred and eight participants for each question.
21 E. Robinson et al./Appetite 83 (2014) 1925
Study 2
Overview
The aim of Study 2 was to examine whether manipulating the
extent to which participants believed their food consumption would
be recorded in the lab would inuence the amount of food they sub-
sequently ate. In order to be comparable to other laboratory eating
behaviour studies, we used the commonly employed taste-test par-
adigm (see Boon, Stoebe, Schut, & Ijntema, 2002; Oldham-Cooper,
Hardman, Nicoll, Rogers, & Brunstrom, 2011; Robinson, Benwell, &
Higgs, 2013) in which participants are asked to make sensory ratings
about a food as part of a taste test (cover story), and intake of the
leftover food is recorded. The amount of food consumed is usually
the primary dependent measure. We compared a standard cover
story (control condition) to two other experimental conditions. In
one condition participants were told that their food consumption
would be recorded (monitored condition) and in another they were
led to believe that their food intake would not be recorded
(unmonitored condition). The rationale for the unmonitored con-
dition was to attempt to eliminate perceptions of monitoring. We
opted to include all three conditions so that we could examine food
intake under standard conditions (control) and then directly compare
this to experimental manipulations designed to reduce and in-
crease monitoring of perceived intake. Based on the results of Study
1, our primary hypothesis was that participants in the monitored
condition would eat signicantly less food than participants in the
other conditions. We also hypothesised that participants would eat
more cookies in the control vs. unmonitored condition, as feelings
of being completely unmonitored may encourage participants to eat
more freely (i.e. we reasoned there still may have been suspicion
of monitoring in the control condition).
Study 2: Method
Participants
Based on the large effect sizes (Cohens d = 0.760.91) of in-
creased awareness on food intake that Polivy et al. (1986) observed
and the large effect that the presence of an experimenter had on
food intake (Cohens d = 0.97) in Roth et al. (2001), we calculated
that we would need a minimumsample size of 66 participants (80%
power, p < 0.05). We recruited slightly above this gure (n = 72) to
compensate for participants who might withdraw during the study
or provide incomplete data. All participants were female (mean
age = 20.0, SD = 1.3). Recruitment was limited to females in order
to promote homogeneity in food intake across participants and
because it is fairly common to use female only samples in eating
behaviour studies (e.g. see Koh & Pliner, 2007; Spiegel, Kaplan,
Tomassini, & Stellar, 1993). Participants were undergraduate stu-
dents and the study was approved by the Research Ethics Committee
at the University of Liverpool.
Design and experimental conditions
The study adopted a between-subjects design with three con-
ditions. In the control condition participants completed a standard
taste-test in a test cubicle alone. Specically, they were offered a
bowl of cookies and were told that they will be asked to rate their
avour. Participants were also told that they were free to eat as many
cookies as they liked and that any remaining food would be thrown
away. Participants were then told that the researcher would return
(after 7 minutes) and the remainder of the study would involve com-
pleting some questionnaires. The monitored condition was identical
except that participants were told that the researcher would make
a note of how many cookies had been eaten. In the unmonitored
condition the participants were told that once they have nished
eating they should dispose of any remaining food (whilst pointing
at a waste bin in the corner of the room, which was also present
in the other two conditions), because the researchers were only in-
terested in measuring the avour ratings. We reasoned that this
would convince the participants that their food consumption was
not being monitored.
Procedure
As a cover story, prior to taking part, participants were in-
formed that the study was exploring mood, personality and avour
perception. The study took place in a testing cubicle with a single
chair and table, with a small push-top bin in the corner of the room.
After being seated, participants were told that the study would
involve completing mood ratings and self-report questionnaires, and
that they would be rating the avour of cookies. Participants were
provided with an initial questionnaire which measured eleven mood
Table 2
Study 1 frequency of responses to awareness questions ve and six.
In the session in which the hypothesis
was for me to eat less, awareness would/
would not inuence my behaviour by
causing me to eat
In the session in which the hypothesis
was for me to eat more, awareness would/
would not inuence my behaviour by
causing me to eat
The same More Less Unsure The same More Less Unsure
Eating rate hypothesis awareness Q You are asked to eat at a
normal speed on one day, very fast on another day and very slow
on another day. You think that the study is probably testing
whether how fast you eat affects how much you eat. Knowing this,
do you think it would inuence how much you eat during any of
the sessions?
