Adolescent risk behaviours and protective factors against peer influence

24 July 2021

AperTO - Archivio Istituzionale Open Access dell'Università di Torino

Adolescent risk behaviours and protective factors against peer influence / Elena Cattelino;Fabienne Glowacz;Michel

Born;Silvia Testa;Manuela Bina;Emanuela Calandri. In: JOURNAL OF ADOLESCENCE. ISSN 01401971.

-37:8(2014), pp. 1353-1362.

Original Citation:

Adolescent risk behaviours and protective factors against peer influence

Published version:

DOI:10.1016/j.adolescence.2014.09.013

Terms of use:

Open Access

(Article begins on next page)

Anyone can freely access the full text of works made available as "Open Access". Works made available under a

Creative Commons license can be used according to the terms and conditions of said license. Use of all other works

requires consent of the right holder (author or publisher) if not exempted from copyright protection by the applicable law.

Availability:

This is the author's manuscript

This full text was downloaded from iris - AperTO: https://iris.unito.it/

iris - AperTO

University of Turin’s Institutional Research Information System and Open Access Institutional Repository

This Accepted Author Manuscript (AAM) is copyrighted and published by Elsevier. It is

posted here by agreement between Elsevier and the University of Turin. Changes resulting

from the publishing process - such as editing, corrections, structural formatting, and other

quality control mechanisms - may not be reflected in this version of the text. The definitive

version of the text was subsequently published in JOURNAL OF ADOLESCENCE, 37

(8), 2014, 10.1016/j.adolescence.2014.09.013.

You may download, copy and otherwise use the AAM for non-commercial purposes

provided that your license is limited by the following restrictions:

(1) You may use this AAM for non-commercial purposes only under the terms of the

CC-BY-NC-ND license.

(2) The integrity of the work and identification of the author, copyright owner, and

publisher must be preserved in any copy.

(3) You must attribute this AAM in the following format: Creative Commons BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/deed.en),

10.1016/j.adolescence.2014.09.013

The publisher's version is available at:

http://linkinghub.elsevier.com/retrieve/pii/S0140197114001705

When citing, please refer to the published version.

Link to this full text:

Adolescent risk behaviours and protective factors against peer

influence

Elena Cattelino

Fabienne Glow acz

M ichel Born , Silvia Testa ,

M anuela Bina

Emanuela Calandri

This stu dy ex a m in e d th e relationships b e tw e e n p ro te c tiv e factors and in v o lv e m e n t in risl< b e h a viou r o f Italian a dolescen ts w ith friends in v o lv e d in risl<. P ro tec tiv e factors w e r e d ra w n fro m m od e ls o f p eers and fro m ind ividu a l sl<ills (p e rc e iv e d regu la tory self-efficacy, in tolera n t attitu des abou t d e v ia n c e ) and o rien ta tio n (t o health, school, re lig io n ). T h e data are fro m t w o w aves, 1 y e a r apart, o f a qu estion n a ire su rvey o f adolescents in n o rth w estern Italy. Participants w e r e 908 a d olescen ts (42% b o ys ) ages 14—16 years. Results o f a hierar­ chical regression rev ea led that re lig io s ity is a p ro te c tiv e fa ctor and that friends' m od e ls fo r c o n ve n tio n a l b eh aviou rs and p o s itiv e a ttitu d e abou t health can m itig a te th e in flu en ce o f d e via n t friends on a d o le sce n t risk b e h a vio u r 1 y e a r later, e v e n a fter c o n tro llin g fo r p rio r levels o f risk behaviou r. Possible im p lication s o f this stu dy su ggest th e im p ortan ce o f im p le m e n tin g p re v e n tiv e in terve n tion s b y in v o lv in g th e p e e r group, e s p e cia lly at abou t 16 years, and w o rk in g w ith h ete ro ge n eo u s (d e v ia n t and n o n d ev ia n t) groups.

