cruwys, 2014

Research report

Social modeling of eating A review of when and why social influenceaffects food intake and choiceTegan Cruwys a Kirsten E Bevelander b Roel CJ Hermans c

a School of Psychology University of Queensland St Lucia 4072 Australiab Communication Science Department Behavioural Science Institute Radboud University Nijmegen PO Box 9104 6500 HE Nijmegen The Netherlandsc Developmental Psychopathology Department Behavioural Science Institute Radboud University Nijmegen PO Box 9104 6500 HE Nijmegen TheNetherlands

A R T I C L E I N F O

Article historyReceived 30 May 2014Received in revised form 23 August 2014Accepted 26 August 2014Available online

KeywordsEatingSocial influenceNormative influenceSocial normsHealth behaviorObesity

A B S T R A C T

A major determinant of human eating behavior is social modeling whereby people use othersrsquo eatingas a guide for what and how much to eat We review the experimental studies that have independentlymanipulated the eating behavior of a social referent (either through a live confederate or remotely) andmeasured either food choice or intake Sixty-nine eligible experiments (with over 5800 participants) wereidentified that were published between 1974 and 2014 Speaking to the robustness of the modeling phe-nomenon 64 of these studies have found a statistically significant modeling effect despite substantialdiversity in methodology food type social context and participant demographics In reviewing the keyfindings from these studies we conclude that there is limited evidence for a moderating effect of hungerpersonality age weight or the presence of others (ie where the confederate is live vs remote) Thereis inconclusive evidence for whether sex attention impulsivity and eating goals moderate modeling andfor whether modeling of food choice is as strong as modeling of food intake Effects with substantial ev-idence were modeling is increased when individuals desire to aliate with the model or perceivethemselves to be similar to the model modeling is attenuated (but still significant) for healthy-snackfoods and meals such as breakfast and lunch and modeling is at least partially mediated through be-havioral mimicry which occurs without conscious awarenessWe discuss evidence suggesting thatmodelingis motivated by goals of both aliation and uncertainty-reduction and outline how these might be the-oretically integrated Finally we argue for the importance of taking modeling beyond the laboratory andbringing it to bear on the important societal challenges of obesity and disordered eating

copy 2014 Elsevier Ltd All rights reserved

Introduction

The consumption of food has implications beyond merely pro-viding nutrients and energy needed to sustain life Food and eatingare also intertwined with our social lives Most eating takes placesin the presence of other people and is often perceived as an enjoy-able part of our cultural experience (Rozin 2005) Therefore it shouldnot be surprising that onersquos eating behavior is profoundly affectedby social factors In addition to processes such as social facilita-tion and impression management (also reviewed in this issue ofAppetite) another social influence phenomenon is modeling of foodintake whereby people directly adapt their food intake to that oftheir eating companion It was forty years ago that evidence first

began to accumulate that modeling1 is a primary determinant ofeating behavior Nisbett and Storms (1974) demonstrated that youngmales consistently atemore when their eating companion ate a largenumber of crackers and less when the other person ate minimally(compared to when eating alone) This so-called modeling effectcaught the attention of other researchers and in subsequent yearsseveral other attempts were made to identify boundary condi-tions for the effect This early modeling research was influenced bythe externality hypothesis (Schachter 1971) which stated that over-weight people are more vulnerable to external food-related cues(such as the social environment) rather than internal cues (such ashunger or satiety) However and in accordance with the work ofNisbett and Storms (1974) no differences were found betweenhealthy and overweight people or between restrained and unre-strained eaters in their extent of modeling (Conger Conger CostanzoWright amp Matter 1980 Polivy Herman Younger amp Erskine 1979Rosenthal amp Marx 1979 Rosenthal amp McSweeney 1979) Research-ers therefore concluded that Schachterrsquos externality hypothesiscannot distinguish between healthy-weight and overweight people

Corresponding authorE-mail address tcruwysuqeduau (T Cruwys)

1 This review uses the term modeling to refer to social modeling that is behav-ioral conformity of eating not statistical modeling

httpdxdoiorg101016jappet2014080350195-6663copy 2014 Elsevier Ltd All rights reserved

Appetite (2014) ndash

ARTICLE IN PRESS

Please cite this article in press as Tegan Cruwys Kirsten E Bevelander Roel CJ Hermans Social modeling of eating A review of when and why social influence affects foodintake and choice Appetite (2014) doi 101016jappet201408035

Contents lists available at ScienceDirect

Appetite

journal homepage wwwelseviercom locate appet

in the case of modeling because both groups are influenced bynormative external cues (Herman amp Polivy 2008) Instead theseeffects were found to have a strong and pervasive influence on bothhealthy-weight and overweight individualsrsquo eating behaviors Al-though the reproducibility of these effects was easily and repeatedlydemonstrated the question of why modeling occurs has provedmoredicult to answer definitively That is what purpose does model-ing serve psychologically that might explain why it is so stronglypreserved and generalizable

Over the decades of modeling research a variety of explana-tions have been put forward to understand the effect The mostdominant interpretation however is that modeling of food intakeis an example of a broader phenomenon of social influence and thatgeneral theories of normative behavior might help to understandwhy people adapt their food intake to that of others Using a nor-mative approach Herman and his colleagues proposed that theprincipal regulatory influence on eating in social contexts is pe-oplersquos beliefs about what or howmuch is appropriate to eat (Hermanamp Polivy 2005 Herman Roth amp Polivy 2003b) According to thismodel people conform to othersrsquo eating because they see the amounteaten by others as an indicator of how much one can or should eatwithout eating excessively

Although the literature seems to approach consensus on theutility of this normative model there has not been a systematicreview of modeling studies The lack of a comprehensive reviewimpedes our ability to ascertain from the extant modeling litera-ture (a) when and why social modeling shapes eating behaviorand (b) how to translate this knowledge to inform applied prac-tice aimed at increasing healthy eating behavior Therefore ouroverarching aim is to review the literature on how peoplersquos foodchoice and intake is affected by modeling and on the basis of thesefindings propose new research directions that might help us togain insight into the robustness or underlying mechanisms of mod-eling We start by reviewing typical methodological approaches tothe study of modeling before summarizing the key findings fromour systematic review of 69modeling experiments We then discusstheoretical and practical implications of these findings

Modeling methodological approaches

In past research several strategies have been used to investi-gate modeling effects on eating Both observational as well ascorrelational studies have found that people adapt their intake tothat of their eating companion and that those who are eating to-gether converge upon an eating norm (eg Salvy Romero Paluchamp Epstein 2007c Salvy Vartanian Coelho Jarrin amp Pliner 2008b)This occurs such that the variance among participants in their foodintake is reduced when eating together However both statisticaland theoretical concerns arise when interpreting research whereparticipants model one another Firstly because food intake is non-independent between participants an appropriate statistical methodof analysis would be multi-level modeling (Luke 2004) ndash al-though often this is not performed Furthermore without randomassignment it is dicult to rule out the possibility that non-socialfactors such as pre-existing similarity or eating attitudes are re-sponsible for conformity effects between eating companions Finallyin a scenario in which both co-eaters are free to choose the typeor amount of food to consume it is dicult to determine whichperson is modeling and which person is being modeled In partbecause of these concerns an experimental design in which theintake andor choice of one co-eater (ie the confederate) is pre-determined by the experimenter has arguably become the gold-standard for research on the modeling of food intake This paradigmenables researchers to investigate modeling behavior without anypotential confounds related to selection or non-social processes Insome studies participants are providedwith a non-food related cover

story for the experiment (eg Bevelander Anschuumltz CreemersKleinjan amp Engels 2013a Cruwys et al 2012 Hermans SalvyLarsen amp Engels 2012c) In these experiments participants believethat food is incidental to the research question In other studies par-ticipants are told that they are participating in a taste-test studyand are asked to complete questionnaires related to their experi-ence of the food items (eg Goldman Herman amp Polivy 1991Vartanian Sokol Herman amp Polivy 2013) In these studies partici-pants are aware of the centrality of the food to the experimenthowever the researchersrsquo interest in social influence and the amountof food consumed remains opaque

The sheer robustness of modeling has allowed researchers toalso develop a more ldquolight-touchrdquo technique for communicatingsocial norms to participants known as the remote-confederateparadigm (cf Roth Herman Polivy amp Pliner 2001) In studies uti-lizing this design the confederate providing the norm for foodchoice or intake is not physically present Rather participants areprovided normative information (while concealing the aim ofthe study with a cover story) by exposing them to either writteninformation about the amount consumed by previous partici-pants (eg in the form of a list on a table which was supposedlyused to determine how much food needed to be ordered by theexperimenters) or by exposure to a remotemodel selecting or eatingfood on a video or computer screen (Bevelander et al 2013aBevelander Anschuumltz amp Engels 2012b Hermans et al 2012cRomero Epstein amp Salvy 2009) Given that both live and remoteconfederate designs have been found to induce modeling effectson eating (cf Feeney Polivy Pliner amp Sullivan 2011) and are ableto infer strong cause and effect relationships we summarize find-ings of studies in which the eating norm is induced by either typeof confederate

Inclusion criteria

To find relevant English-language empirical research on mod-eling effects on food choice and food intake a literature search ofPubMed and Google Scholar was conducted using the following keywords lsquomodelingrsquo lsquomatchingrsquo lsquosocial influencersquo lsquonormative influ-encersquo lsquoeatingrsquo lsquofood choicersquo lsquofood intakersquo These key words wereused in combinations of two to include one theoretical keyword (iemodeling matching social influence normative influence) and onebehavioral keyword (ie eating food choice food intake) The ref-erence lists and citations of eligible publications were also reviewedto identify pertinent literature A criterion for inclusion in the reviewwas that the study had an experimental design in which either foodchoice or food intake was experimentally manipulated by a socialreferent (using either a live or remote confederate) Studies in whichparticipant dyads or groups were examined in a free eating para-digmwithout a confederate were therefore not included (eg SalvyJarrin Paluch Irfan amp Pliner 2007b Salvy Kieffer amp Epstein 2008a)Furthermore we included only those studies with a dependent vari-able that was amount of food consumed or food choice (measuredin a concrete behavioral fashion not intentions only) Table 1 showsa complete list of all the modeling studies that were included inthis review Where possible however we also discuss studies in ourreview that did not meet our inclusion criteria but which provid-ed additional insight into the dynamic process of modeling Sixty-nine studies (in 49 articles) were identified that met these selectioncriteria reporting on over 5800 experimental participants Of thesethe majority (58) measured food intake or whether participants ateat all as the dependent variable of interest whereas only 11 in-vestigated participantsrsquo choice between at least two food alternativesAs can be seen in Table 1 studies conducted with live confeder-ates (42) or with some form of remote confederate (27) are wellrepresented

ARTICLE IN PRESS


2 T Cruwys et alAppetite (2014) ndash

Table 1Summary of 69 modeling studies identified for the review

Study Authors Year Outlet Design N Participantage

Participantgender

Modellive (L) vsremote (R)

DV amount(A) vschoice (C)

Evidenceofmodeling

Moderatorsmediators identified

1 Nisbett andStorms

1974 Book 3 (alone low norm high norm) 3(participant weight)

Young adult M L A Crackers Y Weight status (underweight healthyoverweight) did not moderate

2 Harper andSanders

1975 Journal ofExperimentalChildPsychology

S1 2 (mother vs stranger model) 2(model ate food vs offered only)

80 1ndash4 Boys + girls L C Novel foodsBlue tortillawithham + cheesemacadamiadate

Y Familiarity Increased willingness totry foods that the childrsquos motherrather than a stranger had modeledconsuming

3 S2 2 (male vs female strangermodel) 2 (model ate food vs offeredonly)

140 3ndash4 Boys + girls L Y

4 Rosenthal andMarx

1979 AddictiveBehaviors

3 (participant dieting status) 3 (lownorm high norm no-model)

81 18ndash56 F L A Crackers Y Dieting status (successfulunsuccessful non-dieter) did notmoderate

5 Rosenthal andMcSweeney


S1 2 (modelrsquos eating rate slow vsfast) 2 (participant weight)

31 17ndash24 F L Rate of eating Y Weight status Overweightparticipants did not model as much ashealthy weight participants in fastintake conditions

6 S2 2 (participant sex) 2 (modelrsquossex) 2 (low norm high norm)

79 17ndash28 M + F L A Crackers Y Sex of model High-intake femalemodel had less influence

7 Polivy HermanYounger andErskine

1979 Journal ofPersonality

2 (dieting status of model) 2 (low vshigh norm) 2 (participant restrainthigh vs low)

86 Young adult F L A Sandwichquarters

Y Participant restraint and modelrsquosdieting status did not moderate

8 Conger CongerCostanzoWright andMatter


2 (participant sex) 2 (modelrsquos sex) 2(participant weight) 3 (no intakenorm low norm high norm) + 4controls

114 Young adult M + F L A Crackers Y Sex Stronger modeling effect for maleparticipantsSimilarity Subjects ate more whenmodel was of same-sexWeight status did not moderate

9 Birch 1980 ChildDevelopment

Intervention to seat children withchildren who had opposite foodpreferences for 4 days vs nointervention

39 3ndash10 Boys + girls L C Preferred vsnon-preferredfoods

Y Age Younger children showed moremodeling

10 GoldmanHerman andPolivy

1991 Appetite S1 3 (hunger low moderate high) 2(low norm high norm)

86 Young adult F L A Bite-sizesandwichesfruit cookies

Y Hunger did not moderateFood type No modeling of fruit intake

11 S2 (hunger low high) 3 (controllow norm high norm)


Y

12 Hendy andRaudenbush

2000 Appetite S2 and S3 Silent teacher model eatingvs not eating

34 M = 47 Boys + girls L C Lunch mealfamiliar foods

N No evidence of modeling when teacherate but did not interact with childrenSimilarity Peers modeled more thanteachers (S5)

