03 P-Watel (jr/d) 2/11/03 2:32 PM Page 21

Patrick Peretti-Watel

Neutralization theory and the denial of risk:

some evidence from cannabis use among
French adolescents*

ABSTRACT

In contemporary societies, risk culture and risk profiling lead to the stigmatiza-
tion of unhealthy behaviours as ‘risky’. Risk denial theory refers to a cognitive way
to deal with risky behaviours and can be considered as an updated variant of Sykes
and Matza’s neutralization theory. People neutralize the ‘risky’ label using
specific techniques that must be added to those previously enlisted by Sykes and
Matza. This paper introduces and discusses three techniques of risk denial:
scapegoating, self-confidence and comparison between risks. As it is usually
defined and studied as a ‘risky behaviour’, cannabis use provides a relevant
example to illustrate these types of risk denial, thanks to various ethnographic
studies (including Becker’s seminal work on marijuana smokers) and quantita-
tive French data from the 1999 European School Survey on Alcohol and Other
Drugs (ESPAD). In order to deny the ‘risky’ label, cannabis users scapegoat ‘hard
drugs’ users, they emphasize their own ability to control their consumption
personally, or they compare cannabis and alcohol risks. The paper concludes
with suggestions for further analyses of risk denial.

KEYWORDS: Risk denial; techniques of neutralization; risky behaviours;

cannabis use; adolescence

INTRODUCTION

In an article which became famous, Sykes and Matza (1957) suggested a

new theory of delinquency. According to them, most delinquents are
embedded in the whole society: they don’t live in a separate world which
would allow them to escape from the pressure to conformity to the
dominant social order. This theoretical viewpoint seems especially relevant
to the study of juvenile delinquency because adolescents depend strongly
on adults. Sykes and Matza argued that delinquents often remain sensitive
to law-abiding values but succeed in defusing them by using techniques of

British Journal of Sociology Vol. No. 54 Issue No. 1 (March 2003) pp. 21–42
© 2003 London School of Economics and Political Science ISSN 0007-1315 print/1468-4446 online
Published by Routledge Journals, Taylor & Francis Ltd on behalf of the LSE
DOI: 10.1080/0007131032000045888
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 22

22 Patrick Peretti-Watel

neutralization. These techniques are ‘justifications for deviance that are

seen as valid by the delinquent but not by the legal system or society at
large’ (Sykes and Matza 1957: 666). Such justifications can follow a deviant
act in order to protect the delinquent from remorse or blame, but they can
also precede this act and make it possible. Some subsequent works provided
further evidence for this theory (Reiss 1961; Short and Strodtbeck 1965;
Critchlow 1983), especially in the field of juvenile delinquency.
Neutralization theory demystifies deviance by emphasizing the fact that
delinquents are not so different from law-abiding people, even if common
sense tends to associate a deviant behaviour with an alleged deviant person-
ality (‘evil deeds reveal evil temper’). They are still committed to
conventional norms, but sometimes they neutralize such norms temporarily
by defining them as inapplicable, irrelevant or unimportant. Thus a deviant
act can simply results from the conjunction of opportunity and neutraliza-
tion, without requiring an antisocial personality or family problems.
Demystifying deviance is also the goal pursued by Becker in Outsiders
(1963), another famous work on the sociology of deviance. He considers
rationalization as a crucial dimension for the analysis of any kind of crime
(Becker 1970: 332–3). Becker agrees with Sykes and Matza, and he quotes
at length their article to set out his sequential model of deviance (Becker
1963: 28–9). Dealing with marijuana use in two chapters, Becker doesn’t
refer to ‘techniques of neutralization’ because those chapters were orig-
inally published in 1953, before Sykes and Matza’s article. However, he
employs very close words to describe the ‘subjective’ career of marijuana
users. At each stage of their career, these users must be able to neutralize
the stereotype of the ‘dope fiend’ slave to the drug and unable to control
his behaviour. Many French sociologists have been influenced by Outsiders,
above all those working on drugs users. They also refer to neutralization
theory, especially for cannabis use.
I argue that neutralization techniques listed by Matza and Sykes are not
relevant to study cannabis use because this consumption is labelled as
deviant and risky at once. Cannabis users elude the ‘risky’ label with specific
neutralization techniques which refer to risk denial theory (Peretti-Watel
2000: 192–231). This theory pays attention to the growing place of risk and
risk profiling in contemporary societies, and I propose to consider it as an
updated variant of neutralization theory. I first compare these two theories.
Both refer to a cognitive rather than behavioural adaptation: sometimes
people’s actions are inconsistent with their beliefs, so they must restore
harmony by changing their behaviour (behavioural adaptation) or by
adjusting their beliefs (cognitive adaptation). But the denial of risk obeys
slightly different constraints, and imply specific techniques. Then I
confront risk denial theory to qualitative data about cannabis use (Becker’s
interviews and more recent British and French data). Finally, using French
quantitative data from the 1999 European School Survey on Alcohol and
Other Drugs (ESPAD), I build up statistical tests for some hypotheses
deriving from risk denial theory.
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 23

Neutralization theory and the denial of risk 23

NEUTRALIZATION THEORY AND DENIAL OF RISK

Risk and Deviance

Nowadays, risky behaviours tend to be labelled as deviant behaviours. Of
course, risk-taking often implies lawbreaking (on the road: driving too fast,
driving drunk, blind overtaking . . .). Moreover, as compulsory insurance
covers a wide scope of activities, the monetary cost of many individual risks
is shared between contributors as a whole, so private risky behaviours
become deviant because they affect collective wealth. However, this
labelling must also be linked with the implications of the contemporary
‘risk culture’ on day-to-day life (Giddens 1991): traditional notions of fate
fade away and everyone is exhorted to manage his own ‘reflexive project’
by developing a calculative attitude to the open possibilities of action, by
staying continuously aware of a growing number of risks and opportunities,
in order to colonize and secure the future. This new ‘Moral Requirement’
is illustrated by the figures put forward in tobacco prevention campaigns.
For example, life expectancy for 25-year-olds is 48 years in non-smokers,
versus 40 years in heavy smokers: young smokers are therefore expected to
plan out their long-term health by quitting their risky behaviour now.
People (and especially young people) who indulge in risky behaviours
shorten their life expectancy, they blindly endanger their future rather
than colonizing it, and they show an inability or refusal to manage their
‘reflexive project’; thus, they break the new behavioural norms induced by
the risk culture.
This is the reason why risky behaviours, just as deviant behaviours before,
can be characterized by the fact that they are ‘socially defined as a problem,
a source of concern, or as undesirable by the norms of conventional society
and the institutions of adult authority’ ( Jessor and Jessor 1977: 33). Both
problematic behaviours are connected by theories that postulate the exist-
ence of a ‘general syndrome of deviance’ (Donovan and Jessor 1985) or a
‘problem behaviour syndrome’ (Dryfoos 1990), especially among youth.
Such theories lead to empirical joint analyses of criminal behaviours
(aggression, theft, vandalism . . .), drugs use, dangerous driving and sexual
promiscuity (Osgood et al. 1988; Parker and Auerhahn 1998; Leigh 1999).
As deviance before, risk-taking gives rise to psychological explanations: risk-
taking could be caused by a specific personality trait (‘sensation seeking’
Zuckerman 1979) due to genetic factors and familial and extra-familial
influences (Miles et al. 2001), or by an ‘unrealistic optimism’ (Weinstein
1980).
Risk and deviance are also converging because risk profiling often leads
to stigmatization of a deviant minority. According to Giddens (1991), risk
profiling is now a central part of modernity, a key activity in reflexive
societies. Risk profiling consists of analysing the distribution of a given risk
among a population, in order to define clusters which are more exposed
than others and constitute targets for public health policy. Yet those
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 24

