You are on page 1of 16

Durkheim's Theory of Social Order and Deviance: A Multi-Level Test

Author(s): Thorolfur Thorlindsson and Jn Gunnar Bernburg


Source: European Sociological Review, Vol. 20, No. 4 (Sep., 2004), pp. 271-285
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/3559561
Accessed: 17-02-2016 14:14 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access to European Sociological Review.

http://www.jstor.org

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

2004 271-285
European Sociological Review VOLUME20 | NUMBER4 | SEPTEMBER
DOI:10.1093/esr/jch025,availableonline at www.esr.oupjournals.org

Durkheim's
and

Theory

Deviance:

of

Social

Multi-level

271

Order
Test

ThorolfurThorlindssonandJ6nGunnarBernburg
Inthe present paper, we study the social context of adolescent delinquency in order to
examine Durkheim'stheory of social order. We use survey data on urban adolescents in
Icelandthat allow us to examine key theoretical constructs on both the community and
the individual levels of analysis. Our findings lend substantial support to the multi-level
nature of Durkheim'sconcept of social integration. Using hierarchicallinear regression,
we show that both community and individual level social integration indicators have
negative effects on adolescent delinquency, and that the experience of anomie mediates
a substantial part of these effects. Furthermore,the multi-level context of social
integration and anomie moderates the effect of imitation (peer delinquency) on
delinquent behaviour. The findings highlight important sociological themes that have
been neglected historicallyin scholarlytraditions that have elaborated upon and
extended Durkheim'stheory of social order.

Introduction
In his works on morality and suicide, Durkheim
advancesa generaltheory of social order and deviance,
emphasizing the social against the individual level.
Durkheim's (1895/1982) well-known discussion of
social facts emphasizes the idea that collective life of
groups and social institutions cannot be reduced to the
psychology of individuals. Rather, social groups have
structuralpropertieswhere norms, values and ongoing
social relationshipsplace constraints on the individual.
Social order is rooted in the emergent nature of social
life where the whole is greaterthan the sum of its parts.
Sawyer(2002: 228) has pointed out that, althoughmany
sociologists have acknowledgedDurkheim as a theorist
of emergence, they have often failed to understandthe
centralimportanceof this concept in his work or simply
dismissed it as ambiguous and confusing. In Sawyer's
view, the failure to acknowledgethe importance of the
concept of emergence has been rooted in the broader

bias in American sociology toward individualist sociological theorizing. A leading European sociologist,
Anthony Giddens (1972), has emphasizedthis structural
and emergent character of the social in Durkheim's
work. He points out that Durkheim's objective was to
develop a new theory that transcendedboth utilitarianism and idealism without relapsing into materialism.
Durkheim'sanswerwas to stress the emergent qualities
of the 'social' and external and obligatory characterof
social facts.
Durkheim'stheory of social order is a structuraltheory, but not simply a theory about how individuals
internalize social norms. As Sawyer (2002: 239) has
noted, however, Durkheim has often been misinterpreted to 'locate the source of order in individualinternalization of social facts, following Parson'svoluntarist
readingand the generalindividualismof Americansociology'. In Durkheim'swork 'emergenceprocesses were
centralnot only to the origin of social structure,but also
to its continuedmaintenanceand reproduction'(Sawyer,

EuropeanSociologicalReviewvol. 20 no. 4 ? OxfordUniversityPress2004; all rightsreserved

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

272 | THORLINDSSONAND BERNBURG

2002: 239). Throughouthis work, Durkheim uses many


different concepts, such as collective representations,
social currentsand sui generisto describethe societal or
the group level. However,he never explicitlyworked out
a theory of social interaction, bridging the individual
and societal levels. Instead, he worked out a theory of
emergence that started out from the group level.
Durkheim thinks of social emergence as a downward
influencefrom the societalto the individuallevel (Sawyer,
2002) through specificprocesses.
Throughouthis works,Durkheimdescribesthese processes in different ways. In Suicide (1897/1951) and
MoralEducation(1925/1961) he portrayssocial order in
terms of the now well-known concepts of social integration and regulation.The term social integrationstresses
the social links attaching individuals to social groups
and emphasizesthe importanceof density and duration
of social interaction in the group (see Gibbs, 2000).
Durkheim implies that social integrationmay influence
conformity through interpersonalrelationshipsas well
as through the emergentcharacteristicsof the integrated
social group. Within criminology, contemporarysocial
bonding/control theories (Hirschi, 1967; Sampson and
Laub, 1990) have built upon the formerconcept of integration, focusing on individual-levelprocesses.According to these elaborations, strong and durable social
relationshipsproduce sensitivity to the expectations of
others, and commitment to institutionalized careers
(e.g. education) produces higher stakes in conformity.
The latter process has received much less attention in
devianceresearch.Durkheimimplies that social integration also connects individuals to society by ensuring a
high degreeof attachmentto commonly held values and
beliefs, regardlessof interpersonalrelationships.Social
integrationpromotes the perceptionthat people are part
of a largersocial whole, focusing their interests outside
the individual self and counteracting egoism. Hence,
social organizationwhere social integrationis weak may
foster excessive individualism and weakening of sentiments for collective norms and rules (Durkheim, 1897/
1951:209-216).
Durkheim discusses the term social regulation as the
degree to which society places limits on individual's
desiresand aspirationsthrough normativeor emotional
definitions. Sociologists have discussed regulation
almost exclusivelyin terms of anomie, which describes
the situation when this regulativeprocess fails and the
individual is left without society placing normative
limits on his or her goals and courses of action. While
Durkheim does not develop the connection between
social integration and regulation to any extent, several

scholars have examined such linkages (Johnson, 1965;


Giddens, 1971). Recently, scholars have argued that
social regulationmediates the influence of interpersonal
social integrationon deviant behaviour (Agnew, 1997),
includingsuicidality(Thorlindssonand Bjarnason,1998).
This work indicatesthat interpersonalsocial integration
produces conformity by providing the individual with
sentiments for the social norms and valued goals. Social
regulationmay also mediatethe influence of group-level
social integration on conformity and deviance.
Durkheim's (1897/1951, 1893/1984) discussion of the
role of religion and nationalism in focusing people's
attention on common ideals, and thus strengthening
collective sentiments, implies such a process. Insofar as
integrationof the social group promotes the perception
that people are a part of the group, group level integration should be expected to strengthen the individual's
sentiments for social norms (i.e. reduce anomie),
regardlessof interpersonalsocial integration.
A neglectedtheme in Durkheim'sclassicSuicideis his
discussion of imitation. For Durkheim, imitation is a
social process spreading unstable and transitory social
facts, or social currents.Imitationrefersto the transmittal of deviant behaviours through social contacts,
including the transmittal of attitudes and definitions
favourableto deviance and modeling people of higher
social standing (Tarde, 1903). The theme of imitation is
of vital importance for analysingsocial deviance, and it
is a vital element in Durkheim'sgeneraltheory of social
order, even if Durkheim himself rejectedit on empirical
grounds as an explanation of suicide.' That Durkheim
drops this theme in his study of suicide would seem to
be inconsistentwith the overwhelmingevidence of contemporaryresearchthat suggeststhat deviantbehaviour,
and adolescent delinquency in particular, spreads
through direct social contact (Jensen, 1972;Akers et al.,
1979;Elliot et al., 1985;Tittle et al., 1986;Matsuedaand
Heimer, 1987; Kandel and Davies, 1991; Warr and
Stafford,1991;Warr, 1993;Bernburgand Thorlindsson,
2001; Haynie, 2001). But while Durkheim (1897/1951:
138-142) concluded that imitation has little independent effect on suicide rates,he did recognizethat imitation could be a powerful force influencing social
behaviour when there is a predisposition caused by
other social factors. He implies that social integration
and regulation (anomie) should moderate the effects of
such unstablesocial currents(pp. 123-154).
For Durkheim,then, the level of social integrationin
the group influences individualconformitywithin it. It
is an open empiricalquestion, however,whether group
level integration influences conformity mainly through