56.5% 8.3% 26.9% 8.3% 62.0% 18.5% 13.0% 6.5%
Food cue reactivity hypothesis awareness Q On both visits you
eat a lunch of pizza, although in one of the sessions you are asked
to smell and taste some of the pizza prior to lunch, whilst having
the amount of saliva you are producing measured and in the other
session you have the amount of saliva you are producing
measured, but are not told to smell or taste the pizza rst. You
think the study might be looking at whether the smell and
anticipation of food inuences how much you eat. Knowing this,
do you think it would inuence how much you eat during any of
the sessions?
74.1% 3.7% 15.7% 6.5% 62.0% 21.3% 11.1% 5.6%
Values denote percentage of responses from the one hundred and eight participants for each question.
22 E. Robinson et al./Appetite 83 (2014) 1925
items using a 100-mm visual-analogue scale with anchors of not
at all and extremely (e.g., howexcited do you feel right now?). Em-
bedded in these items was a measure of hunger. This enabled us
to compare baseline hunger across conditions. Consistent with the
cover story, at the end of the questionnaire, the participants were
instructed to describe their current mood in one sentence. The
researcher left the roomwhilst the participant completed the ques-
tionnaire. On re-entering the room the researcher removed the
questionnaire, provided participants with a well-stocked plate of
Maryland chocolate chip cookies (15 cookies, 502 kcal/100 g, ap-
proximate weight per cookie 11 g), and issued instructions tailored
to one of the three conditions (see above). Participants were also
given a taste-rating questionnaire. Using 100-mm line scales par-
ticipants rated the cookies on seven sensory dimensions (e.g., How
crunchy are the cookies?).
When the researcher returned (7 minutes later) the plate was
removed and participants were issued a nal questionnaire. In this
questionnaire participants rst reported their age, weight and height.
Next, they completed ve items taken from the restraint scale of
the Three Factor Eating Questionnaire (Stunkard & Messick, 1985).
This short-formmeasure (e.g., I do not eat some foods because they
make me fat) was included to evaluate levels of dietary restraint
across conditions. The nal page of the questionnaire was entitled
Feedback about taking part in our study. The rst question asked
participants to guess the aim of the study. The next ve items were
included to assess the participants experience during the study (e.g.,
I felt bored during the study, strongly agree to strongly disagree,
ve point Likert scale). The third item served as a manipulation
check; I felt as though the amount of food I was eating would be
measured by the researcher, with a higher score indicating greater
monitoring awareness. After completing the nal questionnaire par-
ticipants were thanked for their time and debriefed. In the
unmonitored condition the researcher retrieved any cookies from
the bin and used these to calculate the amount of food that had been
eaten.
Analysis
We used one-way ANOVAs to explore differences in age, base-
line hunger, BMI and restraint across conditions. We also used one-
way ANOVAs to evaluate the amount of food (g) consumed and to
evaluate responses to our manipulation check (degree of monitor-
ing), across conditions. Where appropriate, Bonferroni corrected
t-tests were used to evaluate differences between individual
conditions.
Study 2: Results
Participant characteristics
ANOVA indicated no signicant group differences (i.e. no main
effects of condition) on baseline hunger [F (2, 71) = 0.15, p = 0.86],
age [F (2, 71) = 1.54, p = 0.22] or restraint [F (2, 71) = 0.87, p = 0.42].
Six participants did not report full weight and height information,
so we were unable to calculate their BMI. For the remaining par-
ticipants, we found no evidence for a signicant effect of condition
on BMI [F (2, 63) = 3.42, p = 0.64]. See Table 3 for means and SDs for
all variables.
Food consumption
There was a signicant effect of condition on grams of cookies
consumed [F (2, 71) = 5.77, p = 0.005]. Bonferroni corrected t-tests
indicated that participants in the monitored condition ate signi-
cantly less food than participants in both the control condition
(p = 0.02, Cohens d = 0.93) and the unmonitored condition (p = 0.009,
Cohens d = 0.92). Participants in the control condition and
unmonitored condition did not differ signicantly in their food intake
(p = 0.99 Cohens d = 0.01). See Table 4 for means and SDs.