24 Patrick Peretti-Watel

‘objective’ categories built up by experts don’t remain morally neutral

because they never escape from dominant values. They often give ‘good
reasons’ to support segregation of marginal groups: just as people used to
blame epidemic diseases on the poor, the Jews or any kind of outsiders over
the past centuries, today heroin users, homosexuals or simply immigrants
have been labelled as minorities ‘at risk’, because they are potential incu-
bators and spreaders of AIDS, and also because they are outsiders (Rosen-
berg 1988; Douglas 1992; Evans 1992).
The seeds of this shift from the ‘risky’ label to scapegoating are sown in
the very science devoted to risk profiling: epidemiology. Reflecting the
contemporary risk culture, modern ‘risk factor epidemiology’ moves the
locus of control of health issues from factors outside human control to
factors inside personal control (Skolbekken 1995). This privatization of risk
‘gives primacy to personal autonomy and action and seeks to induce
personal behaviour change rather than to promote social interventions’
(Rockhill 2001: 365). By focusing on the individual as the sole locus of risk,
epidemiological studies can lead to the assignation of personal responsi-
bility for health problems, which sustains the stigmatization of risk groups
and ‘victim blaming’ (Coughlin 1998). Thus risk profiling often rhymes
with scapegoating, and as the ‘Risk Society’ tends to become a ‘Scapegoat
Society’ (Beck 1992: 75), we can expect that some people try to avoid the
‘risky’ label just as others try to neutralize the ‘deviant’ one.

Neutralization, Risk Denial and Cognitive Dissonance

Sometimes people act in a way which is labelled as risky or deviant by the
dominant social order. If they are committed to this order, such a situation
is a special case of ‘cognitive dissonance’. The theory of cognitive disson-
ance (Festinger 1957) starts from a common sense remark: everyone tries
to avoid inconsistency between his behaviour and his beliefs. In a behav-
iourist schema, beliefs and acts are consistent because acts are guided by
beliefs. Festinger reverses this schema: circumstances cause acts and then
beliefs are made consistent with acts by rationalizations. This theory is
heuristic: assuming that (a): people need consistency between their behav-
iours and their beliefs and (b): this consistency can be reached by adjust-
ing beliefs rather than changing behaviours, we can interpret observable
beliefs as the result of a rationalization process intended for the reduction
of dissonance.
What is the chronological order between behaviours and rationaliza-
tions? Festinger reverses the previous order but both schemas are wrong
because they are too rigid. In fact, beliefs and behaviours are built together
and their shifts are not instantaneous. They result from a process during
which beliefs sway behaviours and reciprocally behaviours affect beliefs.
Sykes and Matza defend this ‘chronological disorder’ (1957: 666) and
Becker develops the same idea in Outsiders. Traditional views promoted by
the dominant social order define drug use as a violation of basic moral
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 25

Neutralization theory and the denial of risk 25

imperatives and describe drug users as dangerous slaves. A marijuana user

must neutralize the dissonance between these stereotyped views and his
consumption by modifying his beliefs about drug and drug users: this
rationalization can be a prerequisite for use, but it is also acquired ‘in the
course of experience’ and can justify a current use a posteriori in order to
maintain it (Becker 1963: 73–8).
This chronological disorder is especially relevant to risky behaviours
because safe/risky categories change quite more quickly than legal/illegal
ones: The carcinogens list is updated daily, formerly harmless products can
become suddenly suspected of being a lethal poison, just as sexual promis-
cuity and beef eating became risky behaviours due to AIDS and Mad Cow
Disease crisis respectively. An individual already committed in a lifestyle
including sexual promiscuity (or beef eating) could not disrupt his habits
overnight. He had to work out a compromise between this new danger and
his lifestyle, in order to justify altogether his past, current and future prac-
tices. Such a compromise could imply the denial of risk by a cognitive adap-
tation. For example: sexual promiscuity is generally risky, but my own
promiscuous behaviour is harmless because in my mind’s eyes I associate
AIDS with stereotypes (homosexuals, heroin users) that don’t fit in with my
selection of sexual partners.

What Must Be Justified? Who Must Be Convinced?

Techniques of neutralization provide delinquents with motives to suspend
their loyalty to dominant values hic et nunc without renouncing them. Such
a temporarily release is possible because values are ‘qualified guides for
actions limited in their applicability in terms of time, place, persons and
social circumstances’ (Sykes and Matza 1957: 666). Therefore, these tech-
niques can easily apply to deviant acts that remain unusual but seem less
relevant to justify systematic deviant behaviours. When deviance becomes
usual as a significant part of a lifestyle, commitment to dominant social
order turns out to be doubtful and sociologists often assume an alternative
commitment to a delinquent sub-culture. Moreover, according to Sykes
and Matza, neutralization is used for protecting the individual from self-
blame but also from the blame of others. Thus neutralization theory can
attract strong criticism if it is reduced to a simplistic version: neutralizations
would just be rationalizations a posteriori (in the Freudian sense) or excuses
offered by an insincere individual after an isolated offence in order to
counter the guilt and offset the censure (Hirschi 1969; Jones 1998). Such
criticism renews with Pareto’s viewpoint: human beings are argumentative
rather than strictly rational. Prone to self-delusion, they always try to find
bad accounts (a ‘logical varnish’) to justify irrational behaviours.
As a variant of neutralization theory, risk denial theory escapes more
easily from such a criticism: on the one hand the denial of risk can support
a whole lifestyle by justifying systematic risky behaviours, on the other hand
the first and (most of the time) the only target of this justification is the
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 26

26 Patrick Peretti-Watel

individual himself. In an article attempting to make use of Mary Douglas’