13 23 M = 44 Boys + girls L C Lunch mealunfamiliarfoods

N

14 S4 Teacher model enthusiasticallyeating vs non-eating teacher

26 M = 44 Boys + girls L A Mangos andcranberries

Y

15 S5 Enthusiastic teacher vs peer 14 M = 43 Boys + girls L A Unfamiliarfruit

Y

16 Roth HermanPolivy andPliner

2001 Appetite 2 (alone vs observed) 3 (norm nolow high)

134 M = 23 F R A Cookies Y Presence of others Modeling did notoccur when participants observed

17 Johnston 2002 Social Cognition S1 2 (model obese vs non-obese) 2(low norm high norm)

48 Unclear F L A Ice cream Y Similarity Normal weight participantswere not influenced by obese modelFacial birthmark of model did notmoderate

18 S2 2 (model appearance birthmarkvs no-birthmark) 2 (low norm highnorm)

84 Unclear F L A Ice cream Y

(continued on next page)

AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

3TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling


19 Hendy 2002 Appetite Intervention participants exposed tono model or model (boy vs girl)modeling acceptance of novel food

22 3ndash6 Boys + girls L A Papayacranberrydried apple

Y Sex of model Females were modeledmore closely

20 Horne et al 2004 EuropeanJournal ofClinicalNutrition

Intervention participants exposedover 16 days to video of heroic peerswho enjoy eating fruit and vegetablesvs no intervention

749 5ndash11 Boys + girls R A Fruit andvegetableintake

Y

21 Pliner andMann

2004 Appetite S1 3 (no norm low norm highnorm) 2 (palatable vs unpalatablefood available)

72 M = 199 F R A Cookies Y Food type Modeling only occurred forfood intake (S1) and not food choice(S2) modeling only occurred forpalatable but not unpalatable food22 S2 Model chose palatable vs

unpalatable +control37 M = 193 F R C Cookies N

23 AddessiGallowayVisalberghiand Birch

2005 Appetite Familiar adult eating nothing vs eatingnovel different colored food vs samecolored food as participant

27 2ndash5 Boys + girls L A Coloredsemolina

Y Similarity Children only modeledwhen food color matched that of theadult model

24 Leone Plinerand Herman

2007 Appetite S1 2 (number of prior participants 3vs 9) 2 (low norm high norm)

75 M = 193 M + F R A Distributionof cookieseaten (notdirectlycomparable toother studies)

Unclear Ambiguity Variation in intake amongmodels reduces social influence leadsto disinhibitionSex and restraint status did notmoderate

25 S2 4 norm conditions (low norm highnorm mixed lowhigh norm verymixed lowmediumhigh norm)

114 M = 208 M + F R Y

26 YamasakiMizdzuno andAoyama

2007 JapaneseJournal of SocialPsychology

Low norm vs high norm 45 M = 189 Female R A Donuts Y Modeling persisted even thoughparticipants believed that their eatingcould not be observed by theexperimenter

27 HermansLarsenHerman andEngels

2008 Appetite 3 (no eating norm low norm highnorm) 2 (model weight slim vsnormal weight)

102 M = 2050 F L A MampMs Y Similarity Participants more likely tomodel normal weight confederate

28 Greenhalgh etal

2009 Appetite Four eating occasions 3 (modelpositive and ate novel food vs negativenon-eating model and later positiveeating model vs control)

35 5ndash7 Boys + girls L A Colorednovel foodsand othersnack foods(ie grapescheese pittabread andcarrot)

Y Negative modeling can be overriddenby later positive modeling (in 5ndash7 yearolds)

29 44 3ndash4 Boys + girls L Y

30 RomeroEpstein andSalvy

2009 Journal ofAmericanDieteticAssociation

2 (participant weight) 2 (low normhigh norm)

44 8ndash12 Girls R A Cookies(S1 + S2)

Y Weight status did not moderate


2009 Appetite 3 (no eating norm low norm highnorm) 2 (model weight) + controlcondition

116 M = 2028 F L A Vegetables Y Model weight status did not moderateHealthy food Size of effect possiblyattenuated for healthy food

32 HermansEngels Larsenand Herman

2009 Appetite 2 (low norm high norm) 2(confederate social nature warm vscold) + control condition

100 18ndash27 F L A MampMs Y Social nature of model Onlyparticipants exposed to coldconfederate modeled

33 McFerranDahlFitzsimons andMorales

2010 Journal ofConsumerResearch

2 (model weight) 2 (low norm highnorm) + 1 control (no confederate)

115 Young adult F L A Candy Y Similarity Less modeling of obeseconfederate


AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

4TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling


34 Burger et al 2010 Journal of Socialand ClinicalPsychology

Healthy norm unhealthy norm control 120 Young adult F R C Healthyversusunhealthy

Y Presence of others did not moderate

35 75 Young adult F R C Healthyversusunhealthy

Y

36 HermansHermanLarsen andEngels

2010 Appetite 3 (no eating norm low norm highnorm)

59 M = 2173 M L A Nuts Y Hunger Modeling only apparent in theconditions where men were fooddeprived


2010 Journal of theAmericanDieteticsAssociation

3 (no eating norm low norm highnorm)

57 M = 2115 F L A Breakfastfoods

Y Meal type Size of effect possiblyattenuated for breakfast foods

38 Brunner 2010 Appetite 2 (low norm high norm) 2 (weightcues present vs not-present)

54 M = 208 F L A Chocolate Y Weight-related cues Eating wassuppressed by weight-related cuessuch that modeling no longer occurred

39 51 M = 213 F L Y

40 Feeney PolivyPliner andSullivan

2011 Eatingbehaviors

3 (no confederate remote confederatelive confederate)

32 M = 186 F Manipulated A Pizza Y Remote vs live confederate did notmoderate

41 BevelanderAnschuumltz andEngels

2011 Appetite 3 (model buys low average or highcalorie products)

89 10ndash12 F L A Total energyin foodpurchases

Y

42 Exline ZellBratslavskyHamilton andSwenson

2012 Journal of Socialand ClinicalPsychology

2 (participant sociotropy low vshigh) 2 (low norm high norm)

109 M = 186 M + F L A Candy Y Sociotropy Modeling enhanced amongparticipants more concerned withmaintaining social harmony


2012 British Journalof Nutrition

2 (model portion size small vs large) (model intake low medium large)

85 M = 2085 F L A Meal Y Source of norm (portion size vs modelintake) did not moderate


2012 Appetite No eating norm low norm high norm 223 6ndash11 M + F L A Snack foods Y Overweight participants were moreresponsive to high norm conditionhealthy weight participants were moreresponsive to no eating normconditionTime delay did not moderate(participants modeled bothimmediately and in delayed testingsession)

45 Stok de Ridderde Vet and deWit

2012 Psychology ampHealth

2 (low frequency norm high frequencynorm) 2 (low identified vs highidentified with referencegroup) + control

119 M = 217 M + F R A Fruit Y Group membership High identifierswith referent group showed moremodeling of majority norm (anddivergence from minority norm)

46 HowlandHunger andMann

2012 Appetite 2 (low intake norm control) 44 18ndash29 yrs M + F L A Snacks Y Presence of others did not moderate47 47 18ndash26 M + F L Y



Familiar food norm vs unfamiliar foodnorm vs control

316 M = 713 M + F R C Snacks(computerbased)

Y Peer increased willingness to tryunfamiliar foods but children preferredhigh energy dense foods

49 Cruwys et al 2012 Appetite 2 (ingroup vs outgroup model) 2 (noeating norm high norm) + control

119 17ndash25 F L A Popcorn Y Group membership Modeling did notoccur for outgroup confederate


AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

5TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling


50 HermansSalvy Larsenand Engels


S1 no eating norm eating norm 77 M = 2029 F R A Candy N Shared social context No evidence ofmodeling when confederate in adifferent situation to participant

51 S2 no eating norm low norm highnorm

51 M = 2043 F R A MampMs N

52 Prinsen deRidder and deVet

2013 Appetite S1 and S2 food wrappers of previousparticipant present vs absent

144 Unclear M + F R Food intake (Yor N)

Y Healthy goal prime did not moderate

53 65 M = 2158 M + F R (S1 and S2) Y54 S3 wrapper of unhealthy vs healthy

snack present90 M = 2187 M + F R C Healthy or

unhealthy inStudy 3

Y

55 RobinsonBenwell andHiggs

2013 Appetite Low norm high norm control 64 M = 192 F R A Cookies Y Trait empathy did not moderate

56 Mollen RimalRuiter and Kok

2013 Appetite Norm type (healthy descriptive normunhealthy descriptive norm healthyinjunctive norm) + control

231 17ndash34 M + F R C Salad orburger

Y Type of norm Descriptive norm moreeffective than injunctive norm

57 Robinson andHiggs


Unhealthy model vs healthy model vscontrol

100 M = 199 F L C Low energydense and highenergy densefoods

Y Food type Modeling was mostnoticeable for low-energy dense food ndashparticipants were readily influenced tonot choose these foods

58 Hermans et al 2013 British Journalof Nutrition

No eating norm low norm high norm 85 M = 2020 F L A MampMs Y Impulsivity Modeling attenuated forthose high in self-reported impulsivityAttention to eating cues and responseinhibition did not moderate

59 BevelanderMeiselmanAnschuumltz andEngels

2013 Appetite 2 (no intake vs standardizedintake) 3 (happy sad or neutralmovie)

110 7ndash10 M + F L A Candy Y Current mood Participants modelednorm but not in neutral videocondition

60 BevelanderAnschuumltzCreemersKleinjan andEngels

2013 PLoS One No eating norm low norm high norm 118 M = 1114 M + F R A Candy Y Children model via video interactionwith remote confederateSelf-esteem High implicit self-esteemand low body esteem associate withgreater modeling

61 VartanianSokol Hermanand Polivy

2013 PLoS One Low norm high norm control 78 Young F R A Cookies (S1and S2)

Y Modeling mediated by perceived normfor appropriate intake in 3 studies62 126 adults F L Y

63 94 F L A MampMs Y64 Florack Palcu

and Friese2013 Appetite 2 (regulatory focus prevention vs

promotion) 2 (no eating norm vseating norm

142 18ndash49 M + F L A Cookies Y Regulatory focus Modeling effect wasmore pronounced when participantshad a prevention focus

65 40 M = 2949 F R A Ice cream Y


2014 British Journalof HealthPsychology

Descriptive norm injunctive normcontrol

80 14ndash17 yrs M + F R A Fruit Y Type of norm Descriptive norm wasmore effective than injunctive norm

67 Salmon Fennisde RidderAdriaanse andde Vet

2014 HealthPsychology

2 (high control low control) 2(healthy descriptive norm [lsquosocialproof heuristicrsquo] no norm control)

177 M = 2047 M + F R C Healthy vsunhealthy

Y Self-control Low self controlassociated with greater modeling

68 RobinsonFleming andHiggs

Inpress

HealthPsychology

Pro-veg norm vs health message(control)

71 M = 196 M + F R A Fruit andvegetables (S1and S2)

Y Type of norm Descriptive norm wasmore effective than injunctive normFood preferences Low consumers offruit and veg were influenced by norm

69 Pro-veg descriptive norm vs pro-veginjunctive norm vs health message(control)

70 M = 191 M + F R Y

AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

6TCruw

yset

alAppetite

(2014)

ndash

Review of the literature

Robustness of modeling

One immediate conclusion that can be drawn from these 69studies is that social modeling is a profound and robust phenom-enon that can determine what and how much people consume Ofthe 69 studies that were reviewed only five studies (in three ar-ticles) found limited evidence of modeling effects on food choiceor intake (Hendy amp Raudenbush 2000 Hermans et al 2012c Plineramp Mann 2004) This is despite diverse samples including males andfemales a wide range of ages ethnicity weight and restraint statusand hungry and satiated individuals Furthermore it emerges thatmany efforts to establish boundary conditions for modeling havefailed For instance researchers have hypothesized that modelingmight be moderated by a personrsquos body weight or sex (Conger et al1980 Nisbett amp Storms 1974) dieting status (Rosenthal amp Marx1979) personality factors (Herman Koenig-Nobert Peterson ampPolivy 2005) and hunger (Goldman et al 1991) and in all cases ithas been found that these variables did not moderate the strengthof modeling Below we review key conclusions that can be drawnfrom the 69 identified studies grouped broadly into sections on con-textual factors individual factors and social factors We aim toprovide insight into the circumstances under which modeling op-erates and how the magnitude of the effect can be affected by avariety of factors

Contextual factors

Type of foodMost studies examining modeling of food intake among adults

as well as young people have largely focused on the intake of high-energy-dense palatable foods (snacks) such as small cookies (LeonePliner amp Herman 2007 Roth et al 2001) chocolate coated peanuts(Bevelander Meiselman Anschuumltz amp Engels 2013d Hermans LarsenHerman amp Engels 2008) popcorn (Cruwys et al 2012) and icecream (Florack Palcu amp Friese 2013 Johnston 2002) These studieshave all found the same pattern people eat more or less when theireating companions eat more or less of these snack foods Given thesubstantial number of such studies it seems safe to conclude thatpeople model their intake of energy-dense snack food on that ofothers