Awareness of monitoring
The effect of condition on self-reported perceptions of being
monitored approached statistical signicance [F (2, 71) = 2.81,
p = 0.067]. Participants in the monitored condition tended to have
a stronger belief that the researcher would record their eating
behaviour than participants in the unmonitored condition (p = 0.06,
Cohens d = 0.8). No signicant difference was observed between par-
ticipants in the monitored condition and the control condition
(p = 0.82, Cohens d = 0.3), or between the participants in the
unmonitored condition and control condition (p = 0.63, Cohens
d = 0.4). See Table 3 for associated means and SDs. A closer exam-
ination of the unmonitored condition suggested that this
manipulation was unsuccessful. Of the 24 participants in this con-
dition, only 14 followed the instruction to dispose of their leftover
cookies by placing them in the waste bin. Thus, the remaining 10
participants may have known that the researcher would see how
much food they had eaten when they returned. In addition, only a
minority of participants (5/24) selected a response (disagree) which
indicated that they believed their intake was not being recorded.
Cover story
Participants were asked to guess the aims of the research
(whether awareness of a researcher monitoring/not monitoring food
intake inuenced behaviour). Two independent coders agreed that
70 of the 72 participants were clearly unaware. Common re-
sponses related to how eating affects mood, suggesting that our
cover story was believed. Removal of the two participants who came
close to identifying the aims of the study (e.g., whether being
watched affects how many cookies are eaten) did not change the
statistically signicant and non-signicant between group differ-
ences observed for food intake. These two participants were in the
unmonitored condition.
Table 3
Study 2 participant age, BMI, hunger and restraint by condition.
Condition Age BMI Hunger Restraint
Monitored condition 20.3 (1.2) 21.8 (3.1) 40.8 (23.1) 13.8 (4.3)
Control condition 19.7 (0.8) 21.6 (3.0) 44.1 (26.5) 12.4 (4.3)
Unmonitored condition 20.2 (1.6) 21.0 (2.0) 40.4 (26.8) 14.0 (4.5)
Values are means (SDs). Age in years, hunger scores = 0100 mm VAS, BMI = self-
reported weight/height
2
restraint scores = 525, lower values denote lower restraint.
Table 4
Study 2 food consumption and reported awareness of eating behaviour monitor-
ing by condition.
Condition Grams of
cookie
eaten (g)
How strongly
participants
believed their
eating was
monitored,
1(strongly
disagree)
5(strongly agree)*
Monitored condition (n = 24) 29.3 (12.4) 4.0 (0.8)
Control condition (n = 24) 45.8 (21.9) 3.7 (1.1)
Unmonitored condition (n = 24) 47.7 (25.3) 3.3 (0.9)
Values are means (SDs). See main text for statistical signicance of between con-
dition comparisons.
* I felt as though the amount of food I was eating would be measured by the
researcher.
23 E. Robinson et al./Appetite 83 (2014) 1925
Study 2: Conclusions
After being led to believe that a researcher would record how
much food they were going to eat, participants ate signicantly fewer
cookies than participants in both a control condition and a condi-
tion in which we attempted to reduce awareness of consumption
monitoring. These effect sizes were statistically large. Contrary to
expectation, we found little evidence that participants in the
unmonitored condition ate more food than those in the control con-
dition. We attribute this to a manipulation failure our manipulation
check indicated that participants in the unmonitored condition
tended to believe that their intake was being monitored.
General discussion
The aimof the present research was to examine whether aware-
ness that ones food consumption is being monitored in a laboratory
study affects eating behaviour. In Study 1 participants were shown
vignettes of typical eating behaviour experiments and were asked
if, and how, they would behave differently if they felt their eating
behaviour was being monitored in an experiment. Across a variety
of vignettes, the majority of participants reported that they would
be likely to change their behaviour if they felt that their eating
was being monitored and that they would eat less. In Study 2 we
tested the effect on food consumption of experimentally manipu-
lating the extent to which participants felt their eating was
being monitored in the lab. Participants ate signicantly less food
after being led to believe that their food consumption was being
recorded.