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

DURKHEIM,SOCIALORDERAND DEVIANCE| 273

interpersonalintegrationand socialization(as in contemporarysocialbondingtheory),or whethergrouplevelintegration influences conformity independent of such


interpersonalfactors. As Thorlindsson and Bjamason
(1998:100) havepointedout, the relationbetweensocietal
integrationand interpersonalintegrationis not deterministic. 'Particularindividuals may become isolated in a
tightly integratedsociety, or they may enjoy a close-knit
personalnetworkin the most disintegratedsociety.'In this
light,we mightframean empiricalquestion:Does a tightly
integratedcommunity increase the individual'sconformity regardlessof personalintegration?Conversely,does a
disintegratedcommunitydecreasethe individual'sconformity regardlessof personalintegration?
The presentpaperattemptsto throw empiricallight on
thesebasic issuesin Durkheim'sworkby studyingadolescent delinquencyas a case of deviantbehaviour.We use a
statisticaltechnique (hierarchicallinear modeling) that
allowsus to addressthe multi-levelnatureof Durkheim's
theory. We examine to what extent conformity (social
order) has its roots not only in personalsocial relationships, as impliedby contemporarysocial control theories
of deviance (Hirschi, 1967), but more generallyin the
level of social integration in the community. We also
examine Durkheim'sproposition that social integration
and anomie contextualize (moderate) social imitation.
We thus use and extend Durkheim'sneglecteddiscussion
of imitation to place the social processesof devianceand
conformityin a generalsociologicalperspective.
It is criticalto measureandanalysekeyvariableson both
the individuallevel and the communitylevel.We use survey dataof Icelandicadolescentsthatoffersan opportunity
to measure and analyse these indicatorson both levels
using multi-levelstatisticaltechniques.Our data includes
whole cohortsof studentsratherthan samples.Therefore,
we can aggregatesurvey responses to the school level
where school characteristicsinclude most students in a
cohort, providinghighly reliablemeasuresof the level of
socialintegrationand anomiein the schoolcommunity.

Social Integration, Anomie and


Imitation
In workssuch as Suicideand MoralEducation,Durkheim
offers a sociological theory of order and deviance. He
does so by defining and discussing the relationships
among three classes of social processes, that is, social
integration, regulation/anomie and, to a lesser extent,
imitation. In the following, we clarifyhow we use these
concepts in studyingadolescentdelinquency.

Social Integration
In Suicide Durkheim distinguishesbetween three types
of integration: religious, familial and political integration.
Religious Integration
Durkheim (1897/1951) emphasizes that religion provides a social context where people interact and form
strong emotional, psychological, and social bonds.
Although Durkheim did not discuss adolescents in this
context, it is clearthat his formulationof religious integration applies as much to them as other members of
society. Below, we directly test the process of religious
integration,focusing on religiousactivity.
FamilyIntegration
While Durkheim's empirical analysis focused on marriage, his theory focuses on density of the family group.
Density derives from the quantity and the intensity of
the relationshipsthat attaches family members to each
other and to common goals, and strengthensthe collective sentiments.Moreover,the densityof the familysociety depends upon the unity and the active participation
of individual members in family life. A well-integrated
family has social control mechanismsconstrainingindividuals and intervening if they break the rules of the
group (1897/1951: 209). Clearly,Durkheim'stheory of
family integrationappliesto all family members,including adolescents, while the theory can have different
meaning for variousfamily members (Thorlindssonand
Bjarnason,1998).
PoliticalIntegration
Political integration is a theme often neglected by
Durkheimianscholars,and it is not as developedas religious and family integration.Durkheimexpects that the
same general principles of participation in organized
social life and commitment to common social and political institutions fostersstrongerbonds between the individual and society. Political crisis or upheavalsincrease
the intensity of 'collective sentiments and stimulate
patriotism' (1897/1951: 219). It forces people to recognize common interests and goals, and value political
institutions. Hence, the ties between the individual and
the group are stronger.
Other Formsof Integration
Durkheim's discussion of social integration is not
exhaustive. Durkheim offers a general theory of social
integration that can be extended to areas of social life

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

274 | THORLINDSSONAND BERNBURG

other than those that he discusses. Since Durkheim did


not discuss integration with adolescents especially in
mind, he did not deal with problems of integration in
areasthat are characteristicfor adolescents.Clearly,the
school is a centralinstitution in the life of contemporary
adolescents. The school provides sentiments of obligation and commitment and a set of common goals.
School is a source of organized group activity of high
density providing a variety of rituals, ideas, and sentiments that have controlling and supporting functions
for adolescentsand directsthem towardssociallyaccepted
goals. Commitment to academic achievement and
school performanceare indicatorsof school integration.

Regulation/Anomie
While Durkheim discusses anomie in several places
throughout his works, his treatment of the concept is
neithercumulativenor consistent.Anomie is an important theme in Suicide, portraying a special kind of
relationshipbetween the individualand society, emphasizing regulation as a central aspect of social order.
Anomie is characterizedby unrealistic aspirations and
expectations that often cause conflict between means
and ends and leave individualswith feelingsof meaninglessness and hopelessness and a sense of injustice. 'The
limits are unknown betweenthe possible and the impossible, what is just and what is unjust, legitimate claims
and hopes and those which are immoderate.' (1897/
1951:253). For a social order to be just it has to be fairly
stable and consistent, it has to be characterizedby equal
opportunityand a predictablefuture (p. 271).
Anomie has severalhighly relateddimensions. At the
core of the anomie concept lies the generalidea that the
absence of clear rules of behaviour and ambiguity in
rules and goals create a state where the individual faces
uncertainty, conflicting expectations, and ambiguous
norms and values. This general dimension of anomie
may be especially pertinent to adolescence, which is a
period in life where the individual is struggling with
problemsof meaning and purpose.
Merton's (1967/1994) well known theory of anomie
captures another, more specific dimension of anomie,
where people are denied institutionalizedmeans to reach
culturallydefined goals. As a result, they must abandon
the means, the goals, or both. Merton's interpretation
of anomie holds that deviance is a social structural
problem,an adaptationto a discrepancybetween culturally defined goals and the socially accepted means of
achieving these goals. The school holds a central position in this context in Westerncountries. Schooling and

school credentialsprovide the legitimate means to economic and social success for Icelandicyouth. Therefore,
school can be an importantsource of anomie for adolescents. Experiencing schoolwork as meaningless is an
indicator of anomie among adolescents in contemporary society. The same appliesto adolescentsthat experience conflict within the school institution and want to
quit school and bypassthe means that it offers to obtain
culturallydesirablegoals.