Involvem en t in various types o f risk behaviours increases during adolescence. Research has shown that adolescent risk behaviours vary w id e ly and include, am ong others, deviant behaviour, violence, sexual behaviour, drug use, cigarette smoking, alcohol abuse, and risl<y driving. These behaviours are often exam ined individually. Nevertheless, although th ey differ in h o w th ey are carried out and in their consequences, th ey are strongly related and are linked to com m on problem s that are characteristic o f adolescence. In fact, it has been shown that different behaviours m ay perform similar functions, such as acceptance by the peer group, feelin g like an adult, and asserting their ow n identity (M offitt, 1993; Silbereisen, Eyferth, & Rudinger, 1986). Researchers w h o have investigated specific risk behaviours and those w h o have exam ined cum ulative risk indices have sought to identify possible risk and protective factors in different contexts. One o f the most im portant risk factors is the influence o f deviant peers.

During adolescence, peer influence increases m arkedly and becom es a significant alternative to parental expectations. Friends perform im portant functions in the g row th process and can be a source o f support and w ell-being. Conversely, th ey can becom e m odels o f deviance and sources o f stress and discom fort (Hartup & Stevens, 1997; Rubin, Bukowsl<y, & Parker, 2006).

D evelopm ental theories suggest that affiliation w ith deviant peers and susceptibility to peer influence are notable con­ tributors to adolescent involvem ent in risk behaviour (D odge, Dishion, & Lansford, 2006; Monahan, Steinberg, & Cauffman, 2009; Wanner, Vitaro, Carbonneau, & Tremblay, 2009). M any studies have established that adolescents’ association w ith deviant peers is a m ajor factor in the g ro w th o f deviancy (D odge, Coie, & Lynam, 2006) and in other problem behaviours (Dishion, Ha, & Véronneau, 2012; Keijsers et al., 2012).

Much has been discussed about the similarities b etw een adolescents and their friends w ith respect to tw o main processes operating in sequential and com plem entary ways: selection and socialization (D odge, Coie, et al., 2006, Dodge, Dishion, et al., 2006; Urberg, Luo, Pilgrim, & Degirm encioglu, 2003). Some studies have shown that im itative and socialization mechanisms and processes contribute to adolescents’ tendency to assume the behavioural patterns o f their friends, y et other studies have revealed that, in the choice o f friends, selection processes are at w ork during w h ich the adolescent tends to build and attend to relationships w ith peers w h o have characteristics similar to theirs (e.g., Engels, Knibbe, Drop, & de Haan, 1997; Ennett & Bauman, 1994; Kandel, 1978). In m iddle adolescence both selection and socialization influence youths’ sim ilarity w ith respect to antisocial behaviour (M onahan et al., 2009).

The link b etw een affiliation w ith deviant peers and adolescent problem behaviour is w e ll established, especially the link b etw een peer contagion and delinquent behaviour (e.g., Dishion, Dodge, & Lansford, 2008; Fergusson, Vitaro, Wanner, & Brendgen, 2007; Mathys, Vitaro, & Born, 2014) and b etw een peer influence and the use o f tobacco, marijuana, and alcohol (e.g., Larsen, Engels, Souren, Granic, & Overbeek, 2009; de Leeuw, Engels, Vermulst, & Schölte, 2009). In particular, deviant peer influences appear to be stronger for early and m iddle adolescents, for boys, and for those youths w h o are exposed to peers w h o are slightly m ore deviant than th ey are, as w e ll as in unstructured, unsupervised settings (Dishion et al., 2008). There remains a need for in-depth exam ination o f specific protective factors, especially in the context o f countries other than the United States.

Protective factors

Interest in protective factors em erged initially from studies o f developm ental psychopathology (Rutter, 1987). Protective factors are considered to be independent variables that can have their o w n direct effects on behaviour but, in addition, can m oderate the relation b etw een risk factors and behaviour (Fergusson et al., 2007; Jessor, Van Den Bos, Vanderryn, Costa, & Turbin, 1995). Jessor expanded on problem behaviour th eory to describe the relationship b etw een psychosocial protective factors and risk factors and in volvem ent in problem behaviour. The theoretical m odel consists o f three types o f protection (m odels, controls, and support) and three types o f risk (m odels, opportunity, and vulnerability). Regarding protection, models includes measures o f models, such as friends’ in volvem ent in com m unity groups and volu n teer work; controls includes individual-level measures o f control, such as attitudinal intolerance o f deviance; and support includes measures o f contextual supports, such as fam ily closeness. W ith regard to risk, models includes measures o f models, such as peers’ alcohol use; opportunity includes opportunity measures, such as availability o f alcohol in the hom e; and vulnerability includes measures o f personal vulnerability, such as perceived stress and lo w self-esteem (Jessor et al., 2003). Similar protective and risk factors have been exam ined in several other investigations o f adolescent risk behaviour (e.g., Felix-Ortiz & N ew com b, 1992; Fergusson et al., 2007).