The results of these two studies suggest that if participants believe
that the amount of food they are eating in a laboratory experi-
ment will be recorded, this will result in them changing their
behaviour and consuming less food than normal. These ndings are
also consistent with earlier studies which indicate that height-
ened self-awareness causes individuals to eat smaller meals (Polivy
et al., 1986; Roth et al., 2001, although see Thomas, Dourish, & Higgs,
2013 for a recent conference abstract concerning concealment of
eating topography equipment during a study). Given that in some
study designs it may be apparent that food consumption will be re-
corded, the present ndings may be a cause for concern. It is
relatively rare for researchers to explicitly make participants aware
that their food consumption is being recorded (although see Yip et al.,
2013) and the present studies demonstrate why this would be ill
advised. However, it seems likely that if studies do not employ cover
stories or attempt to conceal the monitoring of meal size (e.g.,
Andrade et al., 2012; Mekhmoukh, Chapelot, & Bellisle, 2012; Shah
et al., 2014; Temple, Johnson, Recupero, & Suders, 2010), it could
be the case that participants will become suspicious that their food
intake is being recorded and suppress their food intake, as was the
case in the monitored condition in Study 2. One of the reasons this
may be problematic is because awareness could mask expected
effects on food consumption by producing a formof oor effect (see
Brunner, 2010 for an example of howarticially created oor effects
on food intake can remove hypothesised between condition differ-
ences). In Roth et al. (2001) the well-replicated effect of social norms
inuencing food intake (Robinson, Thomas, Aveyard, & Higgs, 2014)
was removed when an experimenter was present during eating, pre-
sumably causing participants to feel self-conscious about their food
intake and to suppress their intake. It will be important to test
whether the degree to which participants feel their food intake will
be recorded during a study can weaken or remove hypothesised
effects of experimental manipulations on food intake, as we are not
aware of any direct formal testing of this proposition to date.
Another point of relevance is whether all people react differ-
ently to their food intake being monitored. It may be the case that
some participants (restrained vs. unrestrained, or overweight vs.
healthy weight) react differently than others do and this could in
theory also affect conclusions concerning sub-group differences in
a study. For example, we know that overweight individuals under-
report dietary intake in certain contexts (Mela & Aaron, 1997). In
a food intake study, this group may respond in much the same way
and deliberately reduce their food intake when they suspect that
their food intake is being monitored. It is not clear whether com-
pletely removing participant awareness that food consumption is
being monitored would result in participants eating more freely,
because the unmonitored manipulation in Study 2 was ineffec-
tive. One future possibility would be to attempt to make food
consumption appear incidental to what participants would deem
to be the main experimental task. For example, in Hermans et al.
(2010) participants completed a task together and although par-
ticipants were offered food, it was not apparent to participants that
food was the main focus of the study. Although given the con-
straints of the laboratory, moving outside of these settings and testing
in the eld more often may help to minimize these concerns (see
de Castro, 2000; Meiselman, 1992). Another interesting question is
the extent to which awareness might inuence the types of food
eaten (in addition to or instead of the amount of food eaten). In Study
1 we asked participants if they would be likely to eat less from a
buffet if they felt monitored, but not if it would inuence their food
choices. Given that the types of foods people choose to eat have ste-
reotypes attached to them and can be a way of signalling identity
(Berger & Rand, 2008; Guendelman et al., 2011), this question may
also be of relevance, as in some studies participants have access to
a number of different foods.
A limitation of the present studies was that we were unable to
rule out the possibility that participants from Study 1 also as-
sisted in Study 2. Given that different advertisement routes were
adopted and Study 2 took place several months after Study 1, we
do not believe this is a signicant limitation. Moreover, cover story
checks in Study 2 indicated that participants were deceived by the
cover story used and were unaware of the true aims of the re-
search (only 2/72 participants in Study 2 came close to identifying
the study aims). Further research is required to characterise the
generalizability of the present ndings to other laboratory para-
digms, participant samples and foods. For example, it is important
to establish if similar results would be seen in males, given that
females may be particularly likely to eat smaller meals in order to
portray femininity (Mori et al., 1987; Pliner & Chaiken, 1990). Like-
wise, we used a specic paradigm (a taste-test) in Study 2 and only
measured consumption of a high calorie unhealthy snack food
(cookies). Thus, it is unclear whether awareness of consumption of
other food types would have a similarly large effect on intake. In
mitigation, we note that the design of Study 2 was based on our
appreciation that large numbers of laboratory eating-behaviour
studies rely on predominantly female samples (e.g. Koh & Pliner,
2007; Spiegel et al., 1993), employ taste tests (e.g. Boon et al., 2002;
Oldham-Cooper et al., 2011; Robinson et al., 2013) and study the
consumption of energy dense food (e.g. Blass et al., 2006; Bodenlos
& Wormouth, 2013).