cultural theory of risk, Bellaby (1990) studies two small egalitarian groups
particularly exposed to danger which developed a ‘risk-immune culture’:
slip-house and kiln workers in a pottery factory and hardened bikers. Both
groups valued early exposure to risk because they believed it empowered
them to overcome subsequent hazards, as a sort of vaccination conferring
immunity to accidents. This denial of risk played a central part in their life
since it supported their leading daily activity. Bikers and workers didn’t
have to convince any authority and didn’t try to avoid any blame. First of
all they had to convince themselves that their daily activity was not so risky
in order to reduce their anxiety: the denial of risk commands sincerity.
Goffman (1965) already developed this idea in Relations in Public.
Dealing with ‘normal appearances’, he used an analogy between man and
animal. Both exhibit two basic modes of activity: the quiet and the alarm.
Alarm is a tiring state, only necessary when a close danger needs a fast
reaction. This is the reason why people and animals learn to adapt them-
selves to their environment by switching from unflagging attention to
routine and placidity. When routine is disrupted by something unusual,
some adjustments are necessary to restore it (Goffman 1965: 239–40) but
these adjustments can be cognitive rather than behavioural, by acquisition
of a ‘reassuring information’ equivalent to a denial of risk
When the subject senses that something is up, his attention and concern
are mobilized; adaptive behaviour occurs if the alarm proves ‘real’, but
if reassuring information is acquired, the alarm proving false, his concen-
tration will decay quickly. . . . Here, then, is a standard for measuring
presence of mind, one involving the ability to come up quickly with the
kind of accountings that allow a disturbing event to be assimilated to the
normal; and we have a base against which to perceive bad accounts,
namely, ones that do not dispel the observer’s suspicion. (Goffman 1965:
262–3)
Thus, an individual alarmed by the irruption of a risk into his environ-
ment can deny it thanks to reassuring information, but only if this infor-
mation convinces him. An efficient denial of risk doesn’t tolerate bad
accounts and bad faith but needs convincing arguments and sincerity. This
is the reason why Pareto’s viewpoint should be completed: human beings
are not strictly rational, they are argumentative but also reasonable. They
are endowed with a ‘cognitive rationality’ (Boudon 1993): they endorse or
reject a belief rather than another because they have strong cognitive
reasons for seeing it as valid (but of course they could be wrong).1

Techniques of Neutralization, Types of Risk Denial

Before scrutinizing these ‘strong reasons’, let’s recall quickly the five well-
known techniques of neutralization distinguished by Sykes and Matza.
Psychologists usually explain risk denial by invoking an ‘unrealistic
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 27

Neutralization theory and the denial of risk 27

TABLE I: The five techniques of neutralization (Matza Sykes 1953)

Denial of responsibility Approaching a ‘billiard ball’ conception of himself, the
delinquent asserts that he is victim of circumstances into
which he is pushed or pulled by forces beyond his control
(unhappy childhood, peers pressure, drunkenness . . . ).

Denial of injury The delinquent supposes that his acts don’t cause any real
harm or injury (because the victim can afford the damage,
because a theft can be viewed as a borrowing, or because a
gang fighting is a private problem).

Denial of the victim The delinquent considers that his act is rightful since the
victim is physically absent, unknown or abstract, or since the
victim deserves a punishment (as member of a stigmatized
minority or because of past misdeeds).

Condemnation of the condemners The delinquent counter attacks by accusing the very ones
who accuse him, focusing on their motives and behaviours.

Appeal to higher loyalties The delinquent justifies a temporary violation of the laws
because he had to bow to a stronger commitment (family,
friendship . . . ).

optimism’ (Weinstein 1980; McKenna 1993): Any individual confronted

with a danger overestimates his own odds to master it. Several other biases
have been listed to explain wrong perceptions of risk (Peretti-Watel 2000).
But following Max Weber’s advice, sociologists should always try to explain
human behaviours by rational motives before letting psychologists find
irrational explanations. Risk denials can be sorted into three main types.
Each type reflects a different kind of arguments that are comprehensible
in the Weberian sense because they are supported by ‘strong reasons’.

Scapegoating This is the oldest way to deny a risk: to draw a border between
the stereotyped ‘them’ (risky people) and ‘us’ (safe people). As indicated
before, the ‘risky’ label should not be reduced to the obvious and morally
neutral result from risk profiling. ‘Them’ often refers to a deviant minority
because such a stigmatization is easier and more meaningful: it makes
possible an attribution of responsibility that transforms risk into blame
(Douglas 1992).2 For example, at the beginning of the 1980s AIDS was first
linked with the ‘four H’s’: Haiti, Hemophilia, Heroin and Homosexuality.
Only heroin users and homosexuals failed to suppress this stigma because
they were too weak to lobby efficiently against stigmatization and not
enough of them conform to dominant values to avoid an attribution of
responsibility (AIDS as a castigation for their ‘immoral’ way of life). For a
given risk, this labelling is neither exclusive nor consensual: young drivers
believe that the older ones are overly cautious and too slow to react, thus
more exposed to road accidents; but according to older drivers the younger
ones provoke many accidents because they are too aggressive on the road
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 28

28 Patrick Peretti-Watel

(Nelson et al. 1992). Both opinions are contradictory, even if supported by

valuable arguments.

Self-confidence Rather than drawing a boundary between ‘us’ and a stereo-

typed ‘them’, an individual can deny a risk by distinguishing himself from
anonymous ‘others’ because he trusts his personal ability to avoid or to
control risky situations. Self-confidence is congruent to the contemporary
trend towards individuality and self-determination emphasized by Giddens
(1991) and Beck (1992): today the self is conceived of as something
consciously constructed and reflexively monitored in relation to con-
straints and opportunities provided by its social environment. This trend is
strengthened among car drivers: many of them conceive driving as an
activity requiring several skills, including self-control, mastery of the vehicle
and an acute ability to perceive quickly where and when the Highway Code
must be conformed to – or infringed (Peretti-Watel 2001a). By driving daily
they convince themselves that they possess such abilities which make them
superior to the abstract and anonymous ‘average driver’: among English
drivers 78 per cent think that they drive better than the average motorist
(RAC 2000). In such a case the statistical measure of a risk is no longer
relevant to the individual: he considers himself as less exposed to road acci-
dents than the ‘average man’.3 This denial of risk is illustrated by Tom
Wolfe’s heroes in The Right Stuff (1980)
One theorem was: there are no accidents and no fatal flaws in the
machines; there are only pilots with the wrong stuff. (i.e., blind fate can’t
kill me) . . . Once the theorem and the corollary were understood, the
Navy’s Statistics about one in every four Navy aviators dying meant
nothing. The figures were averages, and averages applied to those with
average stuff. (Wolfe 1980: 27–8)
Comparison between risks A third way to deny a risk consists in comparing it
to similar risks that are already well-accepted by most people. This type of
denial is the least efficient. Firstly, rather than reducing the anxiety gener-
ated by a risk, such a denial drowns it in more anxiety. Secondly, compari-
sons between risks are often used by companies which are responsible for
a specific risk as a means of trying to make it acceptable, but they remain
meaningless for most people.4 At the individual level, this type of risk denial
reveals a rather fatalistic mind. By interviewing workers exposed daily to the
risk of lethal irradiation in an nuclear reprocessing plant, Zonabend (1989)
shows that they sometimes tried to get rid of their anxiety by underlining
the fact that they took higher risks just by driving from their home to the
plant, by smoking cigarettes or by drinking alcohol.
These types of risk denial techniques can cross with one another. Firstly,
individuals can pass from one type to another. For example, the dogma of
predestination made Protestants very anxious: they risked damnation, but
they had no way to solve the uncertainty concerning their status in the next
world (Weber 1976). A life beyond reproach and material success were
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 29

Neutralization theory and the denial of risk 29

believed to be the appropriate means to acquire the certitudo salutis, to be

convinced of belonging to the Few Chosen Ones among the crowd to
escape damnation (i.e. self-confidence). But as soon as Protestants gathered
in sects, they drew a line between ‘them’, as the Elect, and ‘others’, the
damned (i.e. scapegoating). Secondly, an individual can simultaneously use
several types of risk denial. French workers interviewed by Zonabend
(1989) not only compared different risks, they also defined scapegoats. For
wage earners these were interim/temporary workers and for men these
were women who were thought to be more prone to accidents. In addition
each worker believed that he was less exposed to risk than others because
of the personal improvements he had introduced to the standardized safety
regulations.