Peer Group Influences and the


Role of Imitation
Among the themes discussed by Durkheim in Suicide,
which have not receivedmuch attention,is his treatment
of imitation(Durkheim,1897/1951:123-154;Thorlindsson
and Bjarnason,1998). In order to clarifythe concept of
imitation Durkheimdistinguishedbetween three definitions (1897/1951: 123-130). First, imitation describes
the collective action of crowds toward a common goal.
Second, imitation describes the process of modeling a
person or a social group with high prestige or is in a
position to apply social pressure. Finally, imitation,
according to Durkheim, refers to imitation that has no
social aspects involved but works like a 'simple contagion', which Durkheimholds to be more a psychological
than a social phenomenon.
Durkheim dismisses the first definition completely,
arguingthat differentprinciplesgovern collectivebehaviour. He arguesthat the second definition involves various social processes, such as social authority, social
prestige, and social bonds. We may label this type of
imitation as social imitation, putting words into
Durkheim's mouth. In principle, Durkheim acknowledges the importance of imitation based on modeling
and differential association. He argues that imitation
that stems from social bonding or modeling a person
held in high prestige is a social process, that is, social
imitation, not a mindless or 'instinctual'imitation.2
Durkheim proceeds with his analysis of the role of
imitation of suicide using the narrow definition of imitation that emphasizes simple contagion from individual to individual. He recognizesthat imitation, even in
this narrowestsense, can be an important force affecting social behaviour,even suicide (1897/1951: 138-142).
However, since he found no conclusive empirical
evidence relating imitation to general suicide rates, he
felt that imitation had little influence on social behaviour unless there was a predisposition caused by other
social factorsto behave in that particularway.

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

DURKHEIM,SOCIALORDERAND DEVIANCE| 275

Durkheimsuggestedthat integrationcould reduce the


effects of harmful social currentscharacterizingdeviant
groups. However, he never developed this intuitively
promising idea. Since then, sociologists and criminologists have developed the theoretical connections
between anomie and integration on the one hand and
social imitation involving deviant groups in several
ways. Thus, Matza (1964: 49) has argued that when
social control is weakened, adolescents drift towards
delinquent subculturesthat lie between convention and
crime. Brake (1985: 48-53) traced the influence of
Durkheim's anomie theory on subculture studies,
emphasizing the contribution of Robert Merton and
Albert K. Cohen. A carefulreadingof Suicide,also, suggests that imitation could play an importantrole.
Even if it is debatable what Durkheim (and Tarde)
meant precisely by imitation, the resemblanceto contemporarydifferentialassociation(Sutherlandand Cressey,
1984) and social learningtheories (Akers, 1977) is obvious. Some scholars have recognized the similarity
between Durkheim's discussion of imitation, and in
some cases social currents, to theories of differential
association and social learning. Thus, they have
extended Durkheim's theory of suicide to include suggestion in the form of differential association (Stack,
1987, 2003; Stack and Wasserman, 1995; Thorlindsson
and Bjarnason,1998).
Contemporary criminological theories emphasizing
the importanceof associationwith deviantgroupscan help
us make furthersense of Durkheim'sdiscussion of imitation and link it with a prominentand long-standingtradition in sociologicalcriminology(Akers,1977;Sutherland
and Cressey,1984). The basic proposition of these theories is that associating with delinquent peers breeds
delinquency.Researchhas consistentlyproven this relationship to be among the strongestand most consistently
reported in the research on adolescent delinquency
(Jensen, 1972; Akers et al., 1979; Tittle et al., 1986;
Matsueda and Heimer, 1987) and substance use (Elliot
et al., 1985). However, these theories postulate different
mechanismsthroughwhich deviancetransmittaloccurs.
In differential association theory, transmittal occurs
through coming into contact with definitions that are
favourable toward deviance. Social learning theory
argues more generallythat transmittal occurs through
various processes, including modeling, reinforcement,
and contact with definitions. Warr and Stafford (1991)
have suggested that all these mechanisms are involved,
providingevidence indicatingthat the peer group influences behaviourthrough transferenceof attitudes (definitions) as well as modeling and reinforcement. The

concept of social imitation includes all the complex processes that spread delinquency through peer contact.
Below we do not attempt to untangle these processes
empirically.Rather,our goal is to examine whether the
social context of social integration and anomie moderates the transmittalof delinquencythrough peer group
contacts.

Durkheim's Theory of Social


Order Applied to Delinquency
Applying Durkheim's theory of social order and deviance to delinquency allows us to examine important
concepts using contemporary social science methods.
Even though Durkheim's explicit intention in works
such as Suicidewas to develop a macro-sociologicaltheory based on aggregatedata, the result is a highly complex multi-level theory of society that moves between
different levels of analysis, skipping levels or mixing
them unsystematically
(Lukes,1972:213-22;Thorlindsson,
1983; Thorlindsson and Bjarnason,1998). It is important to note in this context that the relationships
between levels of analysesin Durkheim'stheory of suicide are not fixed accordingto some given theory,but to
a considerableextent, empiricallydetermined.The limitation of his data and the data analysestechniquesavailable to him at the time often hinders him from
elaboratingon the relationshipbetween differentlevels.
The micro level in Durkheim's theory is nevertheless
critical in his theory (Thorlindsson and Bjarnason,
1998). Testing Durkheim's theory on both micro and
community levels simultaneouslyis an important contributionto Durkheimianscholarship.
Building on Durkheim's multi-level theory of social
integration and anomie has important implications for
empirical analysis.The theory implies specific patterns
of associations among integration, anomie, imitation,
and deviance on the individualand the community levels of analysis.First, adolescents that have low levels of
social integration should be more likely to engage in
deviant behaviour. While this hypothesis is congruent
with prior researchon social bonding theory (Hirschi,
1967), Durkheim's theory of integration also implies
macro-levelprocesses,a point that social bonding theorists have ignored. A highly integratedcommunity may
foster attachmentto the community and to social values
and norms and increases informal social control (e.g.
see Krohn, 1986), regardlessof personal ties and social
participation. Hence, community-level integration
should negativelyinfluence delinquent behaviourwhile

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

276 | THORLINDSSONAND BERNBURG

controlling for individuallevel integration.We test this


hypothesisby aggregatingthe individuals'level of social
integrationto the community level.
Adolescents experiencing anomie, that is, rule ambiguity and feelings of normlessness and meaningless,
should be more likely to engage in delinquency.Societal
norms, rules, and goals have less control over their
behaviour. Similarly, adolescents who experience
school failure or school work as meaningless should be
likely to drift into delinquency. Low levels of integration should undermine the authority of rules and
norms (Durkheim, 1897/1951: 287-289). Individuals
with low levels of social integration are more likely to
experience feelings of normlessness and meaninglessness (Agnew, 1997;Thorlindssonand Bjarnason,1998).
This argument implies that anomic experiences should
mediate part of the effect of social integration on deviant behaviour.
Our discussion of imitation suggeststhat there should
be a strong, positive relationship between delinquent
behaviour and delinquency among friends. This relationship has been documented consistently in delinquency research(Jensen, 1972;Akers et al., 1979; Elliot
et al., 1985; Tittle et al., 1986; Matsueda and Heimer,
1987;Kandeland Davies, 1991;Warrand Stafford,1991;
Warr, 1993;Bernburgand Thorlindsson,2001; Haynie,
2001). Furthermore,adolescentswho are integratedand
who experiencelow levels of anomie should be less likely
to associate with delinquent peers. Adolescents rarely
reject the dominant culture, rather,they adapt to structurallyinduced situationsby participatingin groupsthat
may collectively rationalize deviance (Kornhauser,
1978). Prior longitudinal research shows that adolescents with strong social ties to conventional institutions
are less likely to associate with delinquent friends
(Thornberry, 1989; Sampson and Laub, 1993). This
researchsuggeststhat (social) imitation should mediate
a substantialpart of social integration and anomie on
delinquency.Our analysisbelow follows this logic.
However, Durkheim's theory implies that imitation
becomes a more powerfulsocial force when the forces of
social solidarity become weaker. This proposition has
micro- and macro-level implications. First, adolescents
who have weak ties to social institutions or experience
anomie should be more vulnerable to imitation processes. The level of personal social integration and
anomie should moderateor contextualizethe effect that
associating with delinquent peers has on self-reported
delinquency. This prediction has some grounding in
prior research.Thus, Warr (1993) has shown that peer
delinquency has less effect on delinquent behaviour as

adolescentsspend more time with their parents (see also


Elliotet al., 1985).
Second, when community integration is weak, and
when the community loses the ability to provide youth
with sentimentsto commonly sharednorms of conduct,
young people should be more vulnerable to imitation
processes, net of personal integration and personal
anomie. When communities have low levels of social
integration or high levels of anomie, imitation should
become a more powerful force influencing their behaviour. Empirically, this implies cross-level statistical
interaction effects; community-level indicators of integration should moderatethe effect of deviant peer associations on individual'sdeviance,while holding constant
interpersonalintegration.We are not awareof any studies that have tested such cross-levelinteractions.

