Addictive Behaviors 35 (2010) 49–52

Contents lists available at ScienceDirect

Addictive Behaviors

Short communication

Peer influence in a micro-perspective: Imitation of alcoholic and

non-alcoholic beverages
Helle Larsen a,⁎, Rutger C.M.E. Engels a, Pierre M. Souren a, Isabela Granic b, Geertjan Overbeek a,1
a
Radboud University Nijmegen, The Netherlands
b
The Hospital for Sick Children, Toronto, Canada

a r t i c l e i n f o a b s t r a c t

Keywords: Ample experimental research has found evidence for imitation of alcohol consumption in social encounters.
Alcohol However, these studies cannot reveal whether imitation is specifically related to alcohol and not to
Imitation consumption in general. We investigated whether imitation is more evident when peers drink alcohol
Ad lib drinking
compared to other beverages. We observed sipping behavior during a 30-minute interaction between same-
Semi-naturalistic context
sex confederates and participants in an ad lib semi-naturalistic drinking context (bar lab). We expected a
Multilevel analysis
stronger imitation effect when both participant and confederate drank alcoholic beverages. A random
occasion multilevel analysis was conducted to take repeated measurements into account. Findings showed
that participants imitated the sips of the confederates, but that the likelihood of participants imitating a sip
was lower when confederates were drinking alcoholic beverages and participants non-alcoholic beverages
compared to when both were consuming alcohol.
© 2009 Elsevier Ltd. All rights reserved.

1. Introduction
Numerous experimental studies demonstrate that individuals
consume more alcohol when they are with someone who drinks
(e.g., Caudill & Kong, 2001; Collins, Parks, & Marlatt, 1985; Larsen,
Engels, Granic, & Overbeek, 2009; Quigley & Collins, 1999). Studies
focusing on dyadic interactions have found that people exposed to a
heavy drinking peer (i.e., a confederate) consumed more alcohol than
those exposed to non- and light-drinking peers. During peer
interactions, two individuals' drinking behavior might become
synchronized through peer imitation — the drinking behavior, and
even sipping behavior, of one person may become contingent on the
other's behavior. We suggest that this synchronization is particularly
pronounced with alcohol consumption, and not drinking behavior in
general. Previous studies on imitation of alcohol use have not been
designed to reveal whether imitation is associated with alcohol use
specifically, or is based on the mere exposure to other's drinking
behavior, regardless of content.
Alcohol is important for social identity, bonding and belongingness
in friendships and peer groups (Engels & Knibbe, 2000). A salient
⁎ Corresponding author. Behavioural Science Institute, Radboud University
Nijmegen, PO Box 9104, 6500 HE Nijmegen, The Netherlands. Tel.: +31 24 361 29 55;
fax: +31 24 361 27 76.
E-mail address: H.Larsen@pwo.ru.nl (H. Larsen).
1
Present address: Developmental Psychology, Utrecht University, The Netherlands.
motive for alcohol use is social; drinking is assumed to make parties
more fun, it makes one more relaxed, makes it easier to approach
others, or to share feelings and experiences (Kuntsche, Knibbe, Gmel,
& Engels, 2005). On the other hand, drinking heavily and quickly
might give others the impression that one lacks self-control and has a
problem limiting one's consumption (Suls & Green, 2003). Because
much of alcohol consumption is social in nature, we assume that in a
given drinking context, people generally monitor other people's
drinking patterns. Compared with contexts in which alcohol is
unavailable, when alcohol is being served individuals may be more
aware of the awkwardness of drinking faster than a partner.
Alternatively, drinking alcohol may be regarded as socially rewarding,
leading to increased bonding. These social comparisons based on
other people's drinking behavior may be less likely to occur when
peers are drinking non-alcoholic beverages (which have fewer social
implications). This increased attention might affect differences in
magnitude of imitation of alcoholic beverages as compared to non-
alcoholic beverages.
In order to capture imitation processes, real-time observations of
dyadic interactions were conducted. The study was conducted in a
semi-naturalistic setting; a bar lab. We used a randomized design in a
controlled setting to investigate imitation of sips among same-sex
participant–confederate dyads during a 30-minute interaction. We
expected participants to imitate drinking on a sipping level (Quigley &
Collins, 1999). Moreover, we expected stronger imitation when both
participants and confederates drank alcohol. Understanding imitation
in the context of drinking alcohol might help explain why people
sometimes have substantial difficulties limiting their consumption.