NEUTRALIZATION, DENIAL AND CANNABIS USE

Neutralization, Denial and Cannabis Use

Drug use is a delinquent act without victim, thus two techniques of neutral-
ization are already offside: there is neither victim nor injury to deny. Of
course, one could consider the user as being his own victim, but as denying
the victim often implies denigrating him, in this case it would mean self-
disparagement. Concerning the denial of injury, generally speaking it is
not relevant to risk neutralization: denying a risk means denying a potential
injury, not an actual one.
According to Becker (1963: 135–7), three American values have
provided legitimacy for the prohibition of marijuana: people should never
lose their self-control nor become dependent on any drug; ecstatic experi-
ences can reward hard work or religious fervour but they should not be
pursued for their own sake; unhealthy behaviours must be prevented. In
fact, marijuana users interviewed by Becker (just as other cannabis smokers
questioned in more recent surveys) are generally aware of the first and
third values, but not of the second one (which is certainly on the decline).
The first rationalization cited in Outsiders rests on a comparison that is
relevant to the third value
One of the most common rationalizations is that conventional persons
indulge in much more harmful practices and that a comparatively minor
vice like marihuana smoking cannot really be wrong when such things as
the use of alcohol are so commonly accepted. (Becker 1963: 74)
The condemnation of the condemners is the most suitable technique of
neutralization for this rationalization: conventional people who would
blame marijuana users could be accused of consuming a more harmful
product. Nevertheless, in this case both blames remain virtual. Since the
user quoted by Becker considered marijuana as less unhealthy than
alcohol, he justified his own ‘vice’ by comparing it to a riskier one (i.e.
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 30

30 Patrick Peretti-Watel

comparison between risks). This denial of risk could also lead to scapegoating:
this user belonged to a group of musicians in which everybody smoked
marijuana but didn’t drink alcohol and disparaged drinking (they drew a
line between themselves and alcohol-drinkers). But most of the rational-
izations quoted by Becker for occasional or regular smokers deal with the
first value: the self-control imperative. Occasional smokers tried to
convince themselves that they controlled their consumption
Finally, the user, at this point [occasional use], is not using the drug all
the time. His use is scheduled . . . The existence of this schedule allows
him to assure himself that he controls the drug and becomes a symbol of
the harmlessness of the practice . . . The fact that there are times when
he does not, on principle, use the drug, can be used as proof to himself
of his freedom with respect to it. (Becker 1963: 74)
A regular user related that, worried about an eventual addiction, he had
decided to stop smoking for a whole week and achieved it easily: ever since
this success he smoked without anxiety. Another regular user, afraid that
he could be addicted to marijuana as a necessary crutch for psychological
problems, finally considered that he mastered this risk simply because he
was aware of. So occasional and regular marijuana smokers maintained self-
confidence regarding their ability to control personally their marijuana use.
They didn’t deny individual responsibility. It was just the opposite: they
asserted complete mastery of their consumption. They didn’t justify it by
appealing to higher loyalties, by underlining peer pressure: even if peers
played a key role in the process of becoming a marijuana smoker, they
conceived this use as a choice they have made freely. Thus the two remain-
ing techniques of neutralization must be turned down.

Other Qualitative Data

Several recent British qualitative studies give special emphasis to self-
confidence shown by adolescents using licit or illicit drugs. Young users
believe that they control their consumption and live it as a mastered succes-
sion of deliberate and well thought-out choices (Boys et al. 1999). They are
adamant that they have made their own drugs decisions for which they take
responsibility and while they acknowledge peer influence they strongly
refuse to explain their use by peer group pressure because such an expla-
nation wrongly portrays them as passive victims. They consider that such
pressure would be a serious encroachment on their individuality, incom-
patible with the notion of friendship (Parker et al. 1998; Denscombe 2001).
These studies underline the difference between peer pressure and peer
influence. This difference is crucial for interpreting quantitative data
because statistical analyses usually conclude that the reported number of
respondents’ friends who use drugs is a strong and reliable predictor of
cannabis use (Kandel 1980; Pearson and Michell 2000).
The relationship between personal use and use among peers is not
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 31

Neutralization theory and the denial of risk 31

reducible to a question of access to the substance since this statistical link

remains significant once controlled the effect of availability (Höfler et al.
1999; Peretti-Watel 2001b). Peers also play a key role in the acquisition and
support of rationalizations, as illustrated by some French ethnographic
studies realized in the suburbs of Paris. Self-control is an important
informal rule among groups of young cannabis users. A member can be
excluded from the group if he is not able to maintain a normal appearance
in public places even when he ‘gets high’ (Aquatias 1999), as if one loss of
control in the group could endanger self-control displayed by other
members. These cannabis users vigorously reject the stigmatized stereotype
of the drug addict (described as sniffing cocaine or taking ‘smack’ by injec-
tion). They consider their consumption as a recreational and convivial
activity, and their belonging to a peer group confirms them in this opinion.
On the other hand, they outline heroin use as a compulsive and solitary
vice, and heroin user as antisocial elements, who have lost the ability to
maintain friendly or affective relations (Coppel 1993; Duprez and Kokoreff
2000). Parker et al. (1998) found the same discourse among British youth
[in the 1990s] adolescents maintain a fairly stereotypical imaging of
‘hard drug’ users as dangerous, diseased, dishevelled injecting ‘junkies’
and ‘saddos’ who commit vast amounts of crime to fund their habit. . . .
If heroin and crack are at one end of the spectrum cannabis is at the
other. Cannabis is regarded, even by many abstainers, as a relatively safe
drug and certainly no more dangerous than alcohol and tobacco.
(Parker et al. 1998: 132)
As well as the marijuana smokers interviewed by Becker, these adoles-
cents justify their practice, but they neither deny their responsibility nor do
they appeal to higher loyalties, even if they belong to a group of cannabis
users. Firstly, they think that they can control their consumption. Secondly,
they scapegoat ‘hard drug’ users in order to keep the stereotype of the drug
addict at a distance. Thus they deny the risk twice, and their peer group
support both denials. These denials are also illustrated by data from
ESCAPAD, a French quantitative survey dealing with drug use among
adolescents. Respondents were invited to add a free comment at the end of
the self-report questionnaire. About three hundred respondents spon-
taneously stood up for cannabis use.5 Of these, 65.1 per cent were regular
cannabis users (at least ten times during the last 30 days), versus 14.1 per
cent in the whole sample. These comments illustrated the three types of
risk denial. Firstly, they often drew a sharp distinction between cannabis
and heroin. They asserted that cannabis was neither an addictive substance
nor a gateway to ‘hard drugs’ and they rejected the stereotype of the drug
addict, claiming that despite cannabis use they had a normal life, with good
family relationships and good marks at school. Secondly, they emphasized
their own ability to control their use of cannabis, even if sometimes they
knew addicts around them (Beck et al. 2000: 181–3). Thirdly, they
compared cannabis to alcohol and tobacco, considering that both licit
drugs were more dangerous than cannabis.
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 32