Method
The data come from a national surveyof Icelandicadolescents conducted in March 1997 (Thorlindssonet al,
1998). The originalsample consisted of all studentsborn
in 1980 and 1981 (14-16 years old), attendingthe compulsoryninth and tenth gradeof the Icelandicsecondary
school. Anonymous questionnaires were administered
to all students that were present in class on one day in
March 1997. Teachers and researchassistantsadministered questionnaires in sealed envelopes (for methodological considerations, see Bjarnason, 1995). The
students who were not present in school on the day of
the surveywere not included.
Iceland only has one urban area located in the southwestern part of the country, that is, Reykjavikand surrounding towns, an area with a population of 175,000.
Almost all childrenand adolescentsin Icelandattendthe
public schools operated by the county government
(sveitarfelag). In the urban area, students are selected
into the public schools based on residencein the neighbourhood in which the school is located. Therefore,the
students in the school, almost without exception,belong
to the neighbourhood community. All schools in Iceland are public schools, which use the same official curriculumand arefundedin the sameway.Manyimportant
social activities, such as parents' associations and sport
and leisure,are organizedin the communities definedby
the school. Hence, the school and the neighbouring
community provide a meaningfullevel of analysisfrom
a sociological point of view. Below, we obtain measures
of community characteristicsby aggregatingstudents'
responsesto the school level.

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

DURKHEIM,SOCIALORDERAND DEVIANCEI 277

Iceland is a nation in Northern Europe with just


under 300,000 inhabitants. It is a very homogeneous
society by Western standards.About 90 percent of the
population belongs to the Lutheran National Church.
The country has low rates of violent crime, and scholars
have describedit as relativelyhomogeneous with regard
to various dimensions of social stratification,culture,
and ethnicity (see Gunnlaugssonand Galliher,2000).
Thirty-fourpublic schools for ninth and tenth graders
are locatedwithin the urbanarea.Valid
(gagnfrwdiskolar)
questionnaireswere obtained from 4213 students, or 93
percent of the two cohorts in these schools. Forty-eight
percentof the respondentswere female.Afterdeletion of
casesthat had missingvalueson the variablesused in this
study,the numberof studentsremainingin the analysisis
3512. Thus, the present analysisuses 78 percent of the
populationof studentsin the two cohorts.The basiccharacteristicsof the final sampleare not differentfrom those
of the total sample,for example,the male-femaleratio is
the same, as well as the level of delinquency.The largest
school has valid data on 288 students,while the smallest
school has data on 40 students.Twelveschools havevalid
data on more than 100 students while 21 schools have
valid dataon 60-100 students.
Measures3
FamilyIntegration
Statistical descriptions of all variables are shown in
Table 1. A dichotomous variable is coded '1' if the
ardiansand '0' if
subject is living with two parents/gu<
otherwise. The following questions -were used to tap
interactiondensity within the family. IRespondentswere
Table 1 Descriptive
statisticsfor the individuallevel
variables(n= 3512)
Variables

Mean

S.D.

Gender
Familydensity
Two-parenthome
Religiousintegration
Politicalintegration
School integration
Grades
Normlessness
School anomie
Peer delinquency
Delinquentbehaviour
Number of school communities

0.50 0.50
0.65
0.85
0.58
8.68
6.34
20.60
18.43
6.11
4.74
7.97 7.97

0.50
7.22
0.35
0.49
3.06
1.99
5.99
4.90
2.64
2.14
3.67
3.67

asked how frequentlythey spent time with their parents


on weekends and on working days (0=almost never,
1 = seldom, 2 = sometimes, 3 = often, 4 = almost always).
Another set of questions asked respondents how frequently they joined their family in various leisure activities: watched TV or VCR, did sports or outdoor
activities,went to shows and events, visited relativesor
friends, played cards, and talked (0= almost never,
1 = seldom, 2 = sometimes, 3 = often). The answer-scores
to these questionswere convertedinto z-scores and then
summed into a scale (Cronbach'sAlpha= 0.76).
Religious Integration
Participation in religious activity was measured on a
four-item summary scale (Cronbach's Alpha=0.70).
Respondentswere askedhow often they went to church,
how often they participatedin other kinds of organized
religiousactivities,how often they say their prayers,and
how often they read the Bible or other religious texts
(0 = almost never, 1 = less than once a month, 2 = one to
three times a month, 3=once a week, 4=more than
once a week).
PoliticalIntegration
Adolescents do not participate in political life in the
same way as adults. Political and governmentalinstitutions are usually not as central in their life as religion
and the family. However, Durkheim'sargumentimplies
that trust in politicaland governmentalinstitutionsis an
indicator of adolescent political integration. Political
integration was measured with a four-item summary
scale (Cronbach's Alpha=0.85). Respondents were
asked how much they trusted political institutions: the
courts, the police, the parliament, and the government
(O= very little trust, 1 = rather little trust, 2 = rather a
lot of trust, 3= very much trust). Respondents were
also asked about participation in political activities,
but the answers were not used in the analysis since
only a handful of them reported any political participation at all.
School Integration
School integration measures commitment to school by
combining the scores from two questions (Pearson's
correlation=0.67). Respondents were asked how well
they felt that the following statementsappliedto them: 'I
do not feel that I make enough effort in my study', and
'I feel that I am badly preparedfor classes' (0=almost
always applies, 1= often applies, 2= sometimes applies,
very poorly).
3 = applies
appliesratherpoorly, 4 = appliesvery

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

278 | THORLINDSSONAND BERNBURG

The second measureof school integrationis academic


achievement. Respondents were asked about their
grades in four main subjects, mathematics, Icelandic,
English and Danish (coded 1-7). The grades were
summed into a scale (Cronbach'sAlpha= 0.86).
Anomie
Two scales were used to measure the general state of
anomie and the specific case of school anomie. First,
anomie was measured on a six-item summary scale
adapted from Dean (1961). Respondents were asked
what they felt about the following statements:'You can
break most rules if they don't seem to apply', 'I follow
whateverrules I want to follow', 'In fact, there are very
few absolute rules in life', 'It's hard to trust anything,
because everything is so variable', 'In fact, nobody
knows what is expected of her or him', 'One can never
be certain of anything in this life'. The response categories where 0=very much disagree, 1= somewhat disagree, 2= don't know, 3= somewhat agree, 4 = very
much agree(Cronbach'sAlpha= 0.77).
Second, a more specificcase of school anomie emphasizing the meaninglessness dimension of anomie was
measuredby three questions. Respondentswhere asked
how well they felt that the following statementsapplied
to them: 'I feel that my studieshave no meaning','I want
to quit the school', 'I don't get along with the school
establishment/teachers'
(0=applies very poorly, = applies
rather poorly, 2= sometimes applies, 3= often applies,
4 = almost alwaysapplies). The scores were summarized
on a scale (Cronbach'sAlpha= 0.73).
Imitation
Following Haynie (2002), respondents were asked how
many of their friends had done any of the following in
the last 12 months: stolen something worth more than
5000 kroners(about $70), burglarizeda buildingor a car
to steal something, damagedthings that did not belong
to them (0 = none of them, 1= few of them, 2 = severalof
them, 3 = most of them, 4 = all of them). The scoreswere
summarizedon a scale (Cronbach'sAlpha= 0.82).
Delinquency
Delinquentbehaviourwas measuredby a five-item summaryscale of how many times respondentshad done the
following in the last 12 months: stolen something that
was worthless than 5000 kroner(about$70), stolensomething that was worth more than 5000 kroner, employed
violence to rob someone, burglarizeda building or a car
to steal something, damagedthings that did not belong

to them (0=never, through to, 6=18 times or more).