0306-4603/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.addbeh.2009.08.002
50 H. Larsen et al. / Addictive Behaviors 35 (2010) 49–52
2. Method
2.1. Participants and procedure
Seventy women (52%) and 65 men were recruited from the
university campus. Participants were 21 years old, on average (range:
18–28; SD = 2.39). All sessions took place in a bar laboratory at the
Radboud University Nijmegen. Ten undergraduate students aged
18 years and older were employed as confederates. Confederates
were trained to act in a socially neutral way and were instructed to
actively take part in the conversation with the participant. We coded
sips during a 30-minute “break” that occurred between two tasks.
During the 30 min, participants could choose alcoholic (wine or beer)
and non-alcoholic beverages (soda or water). Confederates were pre-
instructed regarding their choice of drink. For a detailed description of
the methodology and ecological validity of the bar lab paradigm, see
Larsen et al. (2009) and Bot, Engels, and Knibbe (2005). Participants
who had consumed alcohol during the observational session were
offered a taxi home. Protocols for the study were approved by the
Ethical Committee of the Faculty of Social Sciences, Radboud
University Nijmegen.
2.2. Measures
Three independent coders scored the same 15 (out of 135, 11%)
sessions to assess reliability. The intraclass correlations for confeder-
ates and participants' number of sips ranged from .73 to 1.00
(confederates' sips) and .89 to 1.00 (participants' sips) indicating a
sufficient to high level of agreement.
We examined the number of confederates' sips and the number of
confederates' sips that were imitated by the participants. Imitation
was operationalized as participants' sip taken within 10 s after a
confederate's sip. Imitation was scored as ‘1’, no imitation was scored
as ‘0’.1 We also examined the proportion of participants' previously
imitated sips before their imitated sip (i.e., number of previously
imitated sips divided by the number of confederates' previous sips),
and the type of drink (alcoholic or non-alcoholic) the confederates
and participants consumed at the moment of imitation. This variable
was called “drinking situation.” There were four different drinking
situations: both the confederate and participant consumed alcoholic
beverages (BA); the confederate drank alcohol and the participant did
not (CA); the participant consumed alcohol and the confederate did
not (PA); both the confederate and participant consumed non-
alcoholic beverages (BN).
2.3. Data analysis
Because we had a different number of repeated measurements for
participants (i.e., sips nested in individuals), a random occasion
Hierarchical Linear Model (HLM) conducted with MLWin software,
1
We chose a 10 second time interval based on a preliminary screening of our
observation data. Fewer seconds appeared to be a relatively narrow time frame for
participants to be able to imitate a sip. On the other hand, using a time interval that
lasted for more than 10 seconds would increase the likelihood that different types of
irrelevant or autonomous behaviors (we mean behaviors fully independent of the
other person's behavior) would be falsely scored as imitation. However, to acquire
more information about how different ‘imitation intervals’ would affect the outcomes
of our study, we performed HLM analyses based on different intervals and compared
the outcomes. Specifically, we employed models with imitation of sips defined within
5 and 15 second time intervals. The patterns (coefficients) of findings were similar to
the one with imitation defined within a 10 second time interval. The likelihood of
imitating sips was lower when only the confederate was drinking alcohol and the
participant not, compared to when they were both consuming alcohol. Thus,
participants imitated confederates’ sips less when they were not consuming alcoholic
beverages at the moment of imitation.