32 Patrick Peretti-Watel

CANNABIS USE AND DENIAL OF RISK: THE ESPAD SURVEY

Data Source
Data from the French part of ESPAD allow to build a partial quantitative
test for the risk denial theory. This survey took place in about thirty
European countries, from March to May 1999, based on a standardized
questionnaire focusing upon drug uses, behaviours and opinions (Hibell et
al. 2001). In France, the survey was carried out by the Health and Medical
Research National Institute (INSERM). The sampling was set up randomly
according to several standards: level and type of high school, public or
private sector and urban situation. Students individually answered the ques-
tionnaire within the classroom, in the presence of a health professional.
The main part of the sample was aged 14–19 (91.8 per cent, 10,810
students).
This survey provided three types of variables relevant to the present
article: respondents’ drugs use; opinions about people using various drugs
(with three response levels: don’t disapprove, disapprove, strongly disap-
prove); and perceptions of how much people risk harming themselves (no
risk, slight risk, moderate risk or great risk) for various uses and drugs. Five
levels of cannabis use were defined (see Table II).

Hypotheses and Methods

The ESPAD data provided no direct way to test the second type of risk
denial (self-confidence)6 but a test was workable for the two others types.
Adolescents are usually prepared cognitively for illicit substance use: they
don’t use it without first believing that benefits exceed costs (Petraitis et al.
1998). This belief must be maintained if the use is to continue: among
Australian long-term users (at least 10 years), only 7 per cent said cannabis
had done them more harm than good (Reilly et al. 1998). So we can expect
that perceptions of the risk induced by cannabis use are closely linked to
respondents’ level of use, and decrease when this level increases (hypothesis
1). Of course this hypothesis is not ‘risky’: we need a more specific one.
According to Becker, cannabis users must be emancipated from the moral
standards condemning drugs use in general, which include ‘hard drugs’
use. But for cannabis users, scapegoating ‘hard drugs’ users is a good way
to deny the risk associated with their own behaviour. With the ESPAD data,

TABLE II: Levels of cannabis use: ESPAD 1999, students, age 14–19 (%)
No use during the last year 69.9 (n = 7553)
At least 1 time during the last year but not during the last month 8.7 (n = 939)
1–2 times during the last month 7.4 (n = 795)
3–9 times during the last month 6.7 (n = 722)
10 times or more during the last month 6.4 (n = 693)
Missing values 1.0 (n = 108)
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 33

Neutralization theory and the denial of risk 33

this apparent contradiction can be overcome by distinguishing opinions

about morality from perceptions of risk. So we can expect that: the more a
respondent uses cannabis, the less he disapproves heroin, cocaine or crack
users, but the more he puts emphasis on the fact that they risk harming
themselves (hypothesis 2).
Building up a factor analysis is an easy and relevant way to test both
hypotheses by taking into account simultaneously all the variables involved:
all variables are homogeneous (ordinal opinions and perceptions), and as
we assume neither a chronological order nor a causal relationship between
risk denial and cannabis use (since we consider them as simultaneous and
co-extensive), multivariate regression is inadequate. From one factor to
another, we expect to find a strong correlation between disapproval of use
for all illicit drugs, cannabis included (Becker’s viewpoint), but also an
opposition between risk perceptions of cannabis and ‘hard drugs’ (accord-
ing to the scapegoating hypothesis). As these opinions and perceptions are
measured with ordinal variables, we can treat them in a quantitative way
and carry out a principal components analysis (with numerical values, from
1 for ‘don’t disapprove’ to 3 for ‘strongly disapprove’, and from 1 for ‘no
risk’ to 4 for ‘great risk’: other values, for example an asymmetric distri-
bution, have been tested in order to check the reliability of results). This
method provides a statistical outline of main correlations between quanti-
tative variables and allows illustrative variables to be taken into account (in
this case: levels of cannabis use). The key results of the factorial analysis are
underlined with simple bivariate graphics.
The denial of risk relying on a comparison between different risks can
also be tested. Unlike some of the musicians interviewed by Becker, French
adolescents often use both cannabis and alcohol rather than choosing
between them. In order to get rid of the ‘risky’ label put on cannabis by
common sense, they could compare both risks they take by emphasizing the
fact that drinking alcohol is more risky than cannabis use, and this
emphasis could become more pronounced when the level of cannabis use
increases. Thus we can expect that for a fixed level of alcohol consumption
(regular use: to have drunk alcohol at least 10 times during the last month)
the risk perception for drinking alcohol will be amplified for higher levels
of cannabis use (hypothesis 3).

Results
The statistical package SPSS was used to perform a principal components
analysis on the 32 active variables (16 disapproval items, 16 risk perception
items). The first three factors of the principal components analysis
accounted for 52.8 per cent of the total variance. For each of them, Table
III gives a label and factor loading for items for which the contribution to
the variance carried by a factor (CTR) was at least 4 per cent (the average
contribution is 3.1 per cent).
The 11 items retained for the first factor accounted for 65.6 per cent of
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 34

34 Patrick Peretti-Watel

TABLE III: Factor loadings for items included in principal components analysis of
disapproval and risk perceptions toward people using drugs: ESPAD 1999, students,
age 14–19
Factor
Factors and items loading CTR (%)

General disapproval toward illicit drugs (Eigenvalue = 10.2)

Disapproval of people:
. . . trying Ecstasy once or twice –0.85 7.1
. . . trying cocaine once or twice –0.84 6.9
. . . trying LSD or acid once or twice –0.83 6.8
. . . trying crack once or twice –0.83 6.8
. . . trying heroin once or twice –0.82 6.6
. . . trying amphetamine once or twice –0.82 6.6
. . . trying inhalants once or twice –0.80 6.3
. . . trying tranquilizers or sedatives without prescription once or twice –0.73 5.2
. . . smoking cannabis regularly –0.69 4.7
. . . smoking cannabis occasionally –0.67 4.4
. . . trying cannabis once or twice –0.66 4.3

Riskiness of illicit drugs, except for cannabis (Eigenvalue = 3.6)

People risk harming themselves if they:
. . . take heroin regularly –0.61 10.4
. . . take cocaine or crack regularly –0.61 10.4
. . . try cocaine or crack once or twice –0.60 10.0
. . . try heroin once or twice –0.59 9.7
. . . take Ecstasy regularly –0.57 9.0
. . . try Ecstasy once or twice –0.55 8.4
. . . take inhalants regularly –0.49 6.7
. . . try inhalants once or twice –0.48 6.4
Disapproval of people:
. . . smoking cannabis occasionally 0.43 5.1
. . . trying cannabis once or twice 0.40 4.5
. . . smoking cannabis regularly 0.39 4.2

Riskiness of cannabis vs disapproval of ‘hard drugs’ (Eigenvalue = 3.1)

People risk harming themselves if they:
. . . smoke cannabis occasionally 0.60 11.5
. . . try cannabis once or twice 0.58 10.8
. . . smoke cannabis regularly 0.56 10.0
Disapproval of people:
. . . trying heroin once or twice –0.42 5.7
. . . trying crack once or twice –0.42 5.7
. . . trying cocaine once or twice –0.42 5.7
. . . trying amphetamine once or twice –0.38 4.6

its variance. They all referred to the disapproval of illicit drugs use,
especially trying them once or twice (except for cannabis use, but with low
factor loading and CTR), and were projected on the same side of the factor
(with negative factor loading). These results reveal a general tendency to dis-
approve of illicit drug use among respondents, whatever drug is concerned:
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 35