The scores were summed (Cronbach'sAlpha= 0.77).
Community-level Measures
We createdcommunity-levelmeasuresof social integration and anomie by converting the individual-level
scores on these measuresinto school means. We present
these scores as community-levelvariablesin the statistical analysis.

Results
We use a series of hierarchicallinear regressionmodels
with a randominterceptterm and fixed slopes (Brykand
Raudenbush, 1992). This strategyallows us to examine
the contextual effects of community-level variables on
the individual'sdelinquency,while controlling for individual level processes, and vice versa. Moreover, the
method enables us to examine whether individual-level
and community-levelintegrationand anomie moderate
imitation processes.

Direct Effects
Table 2 shows the statistical effects of individual and
community-levelindicatorson delinquency.In Model 1,
delinquencyis regressedon the individualand community-level indicators of social integration.As predicted,
adolescentsthat have higher levels of integrationreport
less delinquency on average, net of the level of social
integration in their communities. These effects are
significantfor all the indicatorsof interpersonalintegration except for the one indicating whether or not the
subjectlives in a two-parenthousehold (the coefficients
are three to ten times largerthan their respectivestandard errors).
The resultsin Model 1 also provide some evidencefor
the hypothesis that the level of social integrationin the
community has a negative, contextual effect on delinquent behaviour. Two community-level indicators of
social integration thus exhibit significant, contextual
effects on delinquency.Adolescentsbelonging to urban
communities where there is a high level of religious
activity report significantly less delinquent behaviour,
net of their own level of religiousactivityand other controls. Second, youths belonging to urban communities
that have a high proportion of two-parent households
reportsignificantlyless delinquency,net of whetherthey
live in two-parenthouseholds themselves.
These resultsprovide evidence in favour of the multilevel natureof socialintegration,namely,that conformity

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

DURKHEIM,SOCIALORDERAND DEVIANCEI 279

and communityvariables
Table 2 Regressionof individual
delinquencyon individual

Individuallevel
Gender (male= 1)
Familydensity
Two-parent home
Religiousactivity
Politicalintegration
School commitment
Grades
Normlessness
School anomie
Peer delinquency
Communitylevel
Familydensity
Two-parent home
Religiousactivity
Politicalintegration
School commitment
Grades
Normlessness
School anomie
FixedIntercept
Random intercept
Residual
-2 Log-likelihood(Deviance)

Model 3

Model 2

Model 1
1.45*
-0.07*
0.01
-0.19*
-0.20*
-0.31*
-0.03*

(0.12)
(0.01)
(0.16)
(0.12)
(0.02)
(0.03)
(0.01)

1.31*
-0.05*
0.01
-0.13
-0.13*
-0.15*
-0.02
0.11*
0.22*

(0.11)
(0.01)
(0.16)
(0.12)
(0.02)
(0.03)
(0.01)
(0.01)
(0.03)

0.67*
-0.03*
0.10
-0.03
-0.06*
-0.04
0.00
0.05*
0.12*
0.90*

(0.10)
(0.01)
(0.13)
(0.10)
(0.02)
(0.03)
(0.01)
(0.01)
(0.02)
(0.03)

0.06
-3.89*
-3.19*
0.21
0.26
0.05

(0.12)
(1.22)
(1.57)
(0.23)
(0.29)
(0.06)

0.13
-3.57*
-1.30
0.21
0.12
0.07
0.10
0.35
2.42
0.03
10.14
18109

(0.11)
(1.11)
(1.50)
(0.21)
(0.27)
(0.06)
(0.16)
(0.27)

0.12
-1.98*
-1.25
0.08
-0.07
-0.02
0.04
0.05
3.86
0.00
7.42
17007

(0.08)
(0.83)
(1.09)
(0.16)
(0.21)
(0.05)
(0.12)
(0.20)

12.50
0.07
10.62
18281

Note: Coefficientsare maximumlikelihoodestimatorsfrom HLMregression;numbersin parenthesesare standarderrors.


*Absolutevalueof the t-ratiois higherthan2.0.

derivesnot only from interpersonalintegration,but also


from the level of social integrationin the person's community.
Model 2 adds general anomie and school anomie to
the equation. As predicted, individualswho experience
anomie reportsignificantlyhigher involvementin delinquency. School anomie also has a significant, positive
effect on delinquency.
Durkheim's theory implies that social integration
influences delinquencypartlythrough the experienceof
anomie and school anomie. Hence, anomie and school
anomie should mediatesome of the effectsof social integration on delinquent behaviour. As predicted, adding
the anomie indicatorsto the equation reducesthe effects
of the social integration variables. Thus, the negative
effect of school commitment on delinquency decreases
by about half (from -0.31 to -0.15). The effectsof family
density, political commitment, and religious activity
decreasesby about a third. It is noteworthythat adding
anomie to the equation also produces a substantial
decreasein the coefficient for the effect of communitylevel religious activity (from -3.19 to -1.30), indicating

that the individual experience of anomie explains the


contextualeffectthat community-levelreligiousintegration has on delinquency.
These results lend furthersupport to our hypothesis,
indicatingthatlow levelsof socialintegrationweakencollectivesentimentsand thus decreasethe clarityof and sentiments for social norms and rules, and thus increasethe
experienceof anomie and school anomie.In turn, anomie
and school anomie positivelyinfluencedelinquency.
Finally, in Model 3, we add peer delinquency to the
equation. As expected, peer delinquency has a strong
positive effect on self-reporteddelinquency (the coefficient of 0.90 is 30 times the size of the standarderror).
Moreover, the results are consistent with substantial
mediatingeffects.When adding peer delinquencyto the
model, the effects of the social integration and anomie
variablesdecreaseby about half. The coefficient for the
effect of school commitment on delinquency decreases
from -0.15 to an insignificant-0.04. Finally,the community-leveleffect of a two-parenthousehold decreases
considerablywhen addingpeer delinquencyto the equation (from -3.57 to -1.98).

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

280 | THORLINDSSONAND BERNBURG

This finding is consistent with prior research that


suggeststhat contactswith delinquentpeers mediate the
effects of conventional social ties on delinquent behaviour (Elliott et al., 1985). Adolescentswith low levels of
social integration and high levels of anomie and school
anomie are more likely to associate with delinquent
peers. Also, it is noteworthy that the findings suggest
that peer delinquency explains about half of the effects
of the community prevalenceof two-parent households
on delinquency. The effect of this community-level
indicator was not mediated by anomie or school
anomie (compare Models 1 and 2). This result is in
accordance with the celebrated social disorganization
theory (Sampson and Groves, 1989), which argues that
a high proportion of single-parent households in the
community weakens the ability of the community to
control the behaviour of teenage peer groups. Adolescents belonging to communities where two-parent
homes are less common are more likely to associate
with delinquent peers, regardlessof their own family
structure,and more likely to participatein delinquency
as a result.