was used for analysis (Goldstein, 1995; Snijders, 1996). The dependent
variable was dichotomous (i.e., imitation versus no imitation). We
examined whether sex, confederates' sips (to control for the number of
measurement occasions), proportion of participants' previously
imitated sips (because we expected that with a higher total number
of confederates' sips and more previously imitated sips, participants'
tendency to imitate sips would overall be significantly higher), and the
four drinking situations were related to the probability of participants
imitating confederates' sips.
In Model 0, sex and the drinking situations at the moment of
imitation were included in the model. The drinking situation variables
were entered as three dummy variables. The drinking situation where
both confederates and participants consumed alcoholic beverages
(BA) was the reference category in all models and participants'
imitation of sips was the dependent variable. In Model 1, we examined
whether imitation of sips was related to the type of beverage
consumed while controlling for the number of confederates' sips and
to what extent participants had imitated previous sips in the 30-
minute interaction. In Model 2, we added the interactions between
drinking situations and sex. In Model 3, interactions between drinking
situations and proportion of participants' previously imitated sips
were added.
3. Results
Participants consumed on average .82 (SD = 1.14, range: 0–4)
alcoholic beverages in the 30-minute interaction. Men consumed
more alcohol (M = 1.11, SD = 1.22) than women (M = .26, SD = .56; t
(112) = 4.57, p < .001). On average, participants consumed 1.29
(SD = 1.05) non-alcoholic beverages. There was no sex difference in
the amount of non-alcoholic drinks consumed. The means and
standard deviations of confederates' and participants' sips, imitated
sips and proportion of imitated sips are displayed in Table 1. There
was a strong positive correlation between the number of confeder-
ates' sips and the number of participants' sips (r = .75; p < .001).
Model 0 demonstrated that the likelihood of participants imitating
a sip was lower when confederates were drinking alcohol and participants
were not (CA), compared to when both confederates and participants
consumed alcoholic beverages (BA, Table 2). The model intercept
indicated that already at the first sip, participants had different likelihoods
of imitating a sip. In general, men imitated sips more than women. We also
ran Model 0 without sex and the findings were similar; the likelihood of
imitating a sip was lower when confederates were drinking alcohol and
participants not (CA), compared to when they were both consuming
alcohol (b=−.73, SE=.24, p<.01). However, the likelihood of imitating
a sip was also lower when both consumed alcoholic as compared to non-
alcoholic beverages (b=−.51, SE=.21, p<.05).
Model 1 showed that while controlling for the confederates' total
number of sips, the likelihood of participants imitating a sip was,
again, lower when confederates were drinking alcohol and partici-
pants were not (CA), as compared to when both confederates and
participants were consuming alcohol (BA, Table 2). The number of
confederates' sips was negatively related to the likelihood of
participants imitating a sip. This indicated that with an increasing
number of sips, the likelihood of imitating a sip diminished.
Participants were most likely to imitate sips at the beginning of the
interactions. Generally, men imitated sips more than women. None of
the interactions in Model 2 and Model 3 attained significance (Table 2).
4. Discussion
The current study is the first to investigate whether imitation of
sips is related to the beverage content. The results showed that men
were more inclined to imitate the sips of a same-sex partner than
women. By using the proportion of imitated sips, we corrected for
confederates' number of sips and participants' previously imitated
 H. Larsen et al. / Addictive Behaviors 35 (2010) 49–52 51