Neutralization theory and the denial of risk 35

students who disapproved of people trying Ecstasy also tended to disapprove

of people trying cocaine, LSD, and so on. The 11 items retained for the
second factor accounted for 84.8 per cent of its variance. They showed an
opposition between, on the one hand, perceptions of risk induced by illicit
drugs use (especially heroine and cocaine or crack) excepting cannabis, and
on the other hand, disapproval toward people using cannabis. Thus this
factor distinguishes respondents who emphasized the great risk induced by
‘hard drugs’ but didn’t disapprove of cannabis use. The 7 items retained for
the third factor accounted for 53.9 per cent of its variance. They opposed
perceptions of risk induced by cannabis use with disapproval toward people
trying ‘hard drugs’ (especially heroin, crack and cocaine).
Table IV gives co-ordinates of cannabis users in the respondents’ space.
Each level of use is represented in this space by the centre of gravity i.e. an
‘average point’ which co-ordinates are the average co-ordinates computed
from corresponding respondents. In order to determine if a level of use is
significantly linked with one of the first three factors, its co-ordinate was
tested against 0. All co-ordinates given in Table IV were very significantly
different from 0 (p < 0.001), and the five levels of cannabis use were
arranged in order of intensity (from ‘no use during the last 12 months’ to
‘at least 10 times during the last 30 days’) on each factor, revealing that dis-
approval and risk perception toward drug use were closely linked with
cannabis use.
Thus, the more a respondent used cannabis:
• the less he disapproved of illicit drugs users, whatever drug was
concerned (Factor 1);
• the more he emphasized the risk induced by illicit drugs, except for
cannabis, and the less he disapproved of people using cannabis (Factor
2);
• the less he emphasized the risk induced by cannabis use, and the more
he disapproved of people trying ‘hard drugs’ once or twice (Factor 3).
These results are illustrated by Figures I and II. On the one hand (Figure
I), when the level of cannabis use increased, the proportion of respondents
strongly disapproving of people using cannabis decreased sharply, and the
proportion strongly disapproving of people trying heroin or cocaine once

TABLE IV: Co-ordinates for cannabis users on the first three factors of the principal
components analysis: ESPAD 1999, students, age 14–19
Level of use Factor 1 Factor 2 Factor 3

No use during the last 12 months (n = 7553) –0.66 0.43 0.57

During the last 30 days . . .
No use (n = 939) 0.55 –0.57 –0.70
1–2 times (n = 795) 1.24 –0.73 –1.13
3–9 times (n = 722) 1.62 –1.04 –1.42
10 times or more (n = 693) 2.46 –1.19 –1.50
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 36

36 Patrick Peretti-Watel

or twice also decreased (p < 0.001 for both curves). On the other hand
(Figure II), when the level of cannabis use increased, the proportion of
respondents thinking that people using cannabis take a great risk showed
a marked decline, but the proportion who think that people trying heroin
or cocaine/crack once or twice take a great risk rose (p < 0.001 for both
curves).
The third hypothesis can be tested by looking at the relationship between
cannabis use and risk perception for alcohol drinking, for a constant level
of alcohol consumption. Figure III shows this relationship by distinguish-
ing regular drinkers from others. Among regular drinkers, the proportion
who emphasized the risk induced by heavy drinking (4 or 5 drinks nearly
every day) increased with the level of cannabis use (p = 0.009); but among
other respondents this proportion remained constant whatever the level of
cannabis use (p = 0.873).

Discussion
As expected by common sense, ESPAD data confirm that the more
teenagers use cannabis, the less they condemn this behaviour or consider
it as risky. A general tendency to disapprove of people using illicit drugs is
negatively correlated with cannabis use: as suggested by Becker, cannabis
users were emancipated from the moral standards condemning drug use,
and thus were more tolerant toward drug users in general. But our results

60%

KEY
50%
trying once or
twice cannabis
40%
occasional use
of cannabis
30%
regular use of
cannabis
20%
trying once or
twice heroin
10%
trying once or
twice cocaine
0%
0/last 0/last 1-2/last 3-9/last >9/last
year month month month month

FIGURE I: Proportion of respondents who strongly disapprove of people using

various drugs, for different levels of cannabis use
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 37

Neutralization theory and the denial of risk 37

80%

KEY
70%
trying once or
twice cannabis
60%
occasional use
of cannabis
50%
regular use of
cannabis
40%
trying once or
twice heroin
30%
trying once or
twice
20%
cocaine/crack

10%

0%
0/last 0/last 1-2/last 3-9/last >9/last
year month month month month

FIGURE II: Proportion of respondents who consider that people using various drugs
take a great risk, for different levels of cannabis use

also provide empirical evidence for two hypotheses specific to the risk
denial theory. Firstly, cannabis users considered heroin, cocaine or crack
use as more risky than ‘non-users’ (respondents who don’t use cannabis
during the last year) did, and this perception increased with the level of
cannabis use: cannabis users kept the stereotype of the drug addict at a
distance by scapegoating ‘hard drugs’ users. Secondly, among regular users
of alcohol, the risk perception for heavy drinking increased with the level
of cannabis use, whereas this perception didn’t vary with cannabis use
among non-regular drinkers. Cannabis users got rid of the ‘risky’ label
usually put on this drug by comparing cannabis and alcohol: the more they
consumed cannabis, the more they emphasized the risk induced by
drinking alcohol.
With ESPAD data, such a denial of risk was not observed for cigarette
smoking, the damage to health caused by which is frequently underlined by
the French media: adolescents probably found it more relevant to compare
cannabis to alcohol (rather than to tobacco), because both substances have
strong psycho-active effects and can be considered as substitutable (or
complementary) ways to ‘get high’. The ESPAD questionnaire didn’t
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 38

38 Patrick Peretti-Watel

70%

60%

50%

40%

30%

KEY
20%
regular drinkers

10%
other
respondents
0%
0/last 0/last 1-2/last 3-9/last >9/last
year month month month month

FIGURE III: Proportion of regular drinkers and other respondents who consider that
people having 4 or 5 drinks nearly every day take a great risk, for different levels of
cannabis use

include items to test for the second type of risk denial (self-confidence), but
this type provides an explanation to one result represented on Figure I:
among respondents who have used cannabis at least 10 times during the last
month, about 10 per cent believed that people smoking cannabis regularly
take a great risk. How can this 10 per cent be explained? Perhaps the respon-
dents didn’t consider themselves as regular users, or perhaps they sought
risk consciously. However, we can assume that they differentiated between
the risk for ‘people’ in general and the risk for themselves because they
trusted their own ability to master their consumption.7

CONCLUSION

Risk denial theory refers to a cognitive way to deal with risky behaviours,
which can justify a whole lifestyle and must convince the individual himself
before anybody else. In contemporary societies where risk culture and risk
profiling attach a social stigma to unhealthy behaviours, this theory
provides tools to understand how people neutralize the ‘risky’ label by
using specific techniques, distinct from the five ones previously listed by
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 39