We believe that the present ndings have important implica-
tions for howresearchers design laboratory-based eating-behaviour
experiments. Our results demonstrate that it is important to ensure
that participants awareness of monitoring of food intake during a
study is as minimal as possible. For example, a study by Hermans
et al. (2010) observed eating in a semi-naturalistic laboratory and
included the use of cover stories to not only disguise the experi-
mental hypotheses of their study but to also remove attention from
food consumption during that study. Although such measures may
not always be possible, we believe that demand characteristics
deserve attention when designing eating behaviour studies. More-
over, post-study measurement of participant awareness of
behavioural monitoring or self-imposed restriction of eating during
24 E. Robinson et al./Appetite 83 (2014) 1925
that study may provide researchers with additional information
about their experiments. For example, in a similar vein to the method
adopted in Study 2, during the immediate post-study period re-
searchers could measure the extent to which participants believed
that their food consumption was being monitored. Such measures
could act as manipulation checks for any cover stories or be fac-
tored into analyses where appropriate.
Conclusions
Overall, these two studies demonstrate that if participants believe
that the amount of food they are eating during a study is being moni-
tored then they will suppress their food intake, and this may affect
conclusions drawn from such studies.
References
Andrade, A. M., Kresge, D. L., Teixera, P. J., Baptista, F., & Melanson, K. J. (2012). Does
eating slowly inuence appetite and energy intake when water intake is
controlled? International Journal of Behavioural Nutrition & Physical Activity, 21(9).
doi:10.1186/1479-5868-9-135.
Berger, J., & Rand, L. (2008). Shifting signals to help health. Using identity signaling
to reduce risky health behaviors. Journal of Consumer Research, 35, 509518.
Blass, E. M., Anderson, D. R., Kirkorian, H. L., Pempek, T. A., Price, I., & Koleini, M. F.
(2006). On the road to obesity. Television increases intake of high-density foods.
Physiology & Behavior, 88, 597604.
Bodenlos, J. S., & Wormouth, B. M. (2013). Watching a food-related television show
and calorie intake. A laboratory study. Appetite, 61, 812.
Boon, B., Stoebe, W., Schut, H., & Ijntema, R. (2002). Ironic processes in the eating
behaviour of restrained eaters. British Journal of Health Psychology, 7, 110.
Brunner, T. A. (2010). How weight-related cues affect food intake in a modeling
situation. Appetite, 55, 507511.
Conger, J. C., Conger, A. J., Costanzo, P. R., Wright, K. L., & Matter, L. A. (1980). The
effect of social cues on the eating behavior of obese and normal subjects. Journal
of Personality, 48, 258271.
de Castro, J. M. (2000). Eating behavior. Lessons from the real world of humans.
Nutrition (Burbank, Los Angeles County, Calif.), 16, 800813.
Guendelman, M. D., Cheryan, S., & Monin, B. (2011). Fitting in but getting fat. Identity
threat and dietary choice among U.S. immigrant groups. Psychological Science,
22, 959967.
Hermans, R. C., Larsen, J. K., Herman, P. C., & Engels, R. C. M. E. (2010). Modeling of
palatable food intake in female young adults. Effects of perceived body size.
Appetite, 51, 512551.
Higgs, S. (2002). Memory for recent eating and its inuence on subsequent food
intake. Appetite, 39, 159166.
Koh, J., & Pliner, P. (2007). The effects of degree of acquaintance, plate size, and sharing
on food intake. Appetite, 52, 595602.
Laney, C., Kaasa, S. O., Morris, E. K., Berkowitz, S. R., Bernstein, D. M., et al. (2008).
The Red Herring technique. A methodological response to the problem of
demand characteristics. Psychological Research/Psychologische Forschung, 72,
362375.
Meiselman, H. L. (1992). Methodology and theory in human eating research. Appetite,
19, 4955.