M ost w ork w ith respect to adolescent behaviour and d evelopm en t has been confined to W estern, especially North American, populations. Less numerous are studies carried out in European countries, and rarer y et are those concerning the Italian context. As far as risk behaviour is concerned, in Italy, as in other European countries, adolescence is a crucial period for experim entation w ith behaviours that put p eop le’s health and w ell-b ein g at risk. National statistics (ISTAT, 2013) reveal that alcohol consum ption and cigarette smoking are w idespread, w ith boys and girls show ing no m ajor statistical differences in usage, and these behaviours usually occur in social settings w ith friends. Cannabis is the m ost com m only used drug, w ith initiation m ost often occurring b etw een ages 15 and 17 years. The rate o f ju ven ile delinquency is som ew hat low, and boys are m ore often im plicated than are girls. Theft is the crim e m ost frequently com m itted by young people, particularly in a group situation, although the phenom enon o f crim inal gangs is lim ited. Although the number o f autom obile-related deaths is progressively decreasing, road accidents rem ain the main cause o f death am ong adolescents and youn g peop le b etw een ages 14 and 24 years. For those ages 14 to 17, m otorcycle accidents are m ost com m on. The causes are m ost often associated w ith lack o f respect for the traffic code, and lack o f attention to safety increases the seriousness o f the effects. Nonuse o f safety systems and risl<y drivin g increase w h en adolescents are in the com pany o f their peers (Bonino, & Cattelino, 2012; ISTAT, 2013).

Progressing from Jessor’s problem behaviour th eory and its reform ulation and extension w ith regard to the protective-risk m odel (Jessor et al., 2003), Bonino, Cattelino, and Ciairano (20 05) studied the function o f risk behaviours and risk and pro­ tective factors in the Italian context; that is, seven risk behaviours (cigarette smoking, alcohol consumption, marijuana and other drug use, risl<y driving, antisocial behaviour, risl<y sexual behaviour, disturbed eating) and their relation to the per­ sonality system (values, attitudes, expectations, self-perception), as w e ll as three main social contexts in w hich adolescents are em bedded (fam ily, peers, school). W ith respect to this explanatory, theory-based, psychosocial m odel used in various cultural contexts, such as the United States and the People’s Republic o f China (Costa et al., 2005; Jessor et al., 2003), the main risk and protective factors for Italian adolescents’ risk behaviours w ere found in a cross-sectional design. Individual risk behaviours (cigarette smoking, alcohol consumption, marijuana and other drug use, risl<y driving, antisocial behaviour, sexual behaviour, and disturbed eating) w e re analysed separately, and som e risk and protective factors w ere found to exert a

transversal effect; in particular, it was found that for Italian adolescents a link exists b etw een various adolescent problem behaviours and association w ith deviant peers in m iddle adolescence (Bonino et al., 2005; Cattelino, 2010). M oreover, many variables, drawn from Jessor’s m odel (pertaining to both individual level and social con text) and integrated by measures o f self-efficacy (Bandura, 1997), w e re found to play a protective role against risk behaviours.

Our study focused on six protective factors selected on the basis o f the theoretical m odel and on the basis o f the results o f previous cross-sectional studies (Cattelino, Calandri, Bina, & Graziano, 2011): friends’ m odels o f conventional behaviour (i.e., involvem ent in various conventional organizations and prosocial pastimes, good grades at school, regular church attendance), positive orientation to health, positive orientation to school, religiosity, intolerant attitudes about deviance, and regulatory self-efficacy. Friends’ m odels o f conventional behaviour constitute protection because th ey represent interpersonal control in the face o f deviance and th ey can m oderate the im pact o f deviant peers.