Neutralization theory and the denial of risk 39

Sykes and Matza. As an updated variant of neutralization theory, risk denial

theory improves the former with additional and complementary tech-
niques of neutralization. Cannabis use provided a good opportunity to
illustrate these techniques, thanks to Becker’s interviews, later qualitative
studies and French quantitative data from the 1999 ESPAD survey. In order
to deny the risk induced by cannabis use, people (and especially adoles-
cents) scapegoat ‘hard drugs’ users, emphasize their ability to control it
personally or compare cannabis to alcohol. Other ethnographic studies
reveal a similar risk denial among Ecstasy and amphetamines users (Power
et al. 1996; Klee 1998; McElrath and McEvoy 2001).
Beyond drug use, the risk denial theory can be extended to many other
research fields, including road driving, sexual activity or the workplace. A
single assumption sums up this theory: in general, people neither seek risk
purposely nor endanger themselves unconsciously. They just find ‘good
reasons’ to deny it. Of course, they can be wrong: considering oneself as a
good car driver doesn’t prevent one from crashing. But taking this denial
of risk into consideration is a first necessary step in order to improve
prevention. Like Sykes and Matza’s techniques of neutralization, risk denial
types can be considered as learnt cognitive skills. Their acquisition
probably depends on various social factors such as gender, socio-economic
status or socialization among peers. Subsequent analyses should focus on
these influences, in order to understand how people learn to deny a risk
and why some of them are unable to do so. Finally, if the denial of risk
implies convincing oneself before anybody else, as a mental process this
soliloquy is deeply shaped by scientists, moral entrepreneurs and legis-
lators’ discourses on ‘risky behaviours’. Thus, sociologists should also
broaden the scope of their enquiries by studying the social construction of
the ‘risky’ label, just as they did with the ‘deviant’ one.

(Date accepted: October 2002) Patrick Peretti-Watel

Regional Centre for Disease Control of South-eastern France
INSERM Research Unit 379

ACKOWLEDGEMENTS

*Grateful thanks to the anonymous reviewers for their helpful suggestions

and comments on an earlier draft of this paper. Thanks also to François
Beck, Emily Hart and Brian South Pole.