Although modeling effects on vegetable and fruit consumptionhave been found among children and adults (Horne et al 2004Howland Hunger amp Mann 2012 McFerran Dahl Fitzsimons ampMorales 2010a Robinson amp Higgs 2013 Salvy et al 2008a) thereis some evidence that people are less likely to model their eatingpartner for healthy or unpalatable foods For example Hermans andcolleagues (Hermans Larsen Herman amp Engels 2009b) found thatthe size of the effect of modeling was small when participants wereoffered cucumber and carrots and three studies have found no ev-idence of modeling for healthy foods (Goldman Herman amp Polivy1991 S1 amp S2 Pliner amp Mann 2004) In children the majority ofstudies have focused on modeling to encourage consumption ofnovelnon-preferred low-energy-dense foods (Reverdy ChesnelSchlich Koumlster amp Lange 2008) These studies have utilized varioustypes of models including live or remote peer models (Birch 1980Greenhalgh et al 2009 Horne et al 2004) (un)familiar adult models(Addessi Galloway Visalberghi amp Birch 2005 Harper amp Sanders1975) and teacher models (Hendy 2002 Hendy amp Raudenbush2000) Although modeling effects do occur in all but two of thesestudies repeated exposure was often needed to maintain the effectwhereas a study using snack food showed that children readilymodeled and social influence was maintained a few days later aftera single exposure (Bevelander et al 2012b Bevelander EngelsAnschuumltz amp Wansink 2013b)

Notably while there is considerable literature onmodeling effectson food intake much less is known about modeling of food choicesfor example when both low- and high-energy-dense foods areoffered We identified only 11 studies with a dependent variable offood choice Although the majority of these studies have shown thatmodeling does occur (eg Mollen Rimal Ruiter amp Kok 2013 Prinsende Ridder amp de Vet 2013 Salmon Fennis de Ridder Adriaanse ampde Vet 2014) three studies found no significant modeling effectson food choice (Hendy amp Raudenbush 2000 S2 and S3 Pliner ampMann 2004 S2) However given that these studies were statisti-cally underpowered (particularly given the dependent variable isbinary Ferraro amp Wilmoth 2000) we do not want to overstate theimportance of these null findings

Nevertheless theoretical reasons have been suggested for whymodeling of food choice may be less prominent than modeling offood intake (Pliner ampMann 2004) That is it has been proposed thatpeople may feel more certain about their food likes and dislikes thanthey do about the appropriate amount of consumption in variouscircumstances and therefore do not look to others for guidance indetermining their choice An example of where peoplersquos pre-existing personal preferencesmight reducemodeling is when peoplehave clear eating routines or scripts regarding regular meals suchas breakfast and lunch These scripts reflect what people have learnedis an appropriate expected or desirable amount to consume andunder these circumstances people may be less susceptible to newnormative information This line of reasoning is supported by thefindings of Hermans Herman Larsen and Engels (2010a) who foundthat breakfast intake was affected by the low- and no-intake normbut not by the high-intake norm The absence of the standard small-large modeling effect might indicate that these females were lesssusceptible to the normative information conveyed by the large-intake model In line with this it has been found that lunch intakewas less influenced by others compared to the intake of palatablesnack food (Clendenen Herman amp Polivy 1994 Salvy Elmo NiteckiKluczynski amp Roemmich 2011) and that choices of lunch foods wereless influenced than choice of snack foods (Bevelander Anschuumltzamp Engels 2011)

Notwithstanding these considerations it should be clear thatmodeling persists in the context of meals (de Castro amp Brewer 1992Hermans Larsen Herman amp Engels 2012a Horne et al 2009)Wepropose however that degree of certainty is the critical modera-tor here whereby people model to a lesser extent when they alreadyhave strong established preferences routines or norms within a par-ticular eating context For instance the consumption of breakfastis often based on preferences and social norms a person might havelearned across many years whereas snacking behavior may be lessroutinized People may therefore be less reliant on the new nor-mative information conveyed by the intake of the model as a meansto reduce uncertainty regarding how much one should appropri-ately consume Given the lack of research on this topic howeverit is dicult to ascertain whether different mechanisms may un-derlie modeling of food choice and intake and whether modelingof food choice is less prominent than modeling of food intake Wereturn to this issue in the Theoretical Implications section

Live versus remote confederateAs can be seen in Table 1 modeling has been studied both using

live confederates (42) as well as using some form of remote con-federate (27) Both types of model have been found to influenceeating behavior that is people adapt their intake to both live andremote confederates (cf Feeney et al 2011) Although live andremote confederates were originally quite distinct categories (con-federate physically present vs not) a number of recent studies blurthis distinction For instance studies have utilized a video confed-erate (Hermans et al 2012a 2012b 2012c Romero et al 2009)social media (Bevelander et al 2013a) or participants speaking with

ARTICLE IN PRESS


7T Cruwys et alAppetite (2014) ndash

a live confederate but not observing the confederate eating (Cruwyset al 2012) These variants have generally found evidence of thesame modeling effect as in live confederate studies Yet it is worthnoting that two (Hermans et al 2012c S1 and S2) of the three studiesthat utilized a remote video confederate did not find any evidenceof modeling In these two studies the confederate was shown in adifferent environment than the participants and ate a different kindof snack food than was available to the participants which perhapscreated contextual differences that were too large for modeling tooccur (Hermans et al 2012c) In a related finding a study with chil-dren found that participants modeled more closely when the modelate the same color food as the participants ndash that is when the con-textual differences were reduced (Addessi et al 2005) Thereforeit seems likely that these non-significant effects reveal the impor-tance of shared social context rather than the physical presence ofthe model being necessary

The success of the remote confederate paradigm has very im-portant implications for our understanding of modeling (rather thanmerely being a more convenient experimental design) More spe-cifically one motive that has been proposed for modeling is thatindividuals might model in order to aliate or ingratiate them-selves with others (Herman et al 2003b Hermans Engels Larsenamp Herman 2009a Robinson Kersbergen Brunstrom amp Field 2014aRobinson Tobias Shaw Freeman amp Higgs 2011) That is peopleattempt to become more attractive or likable to another personthrough modeling However participants adhere to the social normprovided by remote confederates evenwhen eating alone when theybelieve their food intake cannot be observed by researchers andwhen they do not expect to have any future interaction with themodel (Burger et al 2010 Roth et al 2001 Yamasaki Mizdzunoamp Aoyama 2007) Ergo it is implausible that people model purelyto elicit social approval or achieve liking Indeed several research-ers have argued that it is more likely that people model becauseothers provide a point of reference in uncertain situations about whatconstitutes appropriate eating behavior (Cruwys et al 2012 Hermanamp Polivy 2005 Robinson Sharps Price amp Dallas 2014b) This is animportant point that we revisit in the Theoretical Implications sectionbelow

Individual factors

Individual differences that could potentially affect modeling aremultifold In this section we review those that have received themost research attention to date specifically hunger and satiety sexage body weight and the traits of impulsivityself-control and goalsrelated to eating

Hunger and satietyAn early explanation of modeling effects the zone of biological

indifferences model (Herman amp Polivy 1988) proposed that hungerwould moderate social influence on eating This model stated thatbiological signals are not typically a primary determinant of eatingbehavior only becoming important at the extremes of hunger andsatiety (Heatherton Polivy amp Herman 1991) In the context of socialinfluences on eating however not much evidence has been foundthat supports the idea that modeling is moderated by hunger Thatis it has been found that modeling persists even in circumstanceswhere individuals are very hungry (Goldman et al 1991) or veryfull (Herman et al 2003a) Furthermore in many experimentalstudies subjective hunger ratings (measured either before or afterthe study) have been included as covariates in the analyses Onlyone of these studies has found a moderating effect of hunger Thatis Hermans and colleagues (Hermans Herman Larsen amp Engels2010b) found thatmales who at the end of the experimental sessionreported high pre-experimental hunger were more likely to adjusttheir intake to that of their eating companion ndash therefore hunger

had the opposite effect to that proposed by the zone of biologicalindifferences model An important limitation however is that thesample size of this study is small and therefore lacks sucient sta-tistical power to draw firm conclusions Moreover given that thisis the only study that has found an effect of hunger on the likeli-hood of modeling and the findings have not yet been replicated atthis stage it seems safe to conclude that social influences on eatingare not moderated by onersquos level of hunger or satiety

SexTo date a considerable amount of literature has been pub-

lished investigatingmodeling effects amongwomen suggesting thatwomen generally adapt their food intake to that of others Al-though there are many studies that have included males in theirdesign (32 out of 69 reviewed) only two studies have recruited amale-only sample (Hermans et al 2010b Nisbett amp Storms 1974)Very few studies have been conducted with sucient power tocompare male and female participants This is in part for theoret-ical reasons such as the much greater vulnerability of women tovarious kinds of disordered eating (Hoek amp van Hoeken 2003)However it has also been for practical reasons ndash psychology un-dergraduate populations that are easiest for researchers to accessare predominantly female (although this was not always the caseNisbett amp Storms 1974 was conducted at a time and place whereundergraduates were predominantly male) Given these con-straints it is dicult to conclude whether males model to the samedegree as females Of the research that has examined sex differ-ences however there is some indication that men may show anattenuated modeling effect For example Hermans et al (2010b)found that modeling might be weaker among men This is consis-tent with some evidence from research with children that modelingis weaker among boys than girls (Hendy amp Raudenbush 2000)However the majority of studies with children have found no sexdifferences (Bevelander et al 2013a Bevelander Anschuumltz amp Engels2012a Bevelander et al 2012b 2013d Salvy Coelho Kieffer ampEpstein 2007a Salvy et al 2008a 2008b) and one study found thatmen showed a stronger modeling effect (Conger et al 1980)

It has been argued that womenrsquos motivations related to eatingare complicated by the ldquothin idealrdquo This refers to a cultural valueplaced on thinness which is equated with success and attractive-ness and applies predominantly to women (Garner amp Garfinkel1980 Grogan Bell amp Conner 1997 Thompson amp Stice 2001) Con-sistent with the notion that women ndash more than men ndash are underpressure to conform to this thin ideal (Rodin Silberstein ampStriegel-Moore 1984) it has been argued that impression man-agement related to food and eating may be more important forwomen than for men (Herman amp Polivy 2010 Roth et al 2001Vartanian Herman amp Polivy 2007) Therefore we would expect thatwomen would attend more closely to normative information re-garding appropriate food intake and choice This will lead womento adjust their eating more readily to that of others leading to in-creased modeling There is plentiful evidence that eating minimallyallows women to convey an impression of femininity (see Vartanianet al 2007 for a review) whereas less is known about malesrsquo in-tentions in this regard

Taken together although there are theoretical reasons why menmight be less likely to consider their eating companionrsquos intake asa guide for their own behavior the empirical data do not providea clear picture of possible sex differences in the vulnerability tomodeling effects on intake Therefore it is not surprising that nu-merous scholars have suggested amore systematic look at differencesbetween males and females in eating behavior as an importantarea for future research (Exline Zell Bratslavsky Hamilton ampSwenson 2012 Herman amp Polivy 2010 Leone Herman amp Pliner2008)

ARTICLE IN PRESS



AgeStudies to date have been runwith children (15) and young adults

(43) with two studies looking at adolescents This diversity allowsus to be confident that modeling is unlikely to be limited to a par-ticular age group as studies have shown that modeling emerges forchildren as young as 1 year old (Harper amp Sanders 1975) There isalso evidence for developmental stability in modeling although onestudy did find that younger children showed more marked mod-eling than older children (Birch 1980) Some factors that are knownto moderate the strength of modeling differ across age groups suchas self-esteem which changes across the lifespan (RobinsTrzesniewski Tracy Gosling amp Potter 2002) however modera-tors of modeling have never been investigated in different age rangeswithin a single study In addition few studies have investigatedmod-eling in people beyond the age of young adult Although we haveno theoretical reason to expect that modeling would occur differ-ently for older adults at present there is little empirical evidencepertaining to modeling in adults above the age of 30

Body weightSeveral studies have investigated whether the body weight of

confederates or participants moderates the degree of modeling Ofthe studieswe reviewed four (Conger et al 1980 Nisbett amp Storms1974 Romero et al 2009 Rosenthal amp McSweeney 1979) foundno evidence that body weight of participants moderated the mod-eling effect ndash that is all participants exhibitedmodeling regardlessof whether they were slim healthy weight overweight or obeseAlthough the findings of Bevelander et al (2012a) were largely thesamewithbothnormal andoverweight childrenmodeling the intakeof their peers there was some indication that normal-weight chil-dren were more likely to restrict in the no-eating norm conditionwhereas overweight childrenweremore likely to overeat in the highnormcondition These differences however did not persist over timeAll in all these studies suggest that body weight does not on thewhole determine the degree of modeling ndash in contradiction to theexternality hypothesis that motivated early modeling research

There is however evidence that the interaction between par-ticipant body weight and the modelrsquos body weight can influencethe degree of modeling Five studies (De Luca amp Spigelman 1979Hermans et al 2008 Johnston 2002 McFerran et al 2010aRosenthal ampMcSweeney 1979) found evidence of a similarity effectwhere modeling was enhanced when the model was of the sameweight status as the participant That is healthy-weight partici-pants adapted their intake to that of the model but not when themodel was very thin (Hermans et al 2008) or obese (Johnston 2002McFerran et al 2010a) whereas obese participants modeled onlythe intake of an obese participant (De Luca amp Spigelman 1979) Wewill revisit these findings in the broader discussion of similarity andshared group membership below