Table 1
Frequencies of sips in the 30-minute observational period (total sample estimates).

Total Women Men t

n M (SD) n M (SD) n M (SD)

Sips of confederate 2316 12.18 (8.21) 1013 11.57 (7.95) 1303 12.65 (8.38) 3.13**
Sips of participant 2078 10.96 (7.10) 903 8.95 (5.94) 1175 12.51 (8.98) 10.31***
Participants' imitated sips 427 .18 (.39) 155 .15 (.36) 272 .21 (.41) 3.44**
Proportion imitated sips – .24 (.21) – .21 (.21) – .26 (.21) 6.03***
⁎⁎⁎p < .001; ⁎⁎p < .01.

sips. Thus, results are not a function of men's tendency to drink more
than women. Our findings also demonstrated that participants
imitated confederates' sips more when they were both consuming
alcohol compared to when confederates were drinking alcoholic
beverages and participants non-alcoholic beverages. This reveals that
imitation of sipping in an ad lib context might be related to the type of
drink consumed. Indeed, participants imitated sips approximately 7%
more when both consumed alcohol compared to when only
confederates consumed alcohol.
The present experimental findings indicate that people are
compelled to drink alcohol when those around them are also
drinking alcohol. Imitation may be stronger when alcohol is being
consumed because of the social meaning people attribute to alcohol
consumption. People may (non-consciously) monitor others' and
their own drinking behavior in order to keep up a similar drinking
pace and imitate sips because they are “keeping an eye” on other's
drinking behavior. This type of monitoring might occur because
people do not want to seem like they are drinking too much or,
alternatively, because they want to signal that they are having a
good time together.
The sex difference in imitation of sips was not related to the type of
beverage consumed. Even though previous studies have only
examined imitation of alcohol consumption in terms of choice of
drink, they did not find sex differences in imitation of alcohol
consumption in terms of level of use (Caudill & Kong, 2001; Corcoran,
1995; Larsen et al., 2009; Lied & Marlatt, 1979). Our findings indicate
that men imitate sips more in general, but that this is not specifically
related to their alcohol consumption. The meaning of this is not clear
to us and future studies are necessary to explain this.
The participants' proportion of previously imitated sips was not
associated with their subsequent likelihood of imitating a current sip.
So, there did not seem to be a cumulative effect of imitation of sipping.
The findings showed that participants' likelihood of imitating
confederates' sip was highest in the beginning of the 30-minute
interactions, which may be explained by the possibility that especially
in the beginning of interactions with strangers, people want to feel
connected and appear friendly. Another explanation is related to the
disinhibition effect of alcohol (Fillmore, Marczinski, & Bowman, 2005;
Lyvers, 2000). Perhaps when people drink more they lose their self-
control, get more in an automatic mode and follow their normal
pattern of drinking rather than the pace set by other people.
A limitation of this study is that the number of confederates' sips
was not controlled during the interactions. That is, we did not instruct
confederates beforehand with regard to how many sips they had to
take with what pace. Thus, although a strength of the present design is
that it provided a highly natural, and thus generalizable, ad lib
drinking situation, we cannot rule out the possibility that confeder-
ates also imitated participants' drinking behaviors. Future research
might aim to examine the synchronization of drinking behaviors in
more naturalistic dyads with no confederate. Also, in this study
imitation of sips was examined in same-sex dyads only. Imitation
processes might be different in opposite-sex dyads; it is possible, for
example, that men would imitate alcohol sipping of women
throughout an interaction, whereas women would stop imitating
alcohol sipping of men after a while.
Role of funding sources
This research was supported by a grant from The Netherlands
Organisation for Scientific Research (NWO) (# 400-05-086). Geertjan
Overbeek was supported by a fellowship from the Netherlands
Organization for Scientific Research (# 451-05-015).

Table 2
Multilevel analyses on participants' imitation of sipping by proportion imitated sips, sex, and drinking situations.

Model 0 Model 1 Model 2 Model 3

Estimate (SE) Estimate (SE) Estimate (SE) Estimate (SE)

Fixed effects
Intercept − 1.545 (.160)⁎⁎⁎ − 1.175(.197)⁎⁎⁎ − 1.168 (.208)⁎⁎⁎ − 1.209 (.226)⁎⁎⁎
Sips of confederate − .043 (.009)⁎⁎⁎ − .043 (.009)⁎⁎⁎ − .041 (.009)⁎⁎⁎
Sex − .440 (.188)⁎ − .465 (.180)⁎⁎⁎ − .490 (.387) − .471 (.383)
Proportion imitation of participant − .300 (.369) − .290 (.371) − .382 (.549)
Confederate drinking alcohol (participant not) − .576 (.243)⁎ − .569 (.237)⁎ − .601 (.308) − .318 (.350)
Participant drinking alcohol (confederate not) − .361 (.259) − .238 (.252) − .207 (.274) − .088 (.378)
Both drinking non-alcohol − .372 (.213) − .270 (.206) − .289 (.246) − .382 (.284)
Both drinking alcohol (reference category) – – – –
Proportion imitation Confederate drinking − 2.655 (1.720)
Proportion imitation Participants drinking − .398 (1.097)
Proportion imitation Both non-alcohol .620 (.815)
Sex Confederate drinking .072 (.514) − .037 (.510)
Sex Participants drinking .224 (.701) − .316 (.714)
Sex Both non-alcohol .054 (.467) .034 (.463)
Random effects
Variance of intercept .313 (.107)⁎⁎⁎ .236 (.097)⁎⁎ .234 (.097)⁎⁎ .215 (.094)⁎⁎