NOTES actor himself or to the observer an ade-

quate ground for the conduct in question.
1. Boudon’s ‘good reason’ corresponds The interpretation of a coherent course of
to Weber’s ‘motive’: ‘A motive is a complex conduct is “subjectively adequate” (or
of subjective meaning which seems to the “adequate in the level of meaning”) in so
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 40
40
far as, according to our habitual modes of
thought and feeling, its component parts
taken in their mutual relations are recog-
nized to constitute a “typical” complex of
meaning. It is more common to say
correct’ (Weber 1978: 11).
2. There are scapegoats in the Bible (a
goat is symbolically burdened with the sins
of a human community and sent to death
in the desert) and in Greek mythology
(physical and moral stigmata made a
perfect scapegoat of Œdipe for the Plague
ravaging Thebes: he was a lame foreigner,
incestuous and a parricide).
3. This denial cannot be reduced to a
whimsical reasoning. It finds a strong
support from Durkheim (1995), who criti-
cized Quételet’s theor y of the ‘average
man’ (defined by statistical means) as far as
it led to a statistical fatalism denying free
will. According to Durkheim (op cit.:
340–2, 368n), the incidence of suicides
measured for a whole society is not
relevant to estimate the risk of suicide at
the individual level, and his reasoning
could be invoked for car drivers.
4. In 1996 Philip Morris published in
newspapers statistical evaluation of several
risk factors for cancer: the relative risk
induced by passive smoking was estimated
at 1.19 (i.e. the risk of cancer is 1.19 times
as big among passive smokers compared to
non-smokers), against 1.62 for drinking
whole milk daily or 6.14 for a high satu-
rated-fat diet. The argument was clear:
passive smoking is not so risky and should
be accepted, since many people are
already taking a greater risk by their eating
habits. Of course this argument misses a
crucial point: ordinar y people make a
strong difference between a pleasure they
freely choose and a pollution they are pas-
sively subjected to.
5. I refer to the first wave (2000) of the
ESCAPAD annual survey. Methodology
and results are available in English at:
http://www.drogues.gouv.uk/index.html.
Out of 14,469 questionnaires, 1,952
include a free comment, 333 concerning
cannabis. See Beck et al. (2000) for com-
plete results.
6. Nevertheless, the questionnaire pro-
vided Rosenberg Scale items to compute a
self-esteem index (Rosenberg et al. 1995).
If self-confidence and self-esteem are
Patrick Peretti-Watel
connected, self-esteem should increase or
remain constant with the level of cannabis
use. This assumption was confirmed for
ESPAD data (Peretti-Watel 2001b).
7. In the following comment extracted
from ESCAPAD, an adolescent considered
alcohol and cannabis as harmful for close
relatives, but not for himself, even though
he declared himself a regular user of both
drugs: ‘In my family two people are drug
addicts: my step-father (alcohol) and my
brother (cannabis). I use alcohol and
cannabis but sometimes I stop using them
for a moment, because I don’t want to
become an addict. . . . With my friends, we
smoke cannabis to sleep, to have fun
during parties, or just to relax, for example
after sport.’ (Beck et al. 2000: 182).
BIBLIOGRAPHY
Aquatias, S. 1999 ‘Cannabis: du produit
aux usages’, Sociétés contemporaines 36:
53–66.
Beck, F., Legleye, S. and Peretti-Watel, P.
2000 Regards sur la fin de l’adolescence, Paris:
OFDT.
Beck, U. 1992 Risk Society: Towards a New
Modernity, London: Sage Publications.
Becker, H. S. 1963 Outsiders, New York:
The Free Press.
—— 1970 Sociological Work, Chicago:
Aldine Publishing Company.
Bellaby, P. 1990 ‘To risk or not to risk? Uses
and limitations of Mary Douglas on risk-
acceptability for understanding health and
safety at work and road accidents’, Socio-
logical Review 38(4): 465–83.
Boudon, R. 1993 ‘Toward a synthetic
theory of rationality’ and ‘More on “good
reasons”: reply to critics’, International
Studies in the Philosophy of Science 7(1): 5–19
and 87–102.
Boys, A., Marsden, J., Fountain, J., Grif-
fiths, P., Stillwell, G. and Strang, J. 1999
‘What Influences Young People’s Use of
Drugs? A Qualitative Study of Decision-
making’, Drugs: Education, Prevention and
Policy 6(3): 373–87.
Coppel, A. 1993 ‘Les réseaux d’échanges
sexuels et de circulation d’information en
matière de sexualité dans les quartiers à
risque’ in A. Tursz, Y. Souteyrand and
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 41
Neutralization theory and the denial of risk
L. R. Salmi (eds) Adolescence et risque, Paris:
Syros.
Coughlin, S. S. 1998. ‘Scientific Paradigms
in Epidemiology and Professional Values’,
Epidemiology 9(5): 578–80.
Critchlow, B. 1983 ‘Blaming the booze: the
attribution of responsibility for drunken
behavior’, Personality and Social Psychology
Bulletin 9(4): 451–73.
Denscombe, M. 2001 ‘Peer Group
Pressure, Young People and Smoking: new
developments and policy implication’,
Drugs: education, prevention and policy 8(1):
7–32.
Donovan, J. E. and Jessor, R. 1985
‘Structure of Problem Behavior in Ado-
lescence and Young Adulthood’, Journal of
Consulting and Clinical Psychology 53(6):
890–904.
Douglas, M. 1992 Risk and Blame, London:
Routledge.
Dryfoos, J. G. 1990 Adolescents at Risk, New
York: Oxford University Press.
Duprez, D. and Kokoreff, M. 2000 Les
mondes de la drogue, Paris: Odile Jacob.
Durkheim, E. 1995. Le suicide, Paris: PUF.
Evans, R. J. 1992. ‘Epidemics and Revolu-
tions: cholera in nineteenth-centur y
Europe’ in T. Ranger and P. Slack (eds)
Epidemics and Ideas, Cambridge: Cam-
bridge University Press.
Festinger, L. 1957 A Theory of Cognitive Dis-
sonance, Stanford: Evanston, Row and
Peterson.
Giddens, A. 1991 Modernity and Self-Identity,
Stanford: Stanford University Press.
Goffman, E. 1965 Relations in Public, New
York: Basic Books Publishers.
Hibell, B., Andersson, B., Ahlström, S.,
Balakireva, O., Bjarnason, T., Kokkevi, A.
and Morgan, M. 2001 The 1999 ESPAD
Report, Stockholm: Modin Tryck AB.
Hirschi, T. 1969 Causes of Delinquency,
Berkeley: University of California Press.
Höfler, M., Lieb, R., Perkonigg, A., Schus-
ter, P., Sonntag, H. and Wittchen, H. U.
1999 ‘Covariates of cannabis use pro-
gression in a representative population
sample of adolescents: a prospective
examination of vulnerability and risk
factors’, Addiction 94(11): 1679–94.
Jessor, R. and Jessor, S. L. 1977 Problem
Behavior and Psychosocial Development: A
Longitudinal Study of Youth, New York: Aca-
demic Press.
41
Jones, S. 1998 Criminology, London: Butter-
worths.
Kandel, D. B. 1980 ‘Drug and drinking
behavior among youth’, Annual Review of
Sociology 6: 235–85.
Klee, H. 1998 ‘The love of speed: an analy-
sis of the enduring attraction of amphet-
amine sulphate for British youth’, Journal
of Drug Issues 28(1): 33–56.
Leigh, B. C. 1999 ‘The risks of drinking
among young adults. Peril, chance, adven-
ture: concepts of risk, alcohol use and risky
behavior in young adults’, Addiction 94(3):
371–83.
McElrath, K. and McEvoy, K. 2001. ‘Heroin
as Evil: ecstasy users’ perceptions about
heroin’, Drugs: education, prevention and
policy 8(2): 177–189.
McKenna, F. P. 1993 ‘It won’t happen to
me: Unrealistic optimism or illusion of
control?’, British Journal of Psychology 84:
39–50.
Miles, D. R., van den Bree, M. B., Gupman,
A. E., Newlin, D. B., Glantz, M. D. and
Pickens, R. W. 2001 ‘A twin study on
sensation seeking, risk taking behavior and
marijuana use’, Drug and Alcohol Dependence
62: 57–68.
Nelson, T. M., Evelyn, B. and Taylor, R.
1992 ‘Experimental intercomparisons of
younger and older driver perceptions’,
International Journal of Aging and Human
Development 36(3): 239–53.
Osgood, D. W., Johnston, L. J., O’Malley,
P. M. and Bachman, J. G. 1988 ‘The Gen-
erality of Deviance in Late Adolescence
and Early Adulthood’, American Sociological
Review 53(1): 81–94.
Parker, H., Alridge, J. and Measham, F.
1998 Illegal Leisure: the normalization of ado-
lescent recreational drug use, London: Rout-
ledge.
Parker, R. N. and Auerhahn, K. 1998 ‘Alco-
hols, drugs and violence’, Annual Review of
Sociology 24: 291–311.
Pearson, M. and Michell, L. 2000 ‘Smoke
Rings: social network analysis of friendship
groups, smoking and drug-taking’, Drugs:
Education, Prevention and Policy 7(1): 21–38.
Peretti-Watel, P. 2000 Sociologie du risque,
Paris: Armand Colin.
—— 2001a ‘La conduite automobile: un
objet de recherche sociologique?’, Euro-
pean Journal of Sociology XLII(1): 1–38.
—— 2001b ‘Comment devient-on fumeur
03 P-Watel (jr/d) 2/11/03 2:32 PM Page 42
42
de cannabis? Une perspective quantita-
tive’, Revue française de sociologie 42(1):
3–30.
Petraitis, J., Flay, B. R., Miller, T. Q., Torpy,
E. J. and Greiner, B. 1998 ‘Illicit Substance
Use among Adolescents: A Matrix of
Prospective Predictors’, Substance Use and
and Misuse 33(13): 2561–604.
Power, R., Power, T. and Gibson, N. 1996.
‘Attitudes and experience of drug use
amongst a group of London teenagers’,
Drugs: education, prevention and policy 3(1):
71–80.
RAC (Royal Automobile Club) 2000 RAC
Report on Motoring 2000, London: RAC.
Reilly, D., Didcott, P., Swift, W. and Hall,
W. 1998 ‘Long-term cannabis use: charac-
teristics of users in an Australian rural
area’, Addiction 93(6): 837–46.
Reiss, A. J. 1961 ‘The social integration of
queers and peers’, Social Problems 9(2):
102–20.
Rockhill, B. 2001. ‘The Privatization of
Risk.’, American Journal of Public Health
91(3): 365–8.
Rosenberg, C. 1988 ‘The Definition and
Control of Disease’, Social Research 55(3):
329.
Patrick Peretti-Watel
Rosenberg, M., Schooler, C., Schoenbach,
C. and Rosenberg, F. 1995 ‘Global Self-
Esteem and Specific Self-Esteem’, American
Sociological Review 60(1): 141–56.
Short, J. F. and Strodtbeck, F. L. 1965
Group Process and Gang Delinquency,
Chicago: University of Chicago Press.
Skolbekken, J. A. 1995. ‘The risk epidemic
in medical journals’, Social Science and Medi-
cine 40(3): 291–305.
Sykes, G. M. and Matza, D. 1957 ‘Tech-
niques of Neutralization: A Theor y of
Delinquency’, American Sociological Review
22(1): 664–70.
Weber, M. 1976 The Protestant Ethic and the
Spirit of Capitalism, London: Allen and
Unwin.
—— 1978 Economy and Society, Berkeley:
University of California Press.
Weinstein, N. D. 1980 ‘Unrealistic opti-
mism about future life events’, Journal of
Personality and Social Psychology 39: 806–20.
Wolfe, T. 1980. The Right Stuff, New York:
Bantam Books.
Zonabend, F. 1989. La presqu’île au
nucléaire, Paris: Odile Jacob.
Zuckerman, M. 1979 Sensation seeking,
Hillsdale: Lawrence Erlbaum.