ImpulsivityMost recently evidence has been found that individual differ-

ences in the extent to which people are able to control their eatingbehavior might also affect the extent of modeling Hermans et al(2013) showed that low-impulsive women but not high-impulsivewomen modeled the food intake of their same-sex eating com-panion High-impulsive women ate the same amount of foodregardless of howmuch the other was eating Moreover they wereless accurate in their estimations of the amount eaten by the otherperson suggesting that they paid less attention to the otherrsquos intakeIt is possible that lack of impulsivity enabled women to attend toothersrsquo intake and control their own behavior in pursuit of a de-liberate goal such as aliation However this conflicts with findingsof Salmon et al (2014) who found that women who were low inself-control were more subject to normative influence Thereforeit is dicult to draw firm conclusions on this point Research on

this topic needs to be undertaken to determine whether individ-ual differences in self-control or impulsivity increase or decreasemodeling as this issue has implications for whether modeling is aconscious or automatic process Specifically if modeling is con-scious and effortful we would expect it to be associated with highself-control or low impulsivity and reduced under cognitive loadThis issue is discussed further below

Eating goalsFour studies have demonstrated that restraint status does not

moderate modeling (Leone et al 2007 Polivy et al 1979 Rosenthalamp Marx 1979 Roth et al 2001) That is people who have a chronichistory of dieting and struggling to maintain their desired weightare as susceptible to modeling as people with no such history Sim-ilarly participants who were primed with a healthy eating goalshowed the same degree of modeling as did participants who ex-perienced no such prime (Prinsen et al 2013) On the contraryhowever Florack et al (2013) found that participants showed agreater degree of modeling when they had been primed with ahealth prevention focus Relatedly Brunner (2010) found that cuesthat reminded participants of their weight led them to inhibit theirintake and attenuated the modeling effect In sum although thereis no evidence that dietary restraint moderates modeling the currentevidence is mixed for the effect of other types of eating-related goals

Social factors

Type of social normThere is some evidence that the kind of norm communicated by

the model plays a part in determining the degree of modeling Spe-cifically three studies that have compared descriptive norms (whatothers do) and injunctive norms (what others think you should do)demonstrated that descriptive norms are more effective in induc-ingmodeling (Mollen et al 2013 Robinson Fleming amp Higgs 2013bStok de Ridder de Vet amp de Wit 2014) Relatedly Hermans andcolleagues (Hermans et al 2012a) found that social norms led tocomparable modeling effects regardless of whether they were com-municated through portion size or actual intake Howeverambiguous or mixed norms seem to have a disinhibiting effect suchthat participants no longer model (Leone et al 2007) Impor-tantly Vartanian et al (2013) have demonstrated that the perceivednorm for appropriate intake mediates the modeling effect Overallthe few studies that have examined the influence of different typesof norms on food intake support the centrality of descriptive normsand show promise for elucidating the specific normative contentparticipants attend to

Desire to aliateIt has been proposed thatmodeling reflects an attempt to develop

a social bond with onersquos eating companion (eg Exline et al 2012Hermans et al 2009a Robinson et al 2011) It follows then thatindividual differences in the desire to aliate with others ndash relatedto traits such as self-esteem empathy or sociotropy (the need toplease others and maintain social harmony) ndash could affect the mag-nitude of the modeling effect

Self-esteem plays an important role in social interactions andsocial bonding (Baumeister amp Leary 2000 Heatherton amp Wyland2003 Leary Tambor Terdal amp Downs 1995) For example peoplewith high self-esteemmay feel less need to arm their social worththan do people with low self-esteem because they worry less abouthow they are perceived by others and perceive a lower probabilityof rejection (Baumeister amp Leary 1995 Bohrnstedt amp Felson 1983Heatherton amp Vohs 2000 Heatherton amp Wyland 2003 Kenny ampDePaulo 1993) One study investigated the potential relationshipbetween modeling and the need for social acceptance by conduct-ing two experiments focusing on two individual traits (ie empathy

ARTICLE IN PRESS



and self-esteem) that could influence modeling (Robinson et al2011) Self-esteem and empathy were indeed found to be associ-ated with the degree of modeling with lower self-esteem and higherempathy scores associated with a greater degree of modeling Inaddition they found that priming feelings of social acceptance ledto an elimination of the modeling effect The findings of Bevelanderet al (2013a) further support the idea that individualsrsquo level of self-esteem can affect their degree of modeling Children with lowerexplicit body-esteem appeared to be more motivated to model thandid those with higher levels of body-esteem However implicit self-esteemwas found to have the opposite effect onmodeling such thatthose with higher implicit self-esteem were more likely to adjustto the intake of a peer than were those with lower implicit self-esteem However given the limited number of studies assessing therelationship between modeling and self-esteem (particularly im-plicit self-esteem) further work needs to be undertaken to verifythese relationships

The results of a study on sociotropy by Exline et al (2012) furthercorroborate the assumption that people might adjust their foodintake to that of others in order to aliate with them These re-searchers demonstrated that those women with a higher need toplease others and maintain social harmony ate more when they be-lieved that their eating companion wanted them to eat more andreported greater effort to model their food intake on that of theireating companion

Relatedly Hermans and colleagues (Hermans et al 2009a) ex-amined whether the quality of the social interaction affected themagnitude of themodeling effect When the confederate was friend-ly and warm modeling was attenuated Only when the confederatewas unresponsive and cold did participants show the usual mod-eling effect Again this suggests the possibility that the enhancedmodeling in the unsociable condition may reflect an attempt at in-gratiation Studies have been mixed however In a result that wouldseem to contradict the result of Hermans et al (2009a) dyads withlow scores on expressiveness have been found to model less(Brunner 2012) There have also been studies that showed no mod-erating effect of variables related to sociability such as extroversionself-monitoring (Herman et al 2005) and empathy (RobinsonBenwell amp Higgs 2013a) Therefore we might tentatively con-clude that there is some but conflicting evidence to support thenotion that modeling of food intake reflects an attempt to aliatewith the eating companion

FamiliarityThemajority of social modeling studies in adolescents and adults

involve designs in which participants are paired with strangers inunfamiliar laboratory settings This is done in order to isolate thespecific social influences of interest when eating with familiarothers common eating norms could already have been estab-lished between persons and therefore effects could reflect selectionrather than influence processes However people eat among familyand friends most of the time and so it is important that modelingresearch is conducted in these eating contexts Only one study amongadults has used an experimental design to demonstrate that mod-eling occurs in pre-existing friendship groups (Howland et al 2012)In studies with children however it has been more common forresearch to utilize familiar eating models such as peers parentsor teachers (eg Addessi et al 2005 Bevelander et al 2012a 2012b2013a 2013d Birch 1980 Harper amp Sanders 1975 Hendy ampRaudenbush 2000) Only one of these studies found any evidencethat familiarity with the model moderated modeling effects Thatis Harper and Sanders (1975) found that children were more willingto try novel foods when their mother (as opposed to a stranger)offered the food All in all on the basis of these findings we canconclude that people model the eating of familiar as well as un-familiar eating companions

Similarity and shared group membershipOne finding that has emerged acrossmultiple studies is thatmod-

eling appears to be enhanced when individuals are similar eitherin terms of sex (Conger et al 1980) weight (Hermans et al 2008Johnston 2002 McFerran et al 2010a Rosenthal amp McSweeney1979) or age (Hendy amp Raudenbush 2000) This is exactly whatwould be predicted from modern social-psychological theories ofsocial influence which state that other people are seen as provid-ing a relevant reference point (eg for appropriate eating behavior)only when they are categorized as similar to the self on dimen-sions that are contextually relevant This notion was confirmed ina study by Cruwys et al (2012) which found that when partici-pants self-categorized in terms of their university student identitythey modeled confederates who identified themselves as stu-dents of the same university but did not model confederates whoidentified themselves as students of another university SimilarlyStok de Ridder de Vet and de Wit (2012) found that participantsmodeled the eating behavior of majority group members and di-verged from the behavior of minority group members particularlywhen the participants were highly identified with the referencegroup Therefore we may conclude that perceived similarity is animportant moderator of modeling effects Critically in both thesestudies the moderating identities were made salient to partici-pants in the moment ndash it is only when participants see themselvesin terms of their university student identity that we would expectthem to model only those from the same university

Attending to shared group membership can also explain why insome circumstances participants might react against an eating normprovided by others Berger and colleagues (Berger amp Heath 2008Berger amp Rand 2008) found that individuals were more likely to eathealthily when an undesirable out-group provided a norm for un-healthy eating This complements the findings of Oyserman Frybergand Yoder (2007) who found that individuals were less likely to eathealthily when they were reminded that out-group members hada healthy eating norm That is because people do not seek to af-filiate with and may wish to distance themselves from out-groupmembers we do not findmodeling andmay sometimes even expectreactance in such circumstances Therefore an important consid-eration in interpreting modeling effects is the similarity betweenthe model and participants and perhaps more importantly the per-ceived shared group membership

Is modeling conscious or automatic

Several studies have shown that people report that they are notpersonally susceptible to modeling (Croker Whitaker Cooke ampWardle 2010 Vartanian Herman amp Wansink 2008) This is con-sistent with a broader research finding that although peoplegenerally acknowledge that external elements influence others theyreport that these elements do not influence their own behavior (thethird-person effect Davison 1983) What is not clear at this stagehowever is whether this represents a lack of awareness of model-ing (that is it occurs unconsciously) or whether this lack of reportingis due to motivated denial (that is people deny that they model forunknown reasons Spanos Vartanian Herman amp Polivy 2013)

Evidence supporting the idea that modeling might be automat-ic and outside of awareness comes from studies of mimicry It hasbeen suggested that people process the behavior of others andengage in imitation unconsciously (Bargh amp Chartrand 1999 NolanSchultz Cialdini Goldstein amp Griskevicius 2008) There is also ev-idence that mimicry of gestures occurs unconsciously andmoreoverfunctions as a way of aliating with others (Iacoboni 2009) Severalstudies have provided evidence that people are more likely to reachfor food (Bevelander Lichtwarck-Aschoff Anschuumltz Hermans ampEngels 2013c) or take a bite or sip immediately after witnessingsomeone else do so (Hermans et al 2012b Koordeman Kuntsche

ARTICLE IN PRESS



Anschutz van Baaren amp Engels 2011 Larsen Engels Granic ampOverbeek 2009) However mimicry of eating is also responsive toaliation goals in a similar way to traditional social modeling studiesFor example these studies have also shown that people are morelikely to imitate at the start of a social interaction than at the endand that humans automatically and unconsciously try to preventimitation when they do not want a bond with another person (vanBaaren Holland Kawakami amp van Knippenberg 2004) Thereforeif mimicry is a predominantly automatic process modeling mustalso be at least partially automatic at least to the degree that it ismediated by direct behavioral mimicry (although this probably onlyapplies to food intake rather than food choice)

Further evidence that modeling is at least partially automaticcomes from studies looking at cognitive load Cognitive load theorystates that conscious effortful tasks require higher-level cognitiveresources such as attention and self-control It follows that if a personis pre-occupied with a task that uses these (limited) cognitive re-sources they will be unable to perform other conscious effortfultasks (Bargh 1984 Sweller Ayres amp Kalyuga 2011) A study byBevelander et al (2013d) demonstrated that among children watch-ing television led to increased modeling of a peer but only whenthe content of the program was emotionally laden This is consis-tent with the findings of other studies (Bellisle Dalix amp Slama 2004Temple Giacomelli Kent Roemmich amp Epstein 2007) Given thatprevious research has argued that processing emotions requires cog-nitive attention causing people to act automatically or mindlesslyin other ways (Baumeister Vohs DeWall amp Zhang 2007 Wansinkamp Sobal 2007) this study provides evidence that modeling can occurwithout conscious effort

On the other hand evidence that modeling is enhanced whenparticipants are less impulsive (Hermans et al 2013) or better atself-monitoring (Berger amp Rand 2008) suggests that people arecapable of monitoring and exerting control over their modeling be-havior Recent evidence has also shown that people can accuratelyreport that social influence determines other peoplersquos eating andthat some people are strategically motivated to deny social influ-ences over their eating (Spanos et al 2013) In addition this studyfound that participants could accurately report instances of mimicryin observed dyads but not of modeling of intake across the courseof meals

All in all the evidence to date would suggest that althoughmod-eling can be automatic it is also accessible to conscious control Itis unlikely that the majority of modeling behavior is strategic or in-tended but individuals are obviously capable of attending to andmodifying their own eating behavior and therefore there are cir-cumstances in which peoplemight intentionally increase or decreaseintake in response to their eating companion Furthermore al-though modeling might be mediated by automatic processes suchasmimicry this cannot account formodeling effects shown in studiesusing a design in which written information is provided about howmuch previous participants have consumed To gain more insightinto the possible (non-)automaticity of modeling behavior more re-search is needed to (1) conclusively determine how automaticmodeling is and (2) assess the degree to which mimicry underliesmodeling

Theoretical implications

Boundary conditions versus mechanism

A strong effort has been made by previous research to identifya substantial number of moderators that qualify the modeling effectThese studies have been empirically sound and have successfullyidentified a large number of candidate variables For example therehas been a thorough investigation into the way in which weight ofboth the model and the participant play a role in determining the

degree of modeling However what is sometimes lacking is an in-tegrated and parsimonious model that is able to explain why eachof thesemoderatorsmight exist A theoretical formulation that speci-fies the scope of themodeling effect (and therefore what moderatorswe should expect) would also assist in the interpretation of seem-ingly contradictory findings in studies investigating moderators ofmodeling For example how are we to interpret the findings thatmodeling is enhanced among people who are not impulsive(Hermans et al 2013) but also not self-controlled (Salmon et al2014) A strong focus on moderators in the absence of a unifyingtheory is problematic because while researchers focus on ques-tions of when modeling will not occur they are necessarily lessconcerned with explaining why modeling is so robust and how pe-oplersquos eating is socially influenced by others Of course investigatingmoderators can in some circumstances be a means of investigat-ing process That is if a moderator is theorized to be a necessarycondition for modeling to occur and an experimental paradigm canlsquodisablersquo the moderator (or prevent it from functioning) then it ispossible to provide an experimental test of mechanism that is em-pirically superior to mediation (Jacoby amp Sassenberg 2011) Thereare examples of this approach being successfully applied in themod-eling literature For instance if experimentally-induced high self-control reduces the modeling effect (Salmon et al 2014) it followsthat modeling is not a psychologically effortful behavior Similarlyif modeling persists when both the experimenter and fellow par-ticipants are believed to be unaware of the participantrsquos intake(Yamasaki et al 2007) it follows that modeling is not purely dueto aliation motives Unfortunately however the hunt for mod-erators has been unsystematic and the lengthy list of potentialmoderators identified by this review may leave researchers feelinglike the modeling effect is not so robust after all Therefore a keyagenda for future research must be the question of which mecha-nism(s) underlie(s) modeling effects on eating Although broadlyunderstood to be the result of normative influence (Herman et al2003b) there is a clear need for research specifically testing the pos-sible mechanisms underlying modeling effects on food intake andstudies that contrast one mechanism against another

