Вы находитесь на странице: 1из 8

576607

research-article2015
HPQ0010.1177/1359105315576607Journal of Health PsychologyBrindal et al.

Article

Journal of Health Psychology

Eating in groups: Do multiple


2015, Vol. 20(5) 483­–489
© The Author(s) 2015
Reprints and permissions:
social influences affect intake sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1359105315576607
in a fast-food restaurant? hpq.sagepub.com

Emily Brindal1,3, Carlene Wilson2, Philip Mohr3


and Gary Wittert3

Abstract
This study investigated multiple social influences to determine whether they affect amount eaten at a fast-food
environment. Using observational methods, data on meal duration, foods eaten and personal characteristics
were collected for 157 McDonald’s patrons. Analysis of covariance revealed that female diners ate less
kilojoules when eating in mixed- versus same-sex groups (adjusted difference = 967 kJ, p < .05), while male
diners eating in mixed-sex company ate more in groups compared to pairs (adjusted difference = 1067 kJ,
p = .019). Influences to increase and restrict the amount eaten can operate simultaneously in an eating
environment with gender a critical factor for consideration.

Keywords
fast food, groups, minimal eating, norms, social influence

Many consumers remain uncertain about how Chaiken, 1990). In contrast, matching norms dic-
much and what to eat. It has been hypothesised tate changing intake (increasing or decreasing)
that uncertainty surrounding ‘appropriate’ con- relative to that of a co-eater (Roth et al., 2001).
sumption (biologically and socially) drives con- Across multiple experimental studies, a confeder-
sumers to search for contextual norms to guide ate (i.e. an actor following the instructions of the
their behaviour and makes them susceptible to a experimenters) who alters the amount of snack
variety of social influences when eating in items (such as cookies or candies) they consume
social contexts. Three key areas of social influ- has been able to alter the amount a subject con-
ence that could affect eating behaviour are min- sumes regardless of gender effects (e.g. De Luca
imal eating norms, matching norms and social and Spigelman, 1979). Finally, social facilitation is
facilitation (Herman et al., 2003). the documented tendency for the presence of other
In one of the first studies to describe minimal people to increase the amount eaten. De Castro
eating, Chaiken and Pliner (1987) observed that
females described as eating smaller meals were 1CSIRO, Australia
rated as more feminine by both males and females. 2FlindersUniversity, Australia
This result was interpreted as indicative of mini- 3University of Adelaide, Australia

mal eating norms – women were expected to eat


Corresponding author:
less than men to convey a feminine image. Emily Brindal, CSIRO, Food and Nutrition, P.O. Box
Experimental studies have subsequently supported 10041, Adelaide, SA 5000, Australia.
the existence of minimal eating norms (Pliner and Email: emily.brindal@csiro.au
484 Journal of Health Psychology 20(5)