Note. ⁎⁎⁎p < .001; ⁎⁎p < .01; ⁎p < .05. ‘Sips of confederate’ is the time indicator. ‘Confederate drinking’ = participant was not drinking at the moment of imitation. ‘Participants
drinking’ = confederate was not drinking at the moment of imitation. ‘Both drinking soda’ = both confederate and participant consumed soda at the moment of imitation. Reference
category = both confederate and participant consumed alcohol at the moment of imitation.
52 H. Larsen et al. / Addictive Behaviors 35 (2010) 49–52
Contributors
Helle Larsen, Rutger Engels, Isabel Granic and Geertjan Overbeek
designed the study. Helle Larsen wrote the protocol and conducted
literature searches. Pierre Souren and Helle Larsen conducted the
statistical analyses. Helle Larsen wrote the first draft of the manuscript
and all authors contributed to and have approved the final
manuscript.
Conflict of interest
There are no conflicts of interest.
References
Bot, S. M., Engels, R. C. M. E., & Knibbe, R. A. (2005). The effects of alcohol expectancies
on drinking behaviour in peer groups: Observations in a naturalistic setting. Ad-
diction, 100, 1270−1279.
Caudill, B. D., & Kong, F. H. (2001). Social approval and facilitation in predicting
modeling effects in alcohol consumption. Journal of Substance Abuse, 13, 425−441.
Collins, R. L., Parks, G. A., & Marlatt, G. A. (1985). Social determinants of alcohol
consumption: The effects of social interaction and model status on the self-
administration of alcohol. Journal of Consulting and Clinical Psychology, 53, 189−200.
Corcoran, K. J. (1995). Cognitive and situational factors predict alcoholic beverage
selection. Addictive Behaviors, 20, 525−542.
Engels, R. C. M. E., & Knibbe, R. A. (2000). Alcohol use and intimate relationships: When
love comes to town. Addictive Behaviors, 25, 435−439.
Fillmore, M. T., Marczinski, C. A., & Bowman, A. M. (2005). Acute tolerance to alcohol
effects on inhibitory and activational mechanisms of behavioral control. Journal of
Studies on Alcohol, 66, 663−672.
Goldstein, H. (1995). Multilevel statistical models, 2nd ed London: Arnold.
Kuntsche, E., Knibbe, R., Gmel, G., & Engels, R. C. M. E. (2005). Why do young people
drink? A review of drinking motives. Clinical Psychology Review, 25, 841−861.
Larsen, H., Engels, R.C.M.E., Granic, I., & Overbeek, G., (2009). An experimental study on
imitation of alcohol consumption in same-sex dyads. Alcohol and Alcoholism, 44,
250–255.
Lied, E. R., & Marlatt, G. A. (1979). Modeling as a determinant of alcohol consumption:
Effect of subject sex and prior drinking history. Addictive Behaviors, 4, 47−54.
Lyvers, M. (2000). “Loss of control” in alcoholism and drug addiction: A neuroscientific
interpretation. Experimental and Clinical Psychopharmacology, 8, 225−249.
Quigley, B. M., & Collins, R. L. (1999). The modeling of alcohol consumption: A meta-
analytic review. Journal of Studies on Alcohol, 60, 90−98.
Snijders, T. (1996). Analysis of longitudinal data using the hierarchical linear model.
Quality and Quantity, 30, 405−426.
Suls, J., & Green, P. (2003). Pluralistic ignorance and college student perceptions of
gender-specific alcohol norms. Health Psychology, 22, 479−486.