Mekhmoukh, A., Chapelot, D., & Bellisle, F. (2012). Inuence of environmental factors
on meal intake in overweight and normal weight male adolescents. A laboratory
study. Appetite, 59, 9095.
Mela, D. J., & Aaron, J. I. (1997). Honest but invalid. What subjects say about
recording their food intake. Journal of the American Dietetic Association, 97,
791793.
Mori, D., Chaiken, S., & Pliner, P. (1987). Eating lightly and the self-presentation
of femininity. Journal of Personality and Social Psychology, 53, 693702.
Oldham-Cooper, R. E., Hardman, C. A., Nicoll, C. E., Rogers, P. J., & Brunstrom, J. M.
(2011). Playing a computer game during lunch affects fullness, memory for lunch,
and later snack intake. American Journal of Clinical Nutrition, 93, 308313.
Orne, M. T. (1962). On the social psychology of the psychological experiment. With
particular reference to demand characteristics and their implications. American
Psychologist, 17, 776783.
Orne, M. T., Whitehouse, W. G., & Kazdin, A. E. (2000). Demand characteristics. In
Encyclopaedia of psychology (pp. 469470). Washington, DC: American
Psychological Association and Oxford University Press.
Pliner, P., & Chaiken, S. (1990). Eating, social motives, and self-presentation in women
and men. Journal of Experimental Social Psychology, 26, 240254.
Polivy, J., Herman, C. P., Hackett, R., & Kuleshnyk, I. (1986). The effects of self-attention
and public attention on eating in restrained and unrestrained subjects. Journal
of Personality and Social Psychology, 50, 12531260.
Robinson, E., Benwell, H., & Higgs, S. (2013). Food intake norms increase and decrease
snack food intake in a remote confederate study. Appetite, 65, 2024.
Robinson, E., Thomas, J., Aveyard, P., & Higgs, S. (2014). What everyone else is eating.
A systematic reviewand meta-analysis of the effect of informational eating norms
on eating behavior. Journal of the Academy of Nutrition and Dietetics, 114, 414429.
Rolls, B. J., Roe, L. S., Halverson, K. H., & Meengs, J. S. (2007). Using a smaller plate
did not reduce energy intake at meals. Appetite, 49, 652660.
Roth, D. A., Herman, C. P., Polivy, J., & Pliner, P. (2001). Self-presentational conict
in social eating situations. A normative perspective. Appetite, 36, 165171.
Salvy, S. J., Jarrin, D., Paluch, R., Irfan, N., & Pliner, P. (2007a). Effects of social inuence
on eating in couples, friends and strangers. Appetite, 42, 9299.
Salvy, S. J., Coelho, J. S., Kieffer, E., & Epstein, L. H. (2007b). Effects of social contexts
on overweight and normal-weight childrens food intake. Physiology & Behavior,
92, 840846.
Shah, M., Copeland, J., Dart, L., Adams-Huet, B., James, A., & Rhea, D. (2014). Slower
eating speed lowers energy intake in normal-weight but not overweight/obese
subjects. Journal of the Academy of Nutrition and Dietetics, 114, 393402.
Spiegel, T. A., Kaplan, J. M., Tomassini, A., & Stellar, E. (1993). Bite size, ingestion rate,
and meal size in lean and obese women. Appetite, 2, 131145.
Stunkard, A. J., & Messick, S. (1985). The three-factor eating questionnaire to measure
dietary restraint disinhibition and hunger. Journal of Psychosomatic Research, 29,
7183.
Temple, J. L., Johnson, K., Recupero, K., & Suders, H. (2010). Nutrition labels decrease
energy intake in adults consuming lunch in the laboratory. Journal of the American
Dietetic Association, 110, 10941097.
Thomas, J. M., Dourish, C. T., & Higgs, S. (2013). Monitoring eating behaviour in the
laboratory. Do we need to do it covertly? Appetite, 71, 487.
Vartanian, L. R., Herman, C. P., & Polivy, J. (2007). Consumption stereotypes and
impression management. How you are what you eat. Appetite, 48, 265277.
Yip, W., Wiessing, K. R., Budgett, S., & Poppitt, S. D. (2013). Using a small dining plate
does not suppress food intake from a buffet lunch meal in overweight,
unrestrained women. Appetite, 69, 102107.
25 E. Robinson et al./Appetite 83 (2014) 1925