(1990) and De Castro and Brewer (1992) have was published prior to articles on social facilita-
repeatedly shown a positive relationship between tion and the expansion of fast-food restaurants
the number of people present and calorie con- which now provide a wider variety of food
sumption using diet diaries. They have suggested choices that people can use to alter their intake.
that this occurs due to time-extension – eating in In various types of restaurants, these authors
the presence of other people extends meal dura- observed and recorded group size, gender com-
tion, which then increases the amount eaten (De position and foods of the people eating. In fast-
Castro, 1994). food restaurants, people were observed eating
Despite empirical support for these various more when eating in a mixed-sex group com-
social influences, how these effects may occur pared to same-sex and opposite-sex groups.
simultaneously in a real-world eating environ- Consistent with social facilitation, the people
ment is less clear. This is particularly true for eating in groups ate more. Finally, observations
women who may experience conflicting norma- supported minimal eating norms; women ate a
tive pressures. A female eating in a group with similar amount to the men when in a small group,
male companions may (1) eat minimally in but when in larger groups, females ate signifi-
order to create a feminine impression (minimal cantly less than males.
eating), (2) increase her intake to match that of The objective of the current study was to
her male companions to make a good impres- investigate whether multiple social influences
sion (matching) or (3) simply eat more than she on eating behaviour alter intake in a fast-food
would alone because she is in a group and eat- environment while assessing interactions
ing for longer (social facilitation). Matching or between sex of the eater and co-eaters and
minimal eating norms could guide behaviour group size which may all alter eating behaviour
depending on the individual, their eating con- according to various social influence theories.
text and the salience of environmental cues
(Salvy et al., 2007). Yet, it has also been sug-
Method
gested that the effects of social facilitation are
pervasive across a variety of situations and The Human Research Ethics Committee at the
interactions (De Castro, 1994). School of Psychology in the University of
Chronic overconsumption and restrictive Adelaide approved this study.
eating behaviours (which augmentation and
inhibition norms promote) could lead to out-
Sampling
comes such as weight gain and eating disorders,
in the long term. Furthermore, a more in-depth McDonald’s restaurants were selected for
knowledge of when, why and how norms oper- observation due to their large market share and
ate in eating environments is required because menu diversity. All patrons who entered the res-
adherence and violation to norms can result in taurant during an observation session and sat
positive (i.e. reinforcing self-concept) and neg- with at least one other person at one of the
ative outcomes (i.e. judgements of poor inter- tables in a targeted area were observed, and
personal characteristics), respectively, for an information about their behaviour was recorded.
individual (Wood et al., 1997; Yantcheva and Data were recorded for one representative who
Brindal, 2013). sat at a predetermined seat at the table with
Fast-food restaurants provide a unique oppor- other diners. Piloting was used to identify which
tunity for observational research because of seat the observer could occupy to achieve the
their standardised menus and set portion sizes. most privacy and best view of other seats. Any
These styles of food are also the target of many group with a child member was excluded from
public health campaigns. One previous study observation because the focus of the study was
has attempted to understand social influences on on normative influences on adult behaviour.
eating behaviour in a real-world fast-food envi- People who started eating alone but finished in
ronment (Klesges et al., 1984), but this research a group were also excluded from observation.
Brindal et al. 485

For this study, a restaurant in the Central In a very small percentage of cases (six
Business District of Adelaide, South Australia, observations), people returned to the counter
was chosen because initial piloting of several during their eating occasion to order another
locations revealed this to be the busiest. item. All returned to their table to consume their
Approximately 10 eligible people ate in the tar- items. Five returned with dessert items and one
geted area of this restaurant during a 60-minute returned with a drink and wrap. These items
period, while at some of the outer suburban res- were added to respective order data, and the
taurants, as little as one or two observations time taken to order was subtracted from the final
were recorded during the same period. meal duration. In seven cases, people left with
incomplete soft drinks. Sharing items was docu-
mented in 13 cases (12 fries; 1 Nuggets). In
Procedure
these instances, we divided amounts equally by
A researcher sat in the restaurant between 4 and the number of people sharing. Three people did
7 p.m. during weekdays in Autumn 2007 and not complete their foods – order adjustments
unobtrusively entered data directly into a per- were made to estimations of intakes based on
sonal digital assistant (PDA). Categories of data the notes made during the observation.
tested and refined in previous research (Brindal
et al., 2011) included the following: the sex of
Analysis
the nominated individual; their age group (‘15–
25’, ‘26–35’, ‘36–45’ and ‘46+’ years); their esti- Item descriptions of foods eaten were used to
mated weight status (‘overweight/obese’ or calculate the total kilojoules consumed accord-
‘not’); the characteristics of others present in the ing to nutrition information from McDonalds’
group (including their sex, age group and weight website (McDonald’s Corporation, 2006). Soft
status); the time eating started (from the moment drinks were recorded but not included in this
the person started eating at their table); the time calculation due to the inability to reliably distin-
the eating occasion ended (recorded when a per- guish diet and non-diet soft drinks. This also
son left the table) and the foods eaten according resulted in the exclusion of 8 of the initial 165
to their trade names (e.g. Big Mac® and small observations as these people consumed only a
fries). Consistent with previous observational drink. Item customisations (e.g. the removal of
research, if members left or joined the group cheese or a pickle) were not accounted for as
before all members had left the table during an these also could not be observed.
observation, the group size was adjusted accord- A 2 (male or female) × 2 (same-sex or mixed-
ingly (Bell and Pliner, 2003). While data were sex group) × 2 (pair or larger group) analysis of
recorded, the researcher consumed purchased covariance (ANCOVA) of energy intake from
items to justify prolonged occupation of a seat. fast-food items controlling for age (under
Menu items were easily identifiable through 25 years or not) and weight (overweight or not)
unique packaging. A guide including photo- was performed. Observation time encompassed
graphs and descriptions of all packaging and two possible eating occasions (afternoon snack
sizes was produced and included in the PDA to and dinner) and a variable describing ‘time of
aid in the identification of products. A visual day’ (started at 4–5:30 p.m. vs 5:31–7 p.m.) was
guide was also developed to assist in the classifi- created for inclusion in the model as a covari-
cation of body size categories. For nine initial ate. Group size was categorised into two levels
sessions, two observers recorded data until in order to optimise cell sizes for the factorial
100 per cent agreement between recorders was analysis. All statistical analyses were performed
reached. Where differences arose, coders delib- using Statistical Package for Social Sciences
erated and adjusted these data for inclusion in the Version 15 using an alpha level of p < .05. In the
final analyses. After this, all data were taken absence of ad hoc predictions, significant main
from observations recorded by a single observer. and interaction effects were followed up with
486 Journal of Health Psychology 20(5)