Implications for normative theory

The dominant theoretical framework that has aimed to explainmodeling effects is the normative theory of Herman and colleagues(2003b 2005) The authors aimed to reconcile literature on socialfacilitation impression management and social modeling pro-cesses based on the existing studies in (mostly female) young adultsbefore 2003 (Herman et al 2003b) The normative framework hasbeen widely used in research on social norms in eating with thesetwo articles receiving over 370 scientific citations to date More-over it has clearly been generative ndash 47 of the 69 experimentalstudies that we review were published after 2003 These studiesprovide a number of clues for how we might further enhance thetheoretical framework and in this section we discuss these possi-bilities and suggest potential future directions for research

First the reviewed studies clearly support the assumption thatpeople look to others to determine how much they can eat andtherefore individuals will both augment and inhibit their eating inaccordance with social norms However some studies have foundthat intake in ldquobaselinerdquo conditions (where participants eat alone)more closely resembles food intake in the ldquohigh normrdquo conditionthan the ldquolow normrdquo condition (eg Feeney et al 2011 Hermanset al 2009a 2009b Pliner amp Mann 2004 Robinson amp Higgs 2013Roth et al 2001 Vartanian et al 2013) Some researchers (egMcFerran et al 2010a Vartanian et al 2013) have interpreted thisas evidence that modeling primarily inhibits rather than aug-ments intake Their reasoning is that people have an inherenttendency to eat as much as they can within the bounds of what is

ARTICLE IN PRESS



normatively considered appropriate (ie avoid eating ldquoexcessive-lyrdquo) However it is possible that this reasoning overlooks a confoundin many experimental designs that might augment eating in the so-called ldquobaselinerdquo conditions That is experimental designs cancommunicate norms and provide a point of comparison for par-ticipants even if not intentionally If modeling is such a strong drivethen participants will look for information about appropriate con-sumption especially in control conditions where this informationis not provided explicitly by a confederate Therefore experimen-tal designs that provide large portion sizes and instruct participantsthat it is exclusively for their consumption (and in some cases ex-plicitly encourage high consumption eg ldquoHelp yourself we willhave to throw the rest away anywayrdquo De Luca amp Spigelman 1979Robinson et al 2013a 2014b Vartanian et al 2013) provide nor-mative information that encourages high intake over and above anyexperimental manipulation Given this potential confound it is notpossible to conclude on the basis of the current literature whetherinhibition or augmentation effects are more common in the mod-eling paradigm One way to reduce this experimental demand infuture studies might be to design studies where participants canchoose their own portion sizes or use foods where participants donot infer that all the offered food is for them personally (eg cakewhich is typically shared or wrapped sweets)

Second there is an ongoing debate about the motives for pe-oplersquos adherence to social norms At the time that the normativemodel was first outlined in 2003 both aliation (the need to beliked) and uncertainty-reduction (the need to be right Deutsch ampGerard 1955) were put forward as potential reasons for model-ing Later however Herman and Polivy (2005) questioned the roleof aliation as a motive for modeling arguing that this is incon-sistent with the persistence of modeling in the absence of othersWhat at this time can we say about the role of aliation versusuncertainty-reduction asmotives for modeling in light of the studiesthat have been published in the intervening period

A growing body of evidence has borne out the finding that mod-eling persists in situations where it is implausible that individualsare strategically seeking to ingratiate themselves with others Forexample the finding by Roth et al (2001) that modeling effectspersist even when participants are alone and believe their eatingis unobserved has been corroborated by several other research teams(Burger et al 2010 Yamasaki et al 2007) In a related finding re-searchers have found that social norms for a wide range of behaviorsfunction when others are not physically present (Cialdini Reno ampKallgren 1990 Kallgren Reno amp Cialdini 2000 Larimer TurnerMallett amp Geisner 2004 Stok et al 2014) These studies are con-sistent with the view that modeling is underpinned by anuncertainty-reduction motive In other words we can conclude thatindividuals look to others to provide meaningful information aboutwhat is appropriate to eat how much when and how

On the other hand recent evidence has also corroborated theidea that aliation goals do play a role in shaping modeling be-havior In particular modeling is enhanced for those with highempathy or low self-esteem (Bevelander et al 2013a Robinson et al2011) or when people seek a stronger social bond (Exline et al 2012Hermans et al 2009a) These findings suggest that aliationmotives(peoplersquos need to be liked accepted and to belong) cannot be ruledout as a motive for modeling That is even if uncertainty-reductionis the primary motive underlying modeling it still remains possi-ble that aliation goals are a secondary motive under certaincircumstances (or even a primary motive in some contexts)

In a nutshell it is not always clearwhichmotive prevails and underwhat circumstances One diculty inherent in this question is thatthese two motives could be interrelated which makes it dicult toexamine their independent influence on modeling effects on foodintake A hypothesis that one can derive from the social identity ap-proach (Tajfel amp Turner 1979 Turner Hogg Oakes Reicher amp

Wetherell 1987) is thus people model to reduce uncertainty butmodeling will be associated with aliation (and related variables)because aliation (either perceived or sought) is a precondition formodeling to occur This is because modeling can only reduce uncer-tainty to the extent that shared group membership already exists ndashout-groupmembers do not offer a valid guide to appropriate or correctbehavior To take this a step further it follows that when we seek toaliate with others ndash whether because of empathy sociotropy lowself-esteem or contextual factors ndash we also believe that those othersprovide a valid reference point for our own behavior (Turner 1999Turner amp Oakes 1989) That is when we perceive a shared psycho-logical group membership the eating norm provided by in-groupmembers becomes self-relevant and these in-group norms tell uswhatthoughts feelings and behaviors are appropriate in an often unfa-miliar context (Berger amp Heath 2008 Cruwys et al 2012 McFerranDahl Fitzsimons amp Morales 2010b Stok et al 2012)

This theoretical framing is consistent with the normative modelbut allows us to understand the different motives that have beenidentified for modeling not as contradictory but rather as reflect-ing different aspects of the same social influence process On thebasis of this theorizing we can say that models and the norms thatthey communicate will be considered valid reference points onlyto the degree that shared group membership already exists (at leastsubjectively Turner 1991) Furthermore this implies that theldquodefaultrdquo for participants is perceived shared group membershipat least in studies where participants typically share sex age weightstatus educational background ethnicity university student iden-tity etc with confederates any one of which might form a basisfor psychological aliation in the moment

Third more research is needed on modeling of food choice toexamine whether the normative account is applicable and whethercontextual uncertainty might be a critical moderator here Al-though there have been theoretical reasons proposed for why foodchoice might be less affected by social influence than food intakehighly-powered experimental studies are needed to address this em-pirically To date the majority of studies have focused primarily onmodeling of snack foods or modeling to encourage (novel) low-energy-dense food consumption among young people (Hendy 2002Hendy amp Raudenbush 2000 Reverdy et al 2008) These are one-sided approaches because much of a personrsquos eating is determinedby choices made in the grocery store or from restaurant menusrather than simply free-eating from a single type of food To be trulyconfident that modeling effects have a powerful influence on eating-related decisions in peoplesrsquo day-to-day lives and whether themechanisms underlying modeling of food choice and intake are thesame it would be useful to expand this research area

Practical implications

Social influence is a primary determinant of eating

An important finding of this review is that individual factors donot appear to be critical in explaining modeling effects Severalstudies investigating factors such as weight and personality havefound that even when significant moderators have been identi-fied they had a small effect relative to the robustness of modelingThis makes the consistent and substantial effect of social influ-ence on eating behavior all the more marked and important toconsider in public health policy Although questions of mecha-nism and boundary conditions of social influence effects on eatingare of academic interest the simple fact that social influence is aprimary predictor of eating behavior has perhaps not been givenenough emphasis and more must be done to translate this re-search that is make it relevant and accessible to health practitionersand policymakers This is crucial in an environment where the ma-jority of research is concerned with the genetic metabolic and

ARTICLE IN PRESS



personality-based predictors of eating behavior (particularly forpathological eating behavior Hill amp Melanson 1999)

The public health problems of obesity and unhealthy dieting aswell as the clinical problems of eating disorders are partially de-termined by the same basic social influence process that underpinsmodeling of food intake and choice For instance research has dem-onstrated that subclinical indicators of disordered eating (such asdieting bingeing and purging) are also subject to social influenceparticularly from peers and family (Crandall 1988 Hutchinson ampRapee 2007 Paxton Eisenberg amp Neumark-Sztainer 2006 PaxtonSchutz Wertheim amp Muir 1999 Salvy de la Haye Bowker ampHermans 2012) Modeling also extends to food purchasing deci-sions (Bevelander et al 2011) and is therefore likely to affect long-term consumption patterns There are also numerous studiessuggesting that peoplersquos weight can be predicted (at least partial-ly) from that of their friends and that obesity clusters in socialnetworks (Badaly 2013 Christakis amp Fowler 2007) These appliedstudies of social influence and eating suggest that modeling has veryreal consequences for physical and mental health at a populationlevel

Furthermore what has been overlooked by those whowould aimto ldquoinoculaterdquo people against the evils of social influence (eg Badaly2013 Vartanian 2009) is that this powerful determinant of eatingbehavior might be harnessed ldquofor goodrdquo (Rosenberg 2011) forexample to encourage healthy eating It has been demonstrated thatchildren model the healthy eating habits of their peers andorparents leading to an increased vegetable intake and reduced fatintake (Bevelander et al 2012b Tibbs et al 2001) and that stu-dents who reside in colleges with healthy eating norms are morelikely to eat healthily and exercise (Gruber 2008) In part becauseof the focus on harmful social influence efforts to design and im-plement interventions that utilize positive social influence are intheir infancy

At this stage it is also important to realize that when compar-ing themodeling literature with eating behavior in real-life contextsthe food choices and amounts consumed by people in the directsocial environment are not likely to be as uniform as a confede-ratersquos behavior in an experimental setting For example confederatesmostly chose either healthy or unhealthy foods or are instructedto eat a small or large portion A study on young adults in whichseveral confederates ate different amounts of palatable food sug-gested that when norms are ambiguous people are less likely tomodel the amount of food consumed by others (Leone et al 2007)This has clear relevance beyond the laboratory where eatingnorms are rarely overt or uncontested Therefore it may be thatnot all findings can directly be translated to real-world eating be-havior andmore research in applied settings is critical to establishingthe relevance of laboratory studies of eating Nevertheless basedon our current knowledge on modeling of food choice andintake we propose several steps to inform health-promotioninterventions

Designing effective healthy-eating interventions using modeling

One approach to modify the social environment with regard tofood and eating might be the community reinforcement ap-proach According to this approach different reinforcers are usedto assist individuals in the adoption and maintenance of a healthierlifestyle within the context of a supportive social network (MeyersVillanueva amp Smith 2005) As noted above the social network hasthe potential to positively influence onersquos energy balance and dietcomposition in numerous ways ndash particularly among childrenParents for instance may influence the family environment by ex-posing the family members to certain foods and actively or passivelyallow them to eat certain foods (Clark Goyder Bissel Blank amp Peters2007 Golan amp Crow 2004) By doing so parents set social norms

regarding food and eating and these norms are likely to influenceinitiation and maintenance of childrenrsquos regular eating habits Thusit is conceivable that by modifying the behavior of a ldquomodelrdquo (egparent sibling peer) there are flow-on benefits to others in the socialnetwork

A similar approach could be applied to nutritional interven-tions (again the existing evidence base is strongest among children)Most notably the lsquoFood Dudesrsquo a program featuring heroic peersthat model a preference for fruit and vegetables has been trialedwith thousands of schoolchildren and has been shown to influ-ence actual consumption patterns in the short and medium term(Horne et al 2004 2009 Lowe Horne Tapper Bowdery amp Egerton2004) Although current studies do not provide robust evidence thatpeers can reduce preferences for high-energy-dense foods studiesthat investigated peer rejection of foods show that it is possible forchildren (Greenhalgh et al 2009 Horne et al 2004) as well as youngfemale adults (Robinson amp Higgs 2012) to take over a peerrsquos foodaversion In these studies the aversion against certain foods wasprovided by an outspoken peer which might have had a strongerinfluence than merely modeling consumption For example chil-dren were unwilling to eat novel foods after negative comments bytheir peers (Greenhalgh et al 2009) Instead of focusing on en-couragement of low energy-dense foods alone it may be useful toexpand the research area and investigate whether the impact of apeer could also be used to reject high-energy-dense foods ndash at leastnovel ones