Moderating Effects
Durkheim's theory implies that the social context of
integration and anomie should moderate the influence
of imitation on delinquency.First,personalties to social
institutions should decreasethe individual'ssusceptibility to imitation (peer influences). The relationship
between self-reporteddelinquencyand peer delinquency
should be weakeramong stronglyintegratedindividuals.
Conversely, there should be a stronger relationship
between self-reporteddelinquencyand peer delinquency
with increasedanomie.
Second, peer-group social interactions take place in
community contexts. High levels of social integrationin
the community should diminish the effectsof imitation.
In other words, the level of social integrationin the community should moderate the effects of imitation. In
communities with high levels of social integration,peer
delinquencyshould have a weaker effect on delinquent
behaviour,net of the moderatingeffectsof interpersonal
social integration.
Table 3 reports the statistical interaction effects. In
Models 1-8, delinquency is regressedon multiplicative
product terms, while holding constant individual and
community-level indicators of social integration and
anomie (to savespaceonly the effectsof the focalvariables
and the multiplicative product terms are shown in the
table). The findings lend support to our hypotheses,

showing consistent, significant statistical interaction


effects as predicted. First, most of the individual level
interaction terms are significant (except for Model 2,
the interaction terms are three to ten times largerthan
their respective standard errors). Thus, the effect of
imitation (peer delinquency) on delinquency is weaker
as individuals have higher levels of integration into the
family, religion, the polity, and the school (Models 16). Conversely, the effect of imitation on self-reported
delinquency is stronger among youths that experience
anomie and school anomie (Models 7 and 8, respectively).
Second, while controlling for these individual-level
interactioneffects,Table 3 shows a few significantcrosslevel statistical interaction effects. As predicted, imitation (peer delinquency) has significantlyweaker effects
on delinquency in urban communities where the proportion of two-parent households is higher (Model 2).
Imitation also has significantlyweaker effects on delinquent behaviourin urbancommunities in which there is
a higher level of political integration(Model 3) and religious activity (Model 6). High levels of community
social integrationand community control diminish the
effect of having delinquent peers on adolescent delinquency, even while taking into account personal social
integration.

Discussion and Conclusion


We have examined Durkheim'sclassicaltheory of social
order and deviance using adolescent delinquency as a
case of deviant behaviour. Following Giddens and
Sawyer,we have emphasizedthe structuralnatureof the
theory and the emergentpropertiesof the social. In support of these ideas, the community-level indicators of
religious activity and prevalence of two-parent households exhibit a significant, negative (contextual) effect
on delinquency.Moreover, the results indicate that the
experience of anomie partly accounts for the effects of
social integrationon delinquency.This finding supports
the idea that low levels of social integrationweaken collective sentiments for social norms and rules, increasing
the probability of deviant behaviour. It is noteworthy
that individual-levelanomie explains a substantialpart
of the effect of community-level religious activity on
delinquency.This finding lends support to Durkheim's
emphasis on the role of religious life as an important
source for collective authority, strengthening social
norms and collectivesentiments(see Shillingand Mellor,
1998, for a recent discussionof this issue).

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

DURKHEIM,SOCIALORDERAND DEVIANCE | 281

and peerdelinquency
Table 3 Statisticalinteractionsbetween integration/anomie
Selected independent variables

Selected independent variables


Individualfamilydensity
Community familydensity
Peer delinquency
Individualfamilydensityx
peer delinquency
Community familydensityx
peer delinquency
-2 Log-likelihood=
Individualtwo-parenthome
Community two-parenthome
Peer delinquency
Individualtwo-parenthome x
peer delinquency
Communitytwo-parenthome x
peer delinquency
-2 Log-likelihood =

Model 1
-0.03* (0.01)
0.14(0.08)
1.70*(0.28)
-0.02* (0.00)
-0.01 (0.03)
16999
Model 2
0.42 (0.32)
2.23 (1.78)
0.03 (0.12)
-0.06 (0.06)
-0.86* (0.34)
17050

Individualpolitical integration
Community political integration
Peer delinquency
Individualpolitical integrationx
peer delinquency
Community political integrationx
peer delinquency

Model 3
0.25*(0.04)
0.73*(0.28)
2.74*(0.44)
-0.07* (0.01)
-0.15* (0.05)
16935

-2 Log-likelihood =

Individualschool commitment
Community school commitment
Peer delinquency
Individualschool commitmentx
peer delinquency
Community school commitmentx
peer delinquency
-2 Log-likelihood=

Model 4
0.37*(0.06)
0.59 (0.39)
2.33*(0.45)
-0.10* (0.01)
-0.13 (0.07)
16956

Individualgrades
Communitygrades
Peer delinquency
Individualgradesx
peer delinquency
Community gradesx
peer delinquency
-2 Log-likelihood=
Individualreligiousactivity
Community religiousactivity
Peer delinquency
Individualreligiousactivityx
peer delinquency
Community religiousactivityx
peer delinquency
-2 Log-likelihood =

Individualnormlessness
Community normlessness
Peer delinquency
Individualnormlessnessx
peer delinquency
Community normlessnessx peer
delinquency
-2 Log-likelihood =

Individualschool anomie
Community school anomie
Peer delinquency
Individualschool anomie x
peer delinquency
Community school anomie x
peer delinquency
-2 Log-likelihood=

Model 5
0.15*(0.02)
-0.11 (0.09)
1.20*(0.31)
-0.03* (0.00)
0.02 (0.02)
16979
Model 6
1.27*(0.24)
2.36(1.95)
1.53*(0.20)
-0.28* (0.05)
-0.82* (0.35)
17012

Model 7
-0.18* (0.02)
-0.13 (0.21)
-0.94 (0.68)
0.05*(0.00)
0.04 (0.04)
16861

Model 8
-0.28* (0.04)
0.10 (0.34)
0.46 (0.36)
0.08*(0.01)
-0.02 (0.05)
16896

Notes: Thedependentvariableis self-reporteddelinquency.Coefficientsare maximumlikelihoodestimators;numbersin parenthesesare standarderrors.In


Models1-6 all variablesincludedin Model 1 in Table2 are controlled.In Models7-8 all variablesincludedin Model2 in Table2 are controlled.
*Absolutevalueof the t-ratiois higherthan2.0.

These findings underscorethe importanceof specifying and testing empirically the social mechanism
involved in social emergence. Social emergence cannot
be treated as a logical property of social structurebut
needs to be addressedas an empirical problem. These
findings draw attention to an important theme in
Durkheim's work, a theme emphasized by Giddens in
relation to the emergent qualities of social structure,

namely, the ideal or spiritual characterof social facts


(Giddens, 1972: 43). They also highlight the recent
revivalof Durkheim's(1912/1995) work on religionand
ritualin sociologicaltheory (Smithand Alexander,1996).
This work focuses on Durkheim'snotion that social and
moral orders are 'recharged' during ritualistic group
assemblies that harness 'people's passions to the symbolic orderof society' (Shillingand Mellor, 1998: 196).