Table 1.  Means and standard deviations for energy consumed (in kJ) by participant sex, group
composition and group size.

Male Female All

  n M SD n M SD n M SD
Same sex Pair 25 3513 1290.3 22 3005 1137.9 47 3276 1235.1
Group (3+) 16 3700 772.4 12 3651 1593.3 28 3679 1168.9
Total 41 3586 1109.6 34 3233b 1329.7 75 3426 1218.7
Mixed Pair 36 3438a 1399.2 23 2933 899.7 59 3241 1245.1
sex Group (3+) 9 4500a 1482.8 14 2672 805.2 23 3388 1419.5
Total 45 3650 1463.5 37 2834b 863.3 82 3282 1289.1
Overall Pair 61 3469 1345.1 45 2968 1012.0 106 3256 1234.9
Group (3+) 25 3988d 1122.4 26 3124d 1304.4 51 3547 1282.9
Total 86 3620c 1299.7 71 3025c 1121.2 157 3351 1254.0

SD: standard deviation.


Mean values are not adjusted for part of day, age and weight status. Values with a shared superscript are significantly
different at p < .05.

post hoc pairwise comparisons using Bonferroni effect for participant sex (F(1, 156) = 13.47,
adjustments. p < .001), signifying that males consumed sig-
nificantly more than females. There were two
higher order effects for participant sex: a two-
Results way interaction between participant sex and
The final sample (n = 157) comprised slightly group composition (F(3, 153) = 4.05, p = .046),
more males (54.8%) than females. Most people which was qualified by a three-way interaction
were rated as ‘not overweight/obese’ (82.3%) and between participant sex, group composition and
‘15–25’ years old (83.4%). A total of 7.6 per cent group size (F(7, 147) = 4.75, p = .031).
of the sample were ‘26–35’ years old, while 4.5 per The three-way interaction is shown in Figure 1.
cent were in either the ‘36–45’ or ‘45+’ years age Post hoc pairwise comparisons revealed that
group. The mean time spent dining was 15.45 min- group size had a significant effect for males in
utes (standard deviation (SD) = 7.20 minutes). The mixed company, with intake higher in groups ver-
mean energy consumed was 3351 kJ (SD = 1254 kJ, sus dyads (mean difference = 1067.1, p = .019).
range = 582–7500 kJ). Most people ate in a dyad Group composition was significantly associated
(67.5%), with a further 19.7 per cent eating in a with the energy intake of the females observed,
group of three, 9.6 per cent in a group of four and with those in the presence of mixed-sex company
3.4 per cent eating in groups with five or more eating less than those with same-sex company
people. Energy consumed was positively associ- (mean difference = 966.9, p = .043). Finally, sex
ated with the amount of time spent eating differences were only observed in mixed-sex
(Spearman’s ρ = .257, p = .001) but not with the groups with males consuming more than females
total others present. (mean difference = 1869.6, p < .001).
Means for kilojoules consumed are pre-
sented by participant sex, sex composition of
Discussion
the group and group size in Table 1. ANCOVA
revealed no main effects for the sex composi- The aim of this study was to explore whether
tion of the group (F(1, 156) = 0.21, p = .650) or multiple social influences may operate in a fast-
group size (F(1, 156) = 3.27, p = .073) for energy food eating environment. Our results provided
intake from fast-food items. There was a main support for potential social influences on food
Brindal et al. 487

Figure 1.  Mean energy (kJ) consumed at an eating occasion controlling for part of day, age and weight
status displayed to represent the three-way interaction between participant sex (male/female), gender
composition of the group (same/mixed sex) and group size (pair/group).
a: male/mixed sex/group versus male/mixed sex/pair (p < .05); b: male/mixed/group versus female/mixed/group (p < .001);
c: female/mixed sex/group versus female/same sex/group (p < .05).
Error bars indicate ±1 standard error.