Having a positive orientation tow ard health constitutes protection against in volvem ent in risk behaviours that can be dam aging to health, such as substance use or risl<y drivin g practices, because it indicates the personal im portance o f health. It also reflects a com m itm ent to behaviours that prom ote healthful outcom es and that discourage health-com prom ising b e­ haviours (Bonino et al., 2005; Jessor, Turbin, & Costa, 1998; Turbin et al., 2006).

A positive orientation toward school constitutes protection because it represents positive engagem ent w ith a conventional social institution and com m itm ent. Adolescents w ith a positive orientation to school can m ore successfully create a strong personal identity through dedication and com m itm ent to the academ ic institution and do not feel the need to seek out alternative or superficial ways o f affirm ing adulthood or feelin g successful.

Religiosity constitutes protection because it em phasizes the values o f universalism, benevolence, tradition, and con form ity (Caprara, Scabini, Steca, & Schwartz, 2011; Schwartz, 1992). M oreover, religiosity reflects a com m itm ent to conventional values and disapproval o f n orm -violative activities and serves as a personal control against in volvem ent in nonnorm ative behaviours (Jessor et al., 1998).

Intolerance o f deviance constitutes protection because it reflects a com m itm ent to conventional values and disapproval o f antisocial behaviour and norm -violative activities. Adolescents w ith high levels o f deviance disapproval strongly iden tify w ith the values shared by society (Pauwels, Vettenburg, Gavray, & Brondeel, 2011).

Individuals’ perceptions o f their self-regulatory abilities (Bandura, 1997), com m only called self-efficacy beliefs, serve as a regulatory function in all m ajor transitions o f life, and th ey are especially im portant during adolescence. In particular, reg ­ ulatory self-efficacy constitutes protection because it increases the likelihood that adolescents w ill be able to resist peer pressure.

This study

M any previous studies in this area o f research have been conducted, especially in the United States. This article provides inform ation about protective factors against problem behaviours that is specific to populations attending Italian secondary schools. The prim ary aim o f this longitudinal study was to exam ine specific protective factors that m ight m oderate the link b etw een deviant peers and adolescent risk behaviours in the Italian setting that have not received much focus in prior research. Protective factors w ere identified in the fram ew ork o f Jessor’s problem behaviour th eory extended by Bonino et al. (2005). A cum ulative risk index, including cigarette smoking, illicit drug use, problem drinking, deviant behaviour, and risl<y driving practices, was used as dependent variable.

Three key questions w e re addressed in this study:

1. Do friends’ m odels o f conventional behaviour, positive orientation to health, positive orientation to school, religiosity, intolerant attitudes toward deviance, and regulatory self-efficacy constitute factors that protect Italian adolescents against m ultiple risk behaviours?

2. Do these same variables serve as protective factors that buffer adolescents from negative influences o f peer deviance in a longitudinal m odel?

3. Do protective factors m odify the influence o f peer deviance in the same w a y am ong boys and girls and am ong youn ger and o lder adolescents?

W e tested the hypothesis that all these variables maintain a protective effect against adolescent risk behaviour ov er tim e and that th ey also have a m oderating effect by buffering adolescents from negative influences o f deviant peers. Finally, w e tested the hypothesis that the same m oderating effects account for risk behaviour involvem ent across gen der and age.

M ethod

Study design and procedure

The data reported in this article are from a longitudinal questionnaire o f lifestyle and health-related behaviour adm in­ istered to adolescents in northw estern Italy. The questionnaire lo, la scuola e il m io stile di vita (School, my life-style and me; Cattelino, Begotti, & Bonino, 1999), w hich includes various measures o f perceived self-efficacy (Bandura, 1997; adapted by

Caprara, 2001), in volvem ent in various risk behaviours, and internalizing problem s (Jessor, 1992; adapted b y Bonino et al., 2005), along w ith questions and scales related to personal variables and characteristics o f the main life contexts o f adoles­ cents, was used to collect the data. A ll aspects investigated by the questionnaire w ere based on self-report measures, and all the variables related to matters perceived by adolescents. The questionnaire was adm inistered twice, 1 year apart.