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

282 | THORLINDSSONAND BERNBURG

Our findings demonstratethe need to consider imitation effects in the context of social integration and
anomie. While prior work has often ignored the potential role of social imitation in Durkheim's theory (see
Thorlindsson and Bjarnason,1998), Durkheim implies
that social integration and anomie may contextualize
imitation processes. Hence, we have hypothesized that
the effect of imitation on delinquent behaviour should
be contingent on (i) personal social integration and
anomie and (ii) on the level of social integrationin the
community. The findingslend supportto both hypotheses. First, association with delinquent peers has weaker
effectson the delinquencyof adolescentswith high levels
of personalsocial integration,but the effectsare stronger
as adolescentshave higher levels of anomie. These findings generalizeprior researchshowing that the effects of
peer delinquency on self-reported delinquency is contingent on social ties with parents (Warr, 1993;see also
Elliottet al., 1985). Second, our findingssuggestthe subtle role of community-levelsocial integrationin contextualizing imitation effects. The lower the community
levels of religious integration,political integration, and
two-parent households, the stronger are the imitation
effects.It is noteworthythat these cross-levelinteraction
effects are strong enough to emerge as significantin our
sample of 34 urban schools.
Durkheim's theory is a general sociological theory,
combining issues of social order and deviance into one
perspective. It postulates that there are various social
mechanisms operating simultaneously in society, the
balance of which may vary from time to time and from
one social group (society) to another.Thus, our findings
suggestthat Durkheim'stheory of social order is a good
candidate for relating contemporarytheories of social
control, anomie and differentialassociation/sociallearning. Delinquency and crime have traditionallybeen the
subject of criminology, a special subfield of sociology.
Influential criminological theories, including anomie
theory(Merton,1967/1994),socialbonding/controltheory
(Hirschi,1967),differentialassociationtheory(Sutherland
and Cressey, 1984) and social learning theory (Akers,
1977), have been special cases or elaborationsof general
sociological theories (Liska, 1991; Akers, 1992). The
development of these specializedtheories has been the
basis for criminological research, research that indeed
has inspired our empirical analysis. But this development has also encourageda division of labour that has
limited the scope of sociological analysis (see Akers,
1992). Thus, the differenttheories are designed to focus
on one aspect of the social realityof crime while ignoring other importantaspectsof it (Warr,1993).4

Our findings have implications for general sociological theory aimed at understanding social solidarity in
contemporarysociety. Durkheim gave priority to processes of social integration and regulation/anomie,
implying that they channeled imitation and unstable
social currents.Taken in context with his other works,
Durkheim's approach implies that these processes are
partof the broadersocial structuresof societalsolidarity.
He argues that as the economy gains dominance over
other social institutions, social regulation becomes
weakerand social limits on desires and expectationsare
removed. The forces of unchecked economic forces,
combined with limitless expectations and self-interest
can create social anomie and weaken social integration,
leading to suicide and criminal behaviour (Durkheim,
1897/1951: 251-258). Changes in capitalism, with
increasing emphasis on individuality and the domination of the market-orientedeconomy over other social
institutions,draw our attention to the interplayof social
integration, anomie, and imitation as important processes affecting social order and deviance (Bernburg,
2002). These processesshould be analysedin relationto
central characteristicsof social solidarity. Messner and
Rosenfeld (1994, 1997) have recently elaborated on
Merton's anomie theory with referenceto Durkheim's
concept of social integration. These authors argue that
when the balance of power among societal institutions
tilts toward the capitalist economy, the ability of
non-economic institutions (the family, education, and
the polity) to fulfill their function is weakened. Thus,
social integration becomes weaker, resulting in the
weakening of the authorityof social norms throughout
the society. The present study provides a bridge for this
approach,connecting individual and community levels,
underscoringthe role of social integration and anomie
in contextualizingimitation and delinquency.

Notes
1. Severalscholarshave highlightedthis neglected line
of thought in suicide research.Stack (1987, 2003),
elaborating on Durkheim's theory of suicide, uses
the concept of differentialidentification to explain
the suggestion-imitation effect of media coverage
on suicide rates.Thorlindssonand Bjarnason(1998)
have highlighted the role of suicidal suggestion in
mediatingthe effect of family integrationand family
regulationon adolescentsuicidality.
2. Durkheim did not do justice to Tarde's theory of
imitation (Thorlindsson,2000). The theory of imitation that Durkheimrejectsis much narrowerthan

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

DURKHEIM,SOCIALORDERAND DEVIANCE| 283

Tarde himself proposed. In fact, Durkheim's own


category of social imitation is strikingly similar to
Tarde's definition of imitation. Tarde'sfirst law of
imitation emphasizes both differential association
and social learning as mechanisms in the imitation
process. Tarde'ssecond law proposes that individuals' higher social standingservesas a model for individualsof lesserstatus.FollowingDurkheim/Tarde's
terminology, we use what is essentiallythe second
definitionin Durkheim'ssystem,that is, social imitation. This definition means that imitation may
involve various social mechanisms,including differentialassociation(Sutherlandand Cressey,1984) and
sociallearning(Akers,1977).Althoughthese theories
arenot limitedto the influenceof deviantpeers,most
tests have focused on the associationbetween selfreporteddelinquencyand delinquentbehaviour.
3. Parents'educationalattainmentis a more important
predictorof adolescent delinquencyin Iceland than
are traditional class measures (Thorlindsson et al.,
1998). We have conducted supplementaryanalysis
controlling for parents' educational attainment on
the individualand the community levels.This analysis does not change the results presented in the
paper.We do not include a measureof parents'education here due to a high number of missing values
on this measure.
4. Attempts to test two or more theories have either
placed the specialized theories in competition
(Krohn et al., 1984;Matsuedaand Heimer, 1987) or
used different theories as components of an integrated theoretical framework (Elliott et al., 1985).
Scholarshave criticizedthese attempts for lumping
together theories based on contradictory assumptions about human behaviour (Hirschi, 1989; Liska
et al., 1989;Gibbonsand Krohn, 1991).

Acknowledgements
This projectwas supportedby the Universityof Iceland
ResearchFund.We thank three anonomous reviewersof
EuropeanSociologicalReview for helpful comments on
an earlierversion of this paper.

References
Agnew, R. (1997). The Nature and Determinants of
Strain:Another Look at Durkheim and Merton. In
Passas, N. and Agnew, R. (Eds), The Future of
Anomie Theory.Boston, MA: NortheasternUniversity Press.

Akers, R. L. (1977). Deviant Behavior:A SocialLearning


Approach.Belmont, CA:Wadsworth.
Akers,R. L. (1992). LinkingSociologyand its Specialties:
The Case of Criminology.SocialForces,71, 1-16.
Akers, R. L., Krohn, M. D., Lanza-Kaduce,L. and
Radosevich,M. (1979). Social Learningand Deviant
Behavior: A Specific Test of a General Theory.
AmericanSociologicalReview,44, 636-655.
Bernburg, J. G. (2002). Anomie, Social Change, and
Crime: An Examination of Institutional-Anomie
Theory. British Journal of Criminology,42, 729742.
Bernburg, J. G. and Thorlindsson, T. (2001). Routine
Activities in Social Context: A Closer Look at the
Role of Opportunity in Deviant Behavior. Justice
Quarterly,18, 543-567.
Bjarnason,T. (1995). Administration Mode Bias in a
School Surveyon Alcohol, Tobacco and Illicit Drug
Use. Addiction,90, 555-559.
Brake,M. (1985). ComparativeYouthCulture:The Sociology of Youth Subculturein America, Britain and
Canada.London:Routledge.
Bryk,A. and Raudenbush,S. (1992). HierarchicalLinear
Models: Application and Data Analysis Methods.
NewburyPark,CA:Sage.
Dean, D. (1961). Alienation:Its meaning and measurement. AmericanSociologicalReview,25, 753-758.
Durkheim,E. (1893/1984). TheDivisionof Laborin Society. New York:The FreePressof Glencoe.
Durkheim, E. (1895/1982). The Rules of Sociological
Method.New York:The FreePress.
Durkheim,E. (1897/1951). Suicide:A Studyin Sociology.
London:Routledge.
Durkheim, E. (1912/1995). The ElementaryForms of
ReligiousLife.New York:The FreePress.
Durkheim, E. (1925/1961). MoralEducation.New York
and London:The FreePress.
Elliott, D., Huizinga, D. and Ageton, S. (1985). Explaining Delinquencyand Drug Use. Beverly Hills, CA:
Sage.
Gibbons, D. C. and Krohn, M. D. (1991). Delinquent
Behavior.EnglewoodCliffs,NJ:PrenticeHall.
Gibbs, J. (2000). Status Integrationand Suicide. Social
Forces,79, 363-384.
Giddens,A. (1971). Capitalismand ModernSocialTheory.
Cambridge,UK: CambridgeUniversityPress.
Giddens, A. (1972). Emile Durkheim:SelectedWritings.
Cambridge,UK: CambridgeUniversityPress.
Gunnlaugsson, H. and Galliher, J. (2000). Waiward
Icelanders:Punishment,BoundaryMaintenance,and
the Creationof Crime. Madison, WI: University of
Wisconsin Press.
Haynie, D. L. (2001). Delinquent Peers Revisited:Does
Network Structure Matter? American Journal of
Sociology,106, 1013-1057.