Modeling research is also powerful in its capacity to explain theeffectiveness of public health interventions in the domain of eatingFor example we know that interventions such as increasing the avail-ability of fruit and vegetables are effective in improving nutritionalstatus (Hearn et al 1998) However rather than attributing this toautomatic behavior (or stealthy ldquonudgerdquo tactics Hanks Just SmithampWansink 2012 Thaler amp Sunstein 2008) modeling research sug-gests that individuals infer important information about group normsfrom the availability of particular foods (as well as for exampleportion size Hermans et al 2012a) that is then used to inform in-dividual food choices Therefore modeling provides a powerful andexperimentally tested framework for making causal inferences aboutthe relationship between societal norms and population eatingbehaviors

Conclusion

Although social modeling is a complex process (particularly inpredicting the degree to which people will model in particular cir-cumstances) the most important conclusion of this review is thatpeoplersquos food intake is determined in large part by social influ-ence and by modeling in particular We found that across 69modeling studies there were three key conclusions that we can drawfrom this review First there was near universal support for thefinding that peoplersquos food intake and choices are shaped by thenorms provided by others Furthermore we found that many at-tempts to identify moderators of the modeling effect have beenunsuccessful andwhen significantmoderators have been found theytypically account for only a small amount of variance in modeling

Second there is evidence that modeling occurs both because in-dividuals seek information about appropriate behavior (anuncertainty-reduction motive) and because individuals seek to af-filiate with others (an aliation motive) Rather than treating theseas incompatible this evidence is best understood as supporting asocial identity model of social influence whereby individuals lookto similar others (or those with whom they are aliated) to providevalid information about appropriate eating Social influencemay thusbe seen as a fundamental feature of human perception and behav-ior which might explain healthy as well as unhealthy eatingbehaviors

ARTICLE IN PRESS



Third the domain in which modeling has been demonstratedis relatively narrow ndash most studies focus on the snack food intakeof young adult females in a laboratory setting We conclude that itis now time to move out of the lab and into the realm of interven-tion ndash both at the clinical level and at the public health level Oneof the great strengths of the literature on modeling has been its ex-perimental focus and strong empirical controls However morestudies are needed to test the robustness of the modeling effectoutside of the laboratory and even more importantly to deter-mine how knowledge of this effect might enhance our capacity tosupport healthy eating and population health Given the current so-cietal challenges of both obesity and disordered eating it is timelyfor us to demonstrate the utility of modeling research for inter-vention and health promotion

References

Addessi E Galloway A T Visalberghi E amp Birch L L (2005) Specific socialinfluences on the acceptance of novel foods in 2ndash5-year-old children Appetite45(3) 264ndash271 doi101016jappet200507007

Badaly D (2013) Peer similarity and influence for weight-related outcomes inadolescence A meta-analytic review Clinical Psychology Review 33 1218ndash1236

Bargh J A (1984) Automatic and conscious processing of social information In RS Wyer amp T K Srull (Eds) Handbook of social cognition (Vol 3 pp 129ndash178)Hillsdale NJ Erlbaum

Bargh J A amp Chartrand T L (1999) The unbearable automaticity of being AmericanPsychologist 54 462ndash479 doi1010370003-066X547462

Baumeister R F amp Leary M R (1995) The need to belong Desire for interpersonalattachments as a fundamental human motivation Psychological Bulletin 117(3)497ndash529 doi1010370033-29091173497

Baumeister R F amp Leary M R (2000) The nature and function of self-esteemSociometer theory Advances in Experimental Social Psychology 32 1ndash62doi101016S0065-2601(00)80003-9

Baumeister R F Vohs K D DeWall C N amp Zhang L (2007) How emotion shapesbehavior Feedback anticipation and reflection rather than direct causationPersonality and Social Psychology Review 11(2) 167ndash203 lthttppsrsagepubcomcontent112167abstractgt

Bellisle F Dalix A M amp Slama G (2004) Non food-related environmental stimuliinduce increased meal intake in healthy women Comparison of televisionviewing versus listening to a recorded story in laboratory settings Appetite 43(2)175ndash180 doi101016jappet200404004

Berger J amp Heath C (2008) Who drives divergence Identity signaling outgroupdissimilarity and the abandonment of cultural tastes Journal of Personality andSocial Psychology 95(3) 593ndash607 doi1010370022-3514953593

Berger J amp Rand L (2008) Shifting signals to help health Using identity signalingto reduce risky health behaviors Journal of Consumer Research 35(3) 509ndash518doi101086587632

Bevelander K E Anschuumltz D J Creemers D H M Kleinjan M amp Engels R C ME (2013a) The role of explicit and implicit self-esteem in peer modeling ofpalatable food intake A study on social media interaction among youngstersPLoS ONE 8(8) e72481 doi101371journalpone0072481

Bevelander K E Anschuumltz D J amp Engels R C M E (2011) Social modeling of foodpurchases at supermarkets in teenage girls Appetite 57(1) 99ndash104 doi101016jappet201104005

Bevelander K E Anschuumltz D J amp Engels R C M E (2012a) Social norms in foodintake among normal weight and overweight children Appetite 58(3) 864ndash872doi101016jappet201202003

Bevelander K E Anschuumltz D J amp Engels R C M E (2012b) The effect of a fictitiouspeer on young childrenrsquos choice of familiar versus unfamiliar low and highenergy-dense foods British Journal of Nutrition 108(6) 1126ndash1133

Bevelander K E Engels R C M E Anschuumltz D J amp Wansink B (2013b) The effectof an intervention on schoolchildrenrsquos susceptibility to a peerrsquos candy intakeEuropean Journal of Clinical Nutrition 67(8) 829ndash835 doi101038ejcn2013122

Bevelander K E Lichtwarck-Aschoff A Anschuumltz D J Hermans R C J amp EngelsR C M E (2013c) Imitation of snack food intake among normal-weight andoverweight children Frontiers in Psychology 4 949 doi103389fpsyg201300949

Bevelander K E Meiselman H L Anschuumltz D J amp Engels R C M E (2013d)Television watching and the emotional impact on social modeling of food intakeamong children Appetite 63 70ndash76 doi101016jappet201212015

Birch L L (1980) Effects of peer modelsrsquo food choices and eating behaviors onpreschoolersrsquo food preferences Child Development 51(2) 489ndash496

Bohrnstedt G W amp Felson R B (1983) Explaining the relations among childrenrsquosactual and perceived performances and self-esteem A comparison of severalcausal models Journal of Personality and Social Psychology 45(1) 43ndash56

Brunner T A (2010) How weight-related cues affect food intake in a modelingsituation Appetite 55(3) 507ndash511 doi101016jappet201008018

Brunner T A (2012) Matching effects on eating Individual differences do make adifference Appetite 58 429ndash431

Burger J M Bell H Harvey K Johnson J Stewart C Dorian K et al (2010)Nutritious or delicious The effect of descriptive norm information on food choiceJournal of Social and Clinical Psychology 29(2) 228ndash242

Christakis N A amp Fowler J H (2007) The spread of obesity in a large social networkover 32 years The New England Journal of Medicine 357 370ndash379

Cialdini R B Reno R R amp Kallgren C A (1990) A focus theory of normative conductRecycling the concept of norms to reduce littering in public places Journal ofPersonality and Social Psychology 58(6) 1015ndash1026

Clark H R Goyder E Bissell P Blank L amp Peters J (2007) How do parentsrsquochild-feeding behaviours influence child weight Implications for childhoodobesity policy Journal of Public Health 29(2) 132ndash141

Clendenen V I Herman C P amp Polivy J (1994) Social facilitation of eating amongfriends and strangers Appetite 23 1ndash13

Conger J C Conger A J Costanzo P R Wright K L amp Matter J A (1980) Theeffect of social cues on the eating behavior of obese and normal subjects Journalof Personality 48(2) 258ndash271

Crandall C S (1988) Social contagion of binge eating Journal of Personality and SocialPsychology 55(4) 588ndash598

Croker H Whitaker K L Cooke L amp Wardle J (2010) Do social norms affectintended food choice Preventive Medicine 49(2ndash3) 190ndash193 doi101016jypmed200907006

Cruwys T Platow M J Angullia S A Chang J M Diler S E Kirchner J L et al(2012) Modeling of food intake is moderated by salient psychological groupmembership Appetite 58(2) 754ndash757 doi101016jappet201112002

Davison W P (1983) The third-person effect in communication Public OpinionQuarterly 47(1) 1ndash15

de Castro J M amp Brewer E M (1992) The amount eaten in meals by humans is apower function of the number of people present Physiology amp Behavior 51(1)121ndash125 doi1010160031-9384(92)90212-K

De Luca R V amp Spigelman M N (1979) Effects of models on food intake of obeseand non-obese female college students Canadian Journal of Behavioural ScienceRevue Canadienne Des Sciences Du Comportement 11(2) 124ndash129

Deutsch M amp Gerard H B (1955) A study of normative and informational socialinfluences upon individual judgment The Journal of Abnormal and SocialPsychology 51(3) 629ndash636

Exline J J Zell A L Bratslavsky E Hamilton M amp Swenson A (2012) People-pleasing through eating Sociotropy predicts greater eating in response toperceived social pressure Journal of Social and Clinical Psychology 31(2) 169ndash193doi101521jscp2012312169

Feeney J R Polivy J Pliner P amp Sullivan M D (2011) Comparing live and remotemodels in eating conformity research Eating Behaviors 12(1) 75ndash77 doi101016jeatbeh201009007

Ferraro K F amp Wilmoth J M (2000) Measuring morbidity Disease countsbinary variables and statistical power The Journals of Gerontology Series BPsychological Sciences and Social Sciences 55(3) S173ndashS189 doi101093geronb553S173

Florack A Palcu J amp Friese M (2013) The moderating role of regulatory focus onthe social modeling of food intake Appetite 69 114ndash122 doi101016jappet201305012

Garner D M amp Garfinkel P E (1980) Socio-cultural factors in the development ofanorexia nervosa Psychological Medicine 10 647ndash656

Goldman S J Herman C P amp Polivy J (1991) Is the effect of a social model oneating attenuated by hunger Appetite 17 129ndash140

Greenhalgh J Dowey A J Horne P J Fergus Lowe C Griths J H amp WhitakerC J (2009) Positive- and negative peer modelling effects on young childrenrsquosconsumption of novel blue foods Appetite 52(3) 646ndash653 doi101016jappet200902016

Golan M amp Crow S (2004) Targeting parents exclusively in the treatment of obesityLong-term results Obesity Research 12(2) 357ndash361

Grogan S C Bell R amp Conner M (1997) Eating sweet snacks Gender differencesin attitudes and behaviour Appetite 28 19ndash31

Gruber K J (2008) Social support for exercise and dietary habits among collegestudents Adolescence 43(171) 557ndash575

Hanks A S Just D R Smith L E amp Wansink B (2012) Healthy convenienceNudging students toward healthier choices in the lunchroom Journal of PublicHealth 34(3) 370ndash376 doi101093pubmedfds003

Harper L V amp Sanders K M (1975) The effect of adultsrsquo eating on young childrenrsquosacceptance of unfamiliar foods Journal of Experimental Child Psychology 20206ndash214

Hearn M D Baranowski T Baranowski J Doyle C Smith M Lin L S et al (1998)Environmental influences on dietary behavior among children Availability andaccessibility of fruits and vegetables enable consumption Journal of HealthEducation 29(1) 26ndash32 doi10108010556699199810603294

Heatherton T F Polivy J amp Herman C P (1991) Restraint weight loss and variabilityof body weight Journal of Abnormal Psychology 100(1) 78ndash83

Heatherton T F amp Vohs K D (2000) Interpersonal evaluations following threatsto self Role of self-esteem Journal of Personality and Social Psychology 78(4)725ndash736

Heatherton T F amp Wyland C L (2003) Assessing self-esteem In S J Lopez amp C RSnyder (Eds) Positive psychological assessment (pp 219ndash233) Washington DCAmerican Psychological Association

Hendy H M (2002) Effectiveness of trained peer models to encourage foodacceptance in preschool children Appetite 39(3) 217ndash225 doi101006appe20020510

Hendy H M amp Raudenbush B (2000) Effectiveness of teachermodeling to encouragefood acceptance in preschool children Appetite 34(1) 61ndash76 doi101006appet19990286

Herman C P Koenig-Nobert S Peterson J B amp Polivy J (2005) Matching effectson eating Do individual differences make a difference Appetite 45 108ndash109

ARTICLE IN PRESS



Herman C P amp Polivy J (1988) Psychological factors in the control of appetiteCurrent Concepts in Nutrition 16 41ndash51

Herman C P amp Polivy J (2005) Normative influences on food intake Physiologyamp Behavior 86 762ndash772

Herman C P amp Polivy J (2008) External cues in the control of food intake in humansThe sensory-normative distinction Physiology amp Behavior 94 722ndash728

Herman C P amp Polivy J (2010) Sex and gender differences in eating In J C Chrisleramp D R McCreary (Eds) Handbook of gender research in psychology (pp 455ndash469)New York Springer New York doi101007978-1-4419-1465-1

Herman C P Polivy J Kauffman N amp Roth D A (2003a) Is the effect of a socialmodel on eating attenuated by satiety University of Toronto

Herman C P Roth D A amp Polivy J (2003b) Effects of the presence of otherson food intake A normative interpretation Psychological Bulletin 129(6)873ndash886

Hermans R C J Engels R C M E Larsen J K amp Herman C P (2009a) Modelingof palatable food intake The influence of quality of social interaction Appetite52(3) 801ndash804 doi101016jappet200903008

Hermans R C J Herman C P Larsen J K amp Engels R C M E (2010a) Socialmodeling effects on young womenrsquos breakfast intake Journal of the AmericanDietetic Association 110(12) 1901ndash1905 doi101016jjada201009007