consumption, but these effects were qualified were being adhered to, one would reasonably
by both group size and sex. In a larger group, expect the presence of a male to increase intake
females ate less when a male was present rela- in females (as males have higher kilojoule
tive to when eating with same-sex companions. requirements). Our observations, contrary to
Simultaneously, it appeared that males were this, are in line with previous suggestions that
influenced by social facilitation, as those males minimal eating is more likely to occur than
in mixed-sex company consumed more energy matching in the presence of competing norma-
when in larger groups relative to pairs. tive cues (Roth et al., 2001). This is also in line
It was interesting that all of the effects were with the description of injunctive norms which
witnessed when groups were mixed sex. For account for the way one would be expected to
example, it appeared that males ate more than behave versus descriptive norms which simply
females when considering the main effect of sex, guide behaviour in a given situation (Reno
but intake was actually only significantly differ- et al., 1993). Given the negative judgements
ent between males and females in large, mixed- that have been associated with consuming fast
sex groups when factors of group size and gender foods (Dunn et al., 2008), a fast-food environ-
composition were also accounted for. ment may trigger heightened self-presentational
When in larger groups, the intake of female concerns and a female eater’s motivation to
diners supported the presence of a minimal eat- assert their femininity through reduced intake.
ing norm – females ate less in the presence of This could also satisfy injunctive norms con-
mixed-sex company compared to those females cerning how women ‘should’ behave when eat-
eating in same-sex company. If matching norms ing. It is important to note that minimal eating
488 Journal of Health Psychology 20(5)

was only observed in large-sized groups and not operate much earlier than during the actual eat-
in dyads. In larger groups, there may be more ing occasion. The demonstration of social influ-
pressure to adhere to eating norms. Alternatively, ence in shopping behaviour also supports this
familiarity may have been an important moder- suggestion (Sommer et al., 1992). If these effects
ator for those in pairs as there may have been reliably occur before eating, we could be more
higher levels of familiarity between compan- confident that social influences are a result of
ions in smaller groups which reduced norma- adherence to wider normative social cues rather
tive demands (Leary et al., 1994). than micro-level situational demands.
A social facilitation effect was witnessed but In the current research, the use of observa-
only for males and again only in the presence of tional methods, although not as controlled as
mixed-sex company. Males in same-sex com- methods used in experimental settings, allowed
pany ate similar amounts regardless of group different social influences to be witnessed
size. Therefore, it is likely that there is more to simultaneously in a natural environment.
social facilitation effects than the mere presence Observational methods provide an ecologically
of others. For example, it is possible a norm valid ‘snapshot’ data but are limited by the
opposite to minimal eating exists for males – scope of the information that can be recorded
one where males increase their intake to assert and limit generalisability to other contexts. For
their masculinity. Differences between males example, despite approximately 80 per cent of
eating in same versus mixed company failed to the sample ordering a soft drink, these could not
reach significance, reducing the plausibility of be accurately identified as diet or non-diet and
this idea; however, sparse research on the were consequently excluded from kilojoule cal-
effects of social influence on males offers little culations to provide a conservative estimation
insight. Given minimal suggestion of social of energy intake and focus discussion on eating
facilitation, we did not explore the time-exten- behaviour. Furthermore, during piloting, it was
sion hypothesis in depth. Our sample size revealed that it may be the case that the CBD
necessitated the categorisation of group size restaurant attracted less families and more
which may have limited the ability to determine young adults as it is located within walking dis-
linear effects of social facilitation. However, tance of a university.
preliminary correlations revealed limited direct The current results are important for under-
relationship between the number of people pre- standing how social influences on eating
sent and the amount eaten. Structural equation behaviour interact in eating environments.
modelling in a larger sample would be needed They suggest that when eating in a social con-
to assess if this is the result of the mediating text, interactions between the sex composition
effect of time or the absence of an effect of the group and the sex of the eater warrant
(Feunekes et al., 1995). consideration. This study also suggests that
Theorising about social influence has been collection of more detailed diet diary data may
largely on the basis of studies that have created be required to further assess evidence of social
artificial scenarios by offering subjects ‘snacks’ influences, in what circumstances they may be
in a laboratory (Pliner and Chaiken, 1990; Roth witnessed (e.g. when eating with certain types
et al., 2001) or through the use of diet diaries of people or when eating certain types of food)
(De Castro, 1990) which fail to capture the and whether social influences are more potent
amount a person chooses to serve themselves at the time of consumption or at the food selec-
and then actually consumes. In the ‘real-world’, tion stage.
social influences may affect food selection
behaviour rather than intake behaviour at the Acknowledgements
time of eating. We observed that most people This study was performed in partial requirement for
consumed all of their items and few returned for a PhD at CSIRO Animal, Food and Health Sciences
additional items. Therefore, these effects may and The University of Adelaide.
Brindal et al. 489