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

284 j THORLINDSSONAND BERNBURG

Haynie, D. L. (2002). Friendship Networks and


Delinquency:The Relative Nature of Peer Delinquency. Journal of Quantitative Criminology, 18,
99-134.
Hirschi, T. (1967). Causesof Delinquency.Berkeley,CA:
Universityof CaliforniaPress.
Hirschi, T. (1989). Problems with Theoretical Integration. In Messner, S., Krohn, M. D. and Liska,A. E.
(Eds), TheoreticalIntegrationin the Study of Deviance and Crime:Problemsand Prospects.Albany,NY:
StateUniversityof New York.
Johnson, B. (1965). Durkheim's One Cause of Suicide.
AmericanSociologicalReview,30, 875-886.
Jensen, G. F. (1972). Parents, Peers, and Delinquent
Action: A Test of the Differential Association
Perspective.AmericanJournalof Sociology,78, 562575.
Kandel, D. B. and Davies, M. (1991). FriendshipNetworks, Intimacy, and Elicit Drug Use in Young
Adulthood:A comparison of two competing theories. Criminology,29,441-470.
Kornhauser, R. (1978). The Social Sources of Delinquency:An AppraisalofAnalyticModels.Chicago,IL:
Universityof ChicagoPress.
Krohn, M. D. (1986). The Web of Conformity:A Network Approach to the Explanation of Delinquent
Behavior.SocialProblems,33, 581-593.
Krohn,M. D., Lanza-Kaduce,L. and Akers,R. L. (1984).
Community Context and Theories of Deviant
Behavior:An Examination of Social Learning and
Social Bonding Theories. The SociologicalQuarterly,
25,353-371.
Liska,A. E. (1991). From the Section Chair. Newsletter
of the Crime, Law, and Deviance Section of the
American Sociological Association. New York:
AmericanSociologicalAssociation.
Liska,A. E., Krohn, M. D. and Messner, S. F. (1989).
Strategiesand Requisitesfor the TheoreticalIntegration in the study of Crime and Deviant Behavior.In
Messner, S., Krohn, M. D. and Liska,A. E. (Eds),
TheoreticalIntegrationin the Study of Devianceand
Crime: Problemsand Prospects.Albany, NY: State
Universityof New York,pp. 1-19.
Lukes, S. (1972). Emile Durkheim:His Life and Work.
New York:Harperand Row.
Matsueda, R. L. and Heimer, K. (1987). Race, Family
Structure,and Delinquency:A Test of Differential
Association Theory. American SociologicalReview,
52,826-840.
Matza, D. (1964). Delinquencyand Drift. Englewood
Cliffs,NJ:PrenticeHall.
Merton, R. K. (1967/1994). Social Structure and
Anomie. In Traub,S. H. and Little,C. B. (Eds), Theories of Deviance (Fourth Edition). Itasca, IL: F.E.
PeacockPublishers,Inc, pp. 114-148.

Messner, S. F. and Rosenfeld,R. (1994). Crimeand The


American Dream (Second Edition). Belmont, CA:
WadsworthPublishingCompany.
Messner,S. F. and Rosenfeld,R. (1997). Markets,Morality, and an Institutional-AnomieTheory of Crime.
In Passas, N. and Agnew, R. (eds), The Future of
Anomie Theory.Boston, MA: NortheasternUniversity Press.
Sampson, R. J. and Groves, W. B. (1989). Community
Structureand Crime:Testing Social Disorganization
Theory.AmericanJournalof Sociology,94, 774-802.
Sampson, R. J. and Laub,J. H. (1990). Crime and Deviance over the Life Course: The Salience of Adult
Social Bonds.AmericanSociologicalReview,55, 609627.
Sawyer,R. K. (2002). Durkheim's Dilemma: Toward a
Sociology of Emergence. Sociological Theory, 20,
227-247.
Shilling,C. and Mellor, P. A. (1998). Durkheim,Morality and Modernity: Collective Effervescence,Homo
Duplex and the Sources of Moral Action. British
Journalof Sociology,49, 193-209.
Smith, P. and Alexander, J. C. (1996). Review essay:
Durkheim's religious revival. American Journal of
Sociology,102, 585-592.
Stack, S. (1987). Celebritiesand Suicide: A Taxonomy
and Analysis, 1948-1983. American Sociological
Review,52, 401-412.
Stack,S. (2003). Media Coverageas a RiskFactorin Suicide. Journal of Epidemiology and Community
Health,57, 238-240.
Stack, S. and Wasserman, I. (1995). The Effect of
Marriage, Family, and Religious Ties on African
AmericanSuicide Ideology.Journalof Marriageand
the Family,57, 215-222.
Sutherland,E. H. and Cressey,D. R. (1984). Differential
Association Theory. In Kelly, D. H. (Ed.), Deviant
Behavior (Second Edition). New York: St. Martins
Press.
Tarde, G. (1903). The Laws of Imitation. New York:
Holt.
Thorlindsson,T. (1983). Social Organizationand Cognition. Human Development,26, 289-307.
Thorlindsson, T. (2000). The Durkheim Tarde Debate
on the Causes of Suicide. Unpublished Manuscript,
Departmentof Sociology,Universityof Iceland.
Thorlindsson, T. and Bjarnason,T. (1998). Modeling
Durkheim on the Micro Level: A Study of Youth
Suicidality. American SociologicalReview, 63, 94110.
Thorlindsson,T., Sigfusdottir,I. D., Bernburg,J. G. and
Halldorsson,V. (1998). VimuefnaneyslaUngs Folks:
Umhverfiog Adstaedur(The Social Contextof Youth
Drug Use). Reykjavik:Icelandic Institute for Educational Research.

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

DURKHEIM,SOCIALORDERAND DEVIANCEI 285

Thornberry,T. P. (1989). Towardan InteractionalTheory of Delinquency.Criminology,25, 863-891.


Tittle, C., Burke,M. J. and Jackson,E. F. (1986). Modeling Sutherland'sTheory of DifferentialAssociation:
Toward an Empirical Classification. Social Forces,
65, 405-432.
Warr,M. (1993). Parents,Peersand Delinquency.Social
Forces,72, 247-264.
Warr, M. and Stafford, M. (1991). The Influence of
Delinquent Peers:What they Think or What They
Do? Criminology,29, 851-866.

Authors' Addresses
ThorolfurThorlindsson,Facultyof Social Science, University of Iceland, Oddi vid Sturlugotu, 101 Reykjavik, Iceland.Email:thorotho@rhi.hi.is
Jon GunnarBernburg,Facultyof Social Science,University of Iceland, Oddi vid Sturlugotu, 101 Reykjavik,
Iceland.
Manuscriptreceived:July2003

This content downloaded from 144.32.128.70 on Wed, 17 Feb 2016 14:14:32 UTC
All use subject to JSTOR Terms and Conditions

You might also like