Hermans R C J Herman C P Larsen J K amp Engels R C M E (2010b) Socialmodeling effects on snack intake among young men The role of hunger Appetite54(2) 378ndash383 doi101016jappet201001006

Hermans R C J Larsen J K Herman C P amp Engels R C M E (2008) Modelingof palatable food intake in female young adults Effects of perceived body sizeAppetite 51(3) 512ndash518 doi101016jappet200803016

Hermans R C J Larsen J K Herman C P amp Engels R C M E (2009b) Effects ofsocial modeling on young womenrsquos nutrient-dense food intake Appetite 53(1)135ndash138 doi101016jappet200905004

Hermans R C J Larsen J K Herman C P amp Engels R C M E (2012a) Howmuchshould I eat Situational norms affect young womenrsquos food intake during mealtime The British Journal of Nutrition 107(4) 588ndash594 doi101017S0007114511003278

Hermans R C J Larsen J K Lochbuehler K Nederkoorn C Herman C P amp EngelsR C M E (2013) The power of social influence over food intake Examining theeffects of attentional bias and impulsivity The British Journal of Nutrition 109(3)572ndash580 doi101017S0007114512001390

Hermans R C J Lichtwarck-Aschoff A Bevelander K E Herman C P Larsen JK amp Engels R C M E (2012b) Mimicry of food intake The dynamic interplaybetween eating companions PLoS ONE 7(2) e31027 doi101371journalpone0031027

Hermans R C J Salvy S-J Larsen J K amp Engels R C M E (2012c) Examiningthe effects of remote-video confederates on young womenrsquos food intake EatingBehaviors 13 246ndash251 doi101016jeatbeh201203008

Hill J O amp Melanson E L (1999) Overview of the determinants of overweight andobesity Current evidence and research issues Medicine amp Science in Sports ampExercise 31(11) S515 lthttpjournalslwwcomacsm-msseFulltext199911001Overview_of_the_determinants_of_overweight_and5aspxgt

Hoek H W amp van Hoeken D (2003) Review of the prevalence and incidence ofeating disorders International Journal of Eating Disorders 34(4) 383ndash396

Horne P J Hardman C A Lowe C F Tapper K Le Noury J Madden P et al (2009)Increasing parental provision and childrenrsquos consumption of lunchbox fruit andvegetables in Ireland The Food Dudes intervention European Journal of ClinicalNutrition 63(5) 613ndash618 doi101038ejcn200834

Horne P J Tapper K Lowe C F Hardman C A Jackson M C ampWoolner J (2004)Increasing childrenrsquos fruit and vegetable consumption A peer-modelling andrewards-based intervention European Journal of Clinical Nutrition 58(12)1649ndash1660 doi101038sjejcn1602024

Howland M Hunger J M amp Mann T (2012) Friends donrsquot let friends eat cookiesEffects of restrictive eating norms on consumption among friends Appetite 59(2)505ndash509 doi101016jappet201206020

Hutchinson D M amp Rapee R M (2007) Do friends share similar body image andeating problems The role of social networks and peer influences in earlyadolescence Behaviour Research and Therapy 45(7) 1557ndash1577 doi101016jbrat200611007

Iacoboni M (2009) Imitation empathy and mirror neurons Annual Review ofPsychology 60 653-670

Jacoby J amp Sassenberg K (2011) Interactions do not only tell us when but can alsotell us how Testing process hypotheses by interaction European Journal of SocialPsychology 41(2) 180ndash190 doi101002ejsp762

Johnston L (2002) Behavioral mimicry and stigmatization Social Cognition 20(1)18ndash35 doi101521soco2011820944

Kallgren C A Reno R R amp Cialdini R B (2000) A focus theory of normative conductWhen norms do and do not affect behavior Personality and Social PsychologyBulletin 26(8) 1002ndash1012 Retrieved from lthttppspsagepubcomcontent2681002abstractgt

Kenny D A amp DePaulo B M (1993) Do people know how others view them Anempirical and theoretical account Psychological Bulletin 114(1) 145ndash161

Koordeman R Kuntsche E Anschutz D J van Baaren R B amp Engels R C M E(2011) Do we act upon what we see Direct effects of alcohol cues inmovies on young adultsrsquo alcohol drinking Alcohol and Alcoholism 46(4) 393ndash398lthttpalcalcoxfordjournalsorgcontent464393abstractgt

Larimer M E Turner A P Mallett K A amp Geisner I M (2004) Predicting drinkingbehavior and alcohol-related problems among fraternity and sorority membersExamining the role of descriptive and injunctive norms Psychology of AddictiveBehaviors 18(3) 203ndash212

Larsen H Engels R C M E Granic I amp Overbeek G (2009) An experimental studyon imitation of alcohol consumption in same-sex dyads Alcohol andAlcoholism 44(3) 250ndash255 lthttpalcalcoxfordjournalsorgcontent443250abstractgt

Leary M R Tambor E S Terdal S K amp Downs D L (1995) Self-esteem as aninterpersonal monitor The sociometer hypothesis Journal of Personality and SocialPsychology 68(3) 518ndash530 doi1010370022-3514683518

Leone T Herman C P amp Pliner P (2008) Perceptions of undereaters A matter ofperspective Personality and Social Psychology Bulletin 34(12) 1737ndash1746

Leone T Pliner P amp Herman C P (2007) Influence of clear versus ambiguousnormative information on food intake Appetite 49 58ndash65

Lowe C F Horne P J Tapper K Bowdery M amp Egerton C (2004) Effects of a peermodelling and rewards-based intervention to increase fruit and vegetableconsumption in children European Journal of Clinical Nutrition 58(3) 510ndash522doi101038sjejcn1601838

Luke D A (2004) Multilevel modeling Thousand Oaks CA Sage Publications IncMcFerran B Dahl D W Fitzsimons G J amp Morales A C (2010a) Irsquoll have what

shersquos having Effects of social influence and body type on food choices of othersJournal of Consumer Research 36 1ndash15

McFerran B Dahl D W Fitzsimons G J amp Morales A C (2010b) Might anoverweight waitress make you eat more How the body type of others issucient to alter our food consumption Journal of Consumer Psychology 20(2)146ndash151 doi101016jjcps201003006

Meyers R J Villanueva M amp Smith J E (2005) The community reinforcementapproach History and new directions Journal of Cognitive Psychotherapy 19(3)247ndash260

Mollen S Rimal R N Ruiter R A C amp Kok G (2013) Healthy and unhealthy socialnorms and food selection Findings from a field-experiment Appetite 65 83ndash89doi101016jappet201301020

Nisbett R E amp Storms M D (1974) Cognitive and social determinants of food intakeIn H L R E Nisbett (Ed) Thought and feeling Cognitive alteration of feeling statesOxford England Aldine

Nolan J M Schultz P W Cialdini R B Goldstein N J amp Griskevicius V (2008)Normative social influence is underdetected Personality amp Social PsychologyBulletin 34(7) 913ndash923 doi1011770146167208316691

Oyserman D Fryberg S A amp Yoder N (2007) Identity-basedmotivation and healthJournal of Personality and Social Psychology 93(6) 1011ndash1027

Paxton S J Eisenberg M E amp Neumark-Sztainer D (2006) Prospective predictorsof body dissatisfaction in adolescent girls and boys A five-year longitudinal studyDevelopmental Psychology 42(5) 888ndash899

Paxton S J Schutz H K Wertheim E H amp Muir S L (1999) Friendship cliqueand peer influences on body image concerns dietary restraint extreme weight-loss behaviours and binge eating in adolescent girls Journal of AbnormalPsychology 108(2) 255ndash266

Pliner P amp Mann N (2004) Influence of social norms and palatability on amountconsumed and food choice Appetite 42(2) 227ndash237 doi101016jappet200312001

Polivy J Herman C P Younger J C amp Erskine B (1979) Effects of a model on eatingbehavior The induction of a restrained eating style Journal of Personality 47100ndash112

Prinsen S de Ridder D T D amp de Vet E (2013) Eating by example Effects ofenvironmental cues on dietary decisions Appetite 70 1ndash5 doi101016jappet201305023

Reverdy C Chesnel F Schlich P Koumlster E P amp Lange C (2008) Effect of sensoryeducation on willingness to taste novel food in children Appetite 51(1) 156ndash165doi101016jappet200801010

Robins R W Trzesniewski K H Tracy J L Gosling S D amp Potter J (2002) Globalself-esteem across the life span Psychology and Aging 17(3) 423ndash434doi1010370882-7974173423

Robinson E Benwell H amp Higgs S (2013a) Food intake norms increase and decreasesnack food intake in a remote confederate study Appetite 65 20ndash24 doi101016jappet201301010

Robinson E Fleming A amp Higgs S (2013b) Prompting healthier eating Testingthe use of health and social norm based messages Health Psychologydoi101037a0034213

Robinson E amp Higgs S (2012) Liking food less The impact of social influence onfood liking evaluations in female students PLoS ONE 7(11) e48858 doi101371journalpone0048858

Robinson E amp Higgs S (2013) Food choices in the presence of ldquohealthyrdquo andldquounhealthyrdquo eating partners The British Journal of Nutrition 109(4) 765ndash771doi101017S0007114512002000

Robinson E Kersbergen I Brunstrom J M amp Field M (2014a) Irsquom watching youAwareness that food consumption is being monitored is a demand characteristicin eating-behaviour experiments Appetite 83 19ndash25 doi101016jappet201407029

Robinson E Sharps M Price N amp Dallas R (2014b) Eating like you are overweightThe effect of overweight models on food intake in a remote confederate studyAppetite 82C 119ndash123 doi101016jappet201407019

Robinson E Tobias T Shaw L Freeman E amp Higgs S (2011) Social matching offood intake and the need for social acceptance Appetite 56(3) 747ndash752doi101016jappet201103001

Rodin J Silberstein L amp Striegel-Moore R (1984) Women and weight A normativediscontent Nebraska Symposium on Motivation 32 267ndash307

Romero N D Epstein L H amp Salvy S-J (2009) Peer modeling influences girlsrsquo snackintake Journal of the American Dietetic Association 109(1) 133ndash136 doi101016jjada200810005

ARTICLE IN PRESS



Rosenberg T (2011) Join the club How peer pressure can transform the world WWNorton amp Company

Rosenthal B amp Marx R D (1979) Modeling influences on the eating behavior ofsuccessful and unsuccessful dieters and untreated normal weight individualsAddictive Behaviors 4(3) 215ndash221

Rosenthal B amp McSweeney F K (1979) Modeling influences on eating behaviorAddictive Behaviors 4(3) 205ndash214

Roth D A Herman C P Polivy J amp Pliner P (2001) Self-presentational conflictin social eating situations A normative perspective Appetite 36 165ndash171

Rozin P (2005) The meaning of food in our lives A cross-cultural perspective oneating and well-being Journal of Nutrition Education and Behavior 37 S107ndashS112doi101016S1499-4046(06)60209-1

Salmon S J Fennis B M de Ridder D T D Adriaanse M A amp de Vet E (2014)Health on impulse When low self-control promotes healthy food choices HealthPsychology 33(2) 103ndash109 doi101037a0031785

Salvy S J Coelho J S Kieffer E amp Epstein L H (2007a) Effects of social contextson overweight and normal-weight childrenrsquos food intake Physiology amp Behavior92(5) 840ndash846 doi101016jphysbeh200706014Effects

Salvy S J de la Haye K Bowker J C amp Hermans R C J (2012) Influence of peersand friends on childrenrsquos and adolescentsrsquo eating and activity behaviors Physiologyamp Behavior 106(3) 369ndash378 doi101016jphysbeh201203022

Salvy S J Elmo A Nitecki L A Kluczynski M A amp Roemmich J N (2011) Influenceof parents and friends on childrenrsquos and adolescentsrsquo food intake and foodselection The American Journal of Clinical Nutrition 93 87ndash92 doi103945ajcn110002097

Salvy S J Jarrin D Paluch R Irfan N amp Pliner P (2007b) Effects of social influenceon eating in couples friends and strangers Appetite 49 92ndash99

Salvy S J Kieffer E amp Epstein L H (2008a) Effects of social context on overweightand normal-weight childrenrsquos food selection Eating Behaviors 9(2) 190ndash196doi101016jeatbeh200708001

Salvy S J Romero N Paluch R amp Epstein L H (2007c) Peer influence onpre-adolescent girlsrsquo snack intake Effects of weight status Appetite 49 177ndash182

Salvy S J Vartanian L R Coelho J S Jarrin D amp Pliner P P (2008b) The role offamiliarity on modeling of eating and food consumption in children Appetite50(2ndash3) 514ndash518 doi101016jappet200710009

Schachter S (1971) Some extraordinary facts about obese humans and rats AmericanPsychologist 26 129ndash144

Spanos S Vartanian L R Herman C P amp Polivy J (2013) Failure to report socialinfluences on food intake Lack of awareness or motivated denial HealthPsychology doi101037hea0000008

Stok F M de Ridder D T D de Vet E amp de Wit J B F (2012) Minority talks Theinfluence of descriptive social norms on fruit intake Psychology amp Health 27(8)956ndash970 doi101080088704462011635303

Stok F M de Ridder D T D de Vet E amp de Wit J B F (2014) Donrsquot tell me whatI should do but what others do The influence of descriptive and injunctive peernorms on fruit consumption in adolescents British Journal of Health Psychology19(1) 52ndash64 doi101111bjhp12030

Sweller J Ayres P amp Kalyuga S (2011) Cognitive Load Theory (Vol 1) New YorkSpringer

Tajfel H amp Turner J C (1979) An integrative theory of intergroup conflict In WG Austin amp S Worehel (Eds) The social psychology of intergroup relations (pp33ndash47) Monterey BrooksCole