Funding by meal duration. Physiology & Behavior 58:


551–558.
The PhD scholarship was funded by a Linkage Grant
Herman CP, Fitzgerald NE and Polivy J (2003) The
from the Australian Research Council with industry
influence of social norms on hunger ratings and
partners the South Australian Health Department and
eating. Appetite 41: 15–20.
McDonald’s Australia. Neither industry partner had
Klesges RC, Bartsch D, Norwood JD, et al. (1984)
a role in the designing, analysing or interpreting the
The effects of selected social and environmental
data or in the drafting, review or approval of the
variables on the eating behaviour of adults in
manuscript.
the natural environment. International Journal
of Eating Disorders 3: 35–41.
References Leary MR, Nezlek JB, Downs D, et al. (1994) Self-
Bell R and Pliner PL (2003) Time to eat: The rela- presentation in everyday interactions – Effects
tionship between the number of people eat- of target familiarity and gender composition.
ing and meal duration in three lunch settings. Journal of Personality and Social Psychology
Appetite 41: 215–218. 67: 664–673.
Brindal E, Wilson C, Mohr P, et al. (2011) Does McDonald’s Corporation (2006) Nutrition informa-
meal duration predict amount consumed in lone tion. Available at: http://mcdonalds.com.au/
diners? An evaluation of the time-extension PDFs/Master_Nutrition_Information.pdf
hypothesis. Appetite 57: 77–79. Pliner P and Chaiken S (1990) Eating, social motives
Chaiken S and Pliner P (1987) Women, but not men, and self-presentation in women and men.
are what they eat – The effect of meal size and Journal of Experimental Social Psychology 26:
gender on perceived femininity and masculin- 204–254.
ity. Personality and Social Psychology Bulletin Reno RR, Cialdini RB and Kallgren CA (1993)
13: 166–176. The transsituational influence of social norms.
De Castro JM (1990) Social facilitation of duration and Journal of Personality and Social Psychology
size but not rate of the spontaneous meal intake of 64: 104–112.
humans. Physiology & Behavior 47: 1129–1135. Roth DA, Herman CP, Polivy J, et al. (2001) Self-
De Castro JM (1994) Family and friends produce presentational conflict in social eating situa-
greater social facilitation of food-intake than tions: A normative perspective. Appetite 36:
other companions. Physiology & Behavior 56: 165–171.
445–455. Salvy SJ, Jarrin D, Paluch R, et al. (2007) Effects
De Castro JM and Brewer EM (1992) The amount of social influence on eating in couples, friends
eaten in meals by humans is a power function and strangers. Appetite 49: 92–99.
of the number of people present. Physiology & Sommer R, Wynes M and Brinkley G (1992) Social
Behavior 51: 121–125. facilitation effects in shopping behavior.
De Luca RV and Spigelman MN (1979) Effects of Environment and Behavior 24: 285–297.
models on food intake of obese and non-obese Wood W, Christensen PN, Hebl MR, et al. (1997)
female college students. Canadian Journal of Conformity to sex-typed norms, affect, and the
Behavioural Science 11: 124–129. self-concept. Journal of Personality and Social
Dunn KI, Mohr PB, Wilson CJ, et al. (2008) Beliefs Psychology 73: 523–535.
about fast food in Australia: A qualitative analy- Yantcheva B and Brindal E (2013) How much does
sis. Appetite 51: 331–334. what you eat matter? The potential role of meal
Feunekes GI, de Graaf C and Vanstaveren WA (1995) size, fat content, and gender on ratings of desir-
Social facilitation of food-intake is mediated ability. Eating Behaviors 14: 285–290.
Copyright of Journal of Health Psychology is the property of Sage Publications, Ltd. and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.

Вам также может понравиться