Temple J L Giacomelli A M Kent K M Roemmich J N amp Epstein L H (2007)Television watching increases motivated responding for food and energy intakein children The American Journal of Clinical Nutrition 85(2) 355ndash361 lthttpajcnnutritionorgcontent852355abstractgt

Thaler R H amp Sunstein C R (2008) Nudge Improving decisions about health wealthand happiness New Haven CT Yale University Press

Thompson J K amp Stice E (2001) Thin-ideal internalization Mounting evidence fora new risk factor for body-image disturbance and eating pathology CurrentDirections in Psychological Science 10(5) 181ndash183 lthttpwwwjstororgstable20182734gt

Tibbs T Haire-Joshu D Schechtman K B Brownson R C Nanney M S HoustonC et al (2001) The relationship between parental modeling eating patternsand dietary intake among AfricanndashAmerican parents Journal of the AmericanDietetic Association 101 535ndash541

Turner J C (1991) Social influence Milton Keynes UK Open University PressTurner J C (1999) Some current issues in research on social identity and

self-categorisation theories In N Ellemers R Spears amp B Doosje (Eds)Social identity Context commitment content (pp 6ndash34) Oxford BlackwellPublishers

Turner J C Hogg M A Oakes P J Reicher S D amp Wetherell M S(1987) Rediscovering the social group A self-categorization theory OxfordBlackwell

Turner J C amp Oakes P J (1989) Self-categorization theory and social influence InP B Paulus (Ed) The psychology of group influence (2nd ed pp 233ndash275)Hillsdale NJ Erlbaum

van Baaren R B Holland R W Kawakami K amp van Knippenberg A (2004) Mimicryand prosocial behavior Psychological Science 15(1) 71ndash74 lthttppsssagepubcomcontent15171abstractgt

Vartanian L R (2009) When the body defines the self Self-concept clarityinternalization and body image Journal of Social and Clinical Psychology 28(1)94ndash126

Vartanian L R Herman C P amp Polivy J (2007) Consumption stereotypes andimpression management How you are what you eat Appetite 48(3) 265ndash277doi101016jappet200610008

Vartanian L R Herman C P amp Wansink B (2008) Are we aware of the externalfactors that influence our food intake Health Psychology 27(5) 533ndash538doi1010370278-6133275533

Vartanian L R Sokol N Herman C P amp Polivy J (2013) Social models provide anorm of appropriate food intake for young women PLoS ONE 8(11) e79268doi101371journalpone0079268

Wansink B amp Sobal J (2007) Mindless eating The 200 daily food decisions weoverlook Environment and Behavior 39(1) 106ndash123 lthttpeabsagepubcomcontent391106abstractgt

Yamasaki M Mizdzuno K amp Aoyama K (2007) The effect of food consumptionby others on the consumption of food by experimental subjects The studysituation in which the experimenter cannot know how much subjects eat TheJapanese Journal of Social Psychology 23(2) 173ndash180

ARTICLE IN PRESS



Social modeling of eating A review of when and why social influence affects food intake and choice

Introduction


Inclusion criteria



Contextual factors

Type of food

Live versus remote confederate

Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity

Similarity and shared group membership








Conclusion

References

in the case of modeling because both groups are influenced bynormative external cues (Herman amp Polivy 2008) Instead theseeffects were found to have a strong and pervasive influence on bothhealthy-weight and overweight individualsrsquo eating behaviors Al-though the reproducibility of these effects was easily and repeatedlydemonstrated the question of why modeling occurs has provedmoredicult to answer definitively That is what purpose does model-ing serve psychologically that might explain why it is so stronglypreserved and generalizable

Over the decades of modeling research a variety of explana-tions have been put forward to understand the effect The mostdominant interpretation however is that modeling of food intakeis an example of a broader phenomenon of social influence and thatgeneral theories of normative behavior might help to understandwhy people adapt their food intake to that of others Using a nor-mative approach Herman and his colleagues proposed that theprincipal regulatory influence on eating in social contexts is pe-oplersquos beliefs about what or howmuch is appropriate to eat (Hermanamp Polivy 2005 Herman Roth amp Polivy 2003b) According to thismodel people conform to othersrsquo eating because they see the amounteaten by others as an indicator of how much one can or should eatwithout eating excessively

Although the literature seems to approach consensus on theutility of this normative model there has not been a systematicreview of modeling studies The lack of a comprehensive reviewimpedes our ability to ascertain from the extant modeling litera-ture (a) when and why social modeling shapes eating behaviorand (b) how to translate this knowledge to inform applied prac-tice aimed at increasing healthy eating behavior Therefore ouroverarching aim is to review the literature on how peoplersquos foodchoice and intake is affected by modeling and on the basis of thesefindings propose new research directions that might help us togain insight into the robustness or underlying mechanisms of mod-eling We start by reviewing typical methodological approaches tothe study of modeling before summarizing the key findings fromour systematic review of 69modeling experiments We then discusstheoretical and practical implications of these findings


In past research several strategies have been used to investi-gate modeling effects on eating Both observational as well ascorrelational studies have found that people adapt their intake tothat of their eating companion and that those who are eating to-gether converge upon an eating norm (eg Salvy Romero Paluchamp Epstein 2007c Salvy Vartanian Coelho Jarrin amp Pliner 2008b)This occurs such that the variance among participants in their foodintake is reduced when eating together However both statisticaland theoretical concerns arise when interpreting research whereparticipants model one another Firstly because food intake is non-independent between participants an appropriate statistical methodof analysis would be multi-level modeling (Luke 2004) ndash al-though often this is not performed Furthermore without randomassignment it is dicult to rule out the possibility that non-socialfactors such as pre-existing similarity or eating attitudes are re-sponsible for conformity effects between eating companions Finallyin a scenario in which both co-eaters are free to choose the typeor amount of food to consume it is dicult to determine whichperson is modeling and which person is being modeled In partbecause of these concerns an experimental design in which theintake andor choice of one co-eater (ie the confederate) is pre-determined by the experimenter has arguably become the gold-standard for research on the modeling of food intake This paradigmenables researchers to investigate modeling behavior without anypotential confounds related to selection or non-social processes Insome studies participants are providedwith a non-food related cover

story for the experiment (eg Bevelander Anschuumltz CreemersKleinjan amp Engels 2013a Cruwys et al 2012 Hermans SalvyLarsen amp Engels 2012c) In these experiments participants believethat food is incidental to the research question In other studies par-ticipants are told that they are participating in a taste-test studyand are asked to complete questionnaires related to their experi-ence of the food items (eg Goldman Herman amp Polivy 1991Vartanian Sokol Herman amp Polivy 2013) In these studies partici-pants are aware of the centrality of the food to the experimenthowever the researchersrsquo interest in social influence and the amountof food consumed remains opaque

The sheer robustness of modeling has allowed researchers toalso develop a more ldquolight-touchrdquo technique for communicatingsocial norms to participants known as the remote-confederateparadigm (cf Roth Herman Polivy amp Pliner 2001) In studies uti-lizing this design the confederate providing the norm for foodchoice or intake is not physically present Rather participants areprovided normative information (while concealing the aim ofthe study with a cover story) by exposing them to either writteninformation about the amount consumed by previous partici-pants (eg in the form of a list on a table which was supposedlyused to determine how much food needed to be ordered by theexperimenters) or by exposure to a remotemodel selecting or eatingfood on a video or computer screen (Bevelander et al 2013aBevelander Anschuumltz amp Engels 2012b Hermans et al 2012cRomero Epstein amp Salvy 2009) Given that both live and remoteconfederate designs have been found to induce modeling effectson eating (cf Feeney Polivy Pliner amp Sullivan 2011) and are ableto infer strong cause and effect relationships we summarize find-ings of studies in which the eating norm is induced by either typeof confederate

Inclusion criteria

To find relevant English-language empirical research on mod-eling effects on food choice and food intake a literature search ofPubMed and Google Scholar was conducted using the following keywords lsquomodelingrsquo lsquomatchingrsquo lsquosocial influencersquo lsquonormative influ-encersquo lsquoeatingrsquo lsquofood choicersquo lsquofood intakersquo These key words wereused in combinations of two to include one theoretical keyword (iemodeling matching social influence normative influence) and onebehavioral keyword (ie eating food choice food intake) The ref-erence lists and citations of eligible publications were also reviewedto identify pertinent literature A criterion for inclusion in the reviewwas that the study had an experimental design in which either foodchoice or food intake was experimentally manipulated by a socialreferent (using either a live or remote confederate) Studies in whichparticipant dyads or groups were examined in a free eating para-digmwithout a confederate were therefore not included (eg SalvyJarrin Paluch Irfan amp Pliner 2007b Salvy Kieffer amp Epstein 2008a)Furthermore we included only those studies with a dependent vari-able that was amount of food consumed or food choice (measuredin a concrete behavioral fashion not intentions only) Table 1 showsa complete list of all the modeling studies that were included inthis review Where possible however we also discuss studies in ourreview that did not meet our inclusion criteria but which provid-ed additional insight into the dynamic process of modeling Sixty-nine studies (in 49 articles) were identified that met these selectioncriteria reporting on over 5800 experimental participants Of thesethe majority (58) measured food intake or whether participants ateat all as the dependent variable of interest whereas only 11 in-vestigated participantsrsquo choice between at least two food alternativesAs can be seen in Table 1 studies conducted with live confeder-ates (42) or with some form of remote confederate (27) are wellrepresented

ARTICLE IN PRESS





Participantgender



Evidenceofmodeling


1 Nisbett andStorms



2 Harper andSanders







4 Rosenthal andMarx





























Y






N



Y


Y









AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

3TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling








Y

21 Pliner andMann













114 M = 208 M + F R Y







28 Greenhalgh etal





















AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

4TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling






Y











39 51 M = 213 F L Y








Y

























AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

5TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling













unhealthy inStudy 3

Y





































Inpress

HealthPsychology





70 M = 191 M + F R Y

AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

6TCruw

yset

alAppetite

(2014)

ndash




Contextual factors









ARTICLE IN PRESS





Individual factors









ARTICLE IN PRESS













Social factors






ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS












Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References



Participantgender



Evidenceofmodeling


1 Nisbett andStorms



2 Harper andSanders







4 Rosenthal andMarx





























Y






N



Y


Y









AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

3TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling








Y

21 Pliner andMann













114 M = 208 M + F R Y







28 Greenhalgh etal





















AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

4TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling






Y











39 51 M = 213 F L Y








Y

























AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

5TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling













unhealthy inStudy 3

Y





































Inpress

HealthPsychology





70 M = 191 M + F R Y

AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

6TCruw

yset

alAppetite

(2014)

ndash




Contextual factors









ARTICLE IN PRESS





Individual factors









ARTICLE IN PRESS













Social factors






ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS












Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References

Table 1 (continued)


Participantgender



Evidenceofmodeling








Y

21 Pliner andMann













114 M = 208 M + F R Y







28 Greenhalgh etal





















AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

4TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling






Y











39 51 M = 213 F L Y








Y

























AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

5TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling













unhealthy inStudy 3

Y





































Inpress

HealthPsychology





70 M = 191 M + F R Y

AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

6TCruw

yset

alAppetite

(2014)

ndash




Contextual factors









ARTICLE IN PRESS





Individual factors









ARTICLE IN PRESS













Social factors






ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS












Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References

Table 1 (continued)


Participantgender



Evidenceofmodeling






Y











39 51 M = 213 F L Y








Y

























AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

5TCruw

yset

alAppetite

(2014)

ndash

Table 1 (continued)


Participantgender



Evidenceofmodeling













unhealthy inStudy 3

Y





































Inpress

HealthPsychology





70 M = 191 M + F R Y

AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

6TCruw

yset

alAppetite

(2014)

ndash




Contextual factors









ARTICLE IN PRESS





Individual factors









ARTICLE IN PRESS













Social factors






ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS












Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References

Table 1 (continued)


Participantgender



Evidenceofmodeling













unhealthy inStudy 3

Y





































Inpress

HealthPsychology





70 M = 191 M + F R Y

AR

TICLE

INP

RE

SS

Pleasecite

thisarticle

inpress

asTegan

Cruw

ysK

irstenE

B

evelanderR

oelC

JH

ermans

Socialm

odelingof

eatingAreview

ofw

henand

why

socialinfluence

affectsfood

intakeand

choiceAppetite

(2014)doi 101016jappet201408035

6TCruw

yset

alAppetite

(2014)

ndash




Contextual factors









ARTICLE IN PRESS





Individual factors









ARTICLE IN PRESS













Social factors






ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS












Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References




Contextual factors









ARTICLE IN PRESS





Individual factors









ARTICLE IN PRESS













Social factors






ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS












Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References



Individual factors









ARTICLE IN PRESS













Social factors






ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS












Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References











Social factors






ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS












Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References












ARTICLE IN PRESS














ARTICLE IN PRESS














ARTICLE IN PRESS












Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References












ARTICLE IN PRESS














ARTICLE IN PRESS












Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References












ARTICLE IN PRESS












Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References










Conclusion



ARTICLE IN PRESS




References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References


References






















































ARTICLE IN PRESS




























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References


























































ARTICLE IN PRESS





































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References



































ARTICLE IN PRESS




Introduction


Inclusion criteria



Contextual factors

Type of food


Individual factors

Hunger and satiety

Sex

Age

Body weight

Impulsivity

Eating goals

Social factors

Type of social norm

Desire to affiliate

Familiarity









Conclusion

References

cruwys, 2014

Documents

modeling of food intake

modeling of food choice

model modeling

significant modeling

goals moderate modeling

modeling phenomenon

social context

social referent