Participants com pleted an anonym ous self-report questionnaire adm inistered by trained researchers in the schools during classroom tim e w h en teachers w ere not present. Com pleted questionnaires w e re turned in im m ediately to researchers. A ctive parental and personal consent w ere required before the questionnaire was administered, in accordance w ith Italian law and the ethical code o f the Professional Psychologists Association.

Participants

The sample was draw n from five high schools chosen in collaboration w ith the school district adm inistration o f Pied­ m ont and Aosta Valley, tw o regions in northwestern Italy, to best represent variation in the socioeconom ic backgrounds o f the students. In each o f the five schools selected, 15 classes w ere random ly sam pled to participate in the study. The initial sample consisted o f 908 adolescents, 42% o f w h o m w ere boys, ages 14-16 (M = 14.9 years; SD = .79) attending Grades 9,10, and 11 o f different types o f secondary schools (lyceum , technical, and professional schools). The second adm inistration o f the questionnaire involved 845 adolescents; attrition was 6.9%. The longitudinal sample was distributed similarly by gender, age, and type o f school attended to that o f the initial sample; no significant differences b etw een those w h o participated on ly in the first survey and the rem aining sample w ere found w ith respect to level o f involvem ent in individual risk, the risk im plication o f friends, and all m oderators considered. In other words, the longitudinal sample did not differ from the initial sample.

Variables and analytic procedure

Friends’ m odels o f conventional behaviour (7 items; a = .65) w ere measured by using Jessor’s scale (Bonino et al., 2005; Jessor, 1992) concerning the perceived proportion o f friends w h o w e re in school clubs or organizations, attended church regularly, participated in com m u nity youth groups, earned good grades, volunteered at school or in the community, participated in organized sports, and spent tim e w ith their families. Possible answers w ere based on a 4-point Likert scale in w h ich 1 = none o f them, 2 = some o f them, 3 = most o f them, and 4 = all o f them.

Protective factors referring to individual-level protection w ere measured by four variables: positive orientation to health, positive orientation to school, religiosity, and attitudinal intolerance o f deviance. Positive orientation to health (nin e items; a = .70) was measured by using Jessor’s scale (Bonino et al., 2005; Jessor, 1992) concerning h o w im portant various health outcom es are to the respondent, such as “to feel in good shape,” “to keep yourself healthy even i f it takes extra effort,” “to have good health habits about eating,” “to practice physical exercise regularly.” Possible answers w ere based on a 4-point Likert scale in w hich 1 = not im portant at all, 2 = not too im portant, 3 = somewhat important, and 4 = very im portan t

Positive orientation to school (fou r items; a = .72) was measured by using Jessor’s scale (Bonino et al., 2005; Jessor, 1992) concerning h o w im portant it is for the adolescent to g et good grades this year, to be considered a bright student b y his/her teachers, to be one o f the best students in his/her class, and to achieve a specialization or an M.A. after high school. Possible answers w ere based on a 4-point Likert scale in w hich 1 = not im portant at all, 2 = not too important, 3 = somewhat important, and 4 = very important.

Religiosity (fou r items; a = .84) was m easured b y using Jessor’s scale (Bonino et al., 2005; Jessor, 1992) concerning how im portant it is for the adolescent to be able to rely on religious teaching w h en he/she has a problem , to b elieve in God, to rely on religious b e lief as a guide for day-to-day living, and to be able to turn to prayer w h en he/she is facing personal problems. Possible answers w e re based on a 4-point Likert scale in which 1 = not im portant at all, 2 = not too important, 3 = somewhat important, and 4 = very important.

Intolerant attitudes about deviance (13 items; a = .91) w e re measured w ith the relative scale o f Jessor (Bonino et al., 2005; Jessor, 1992) concerning to w h at d egree the adolescent considers a series o f antisocial behaviours to be w rong, including physical aggression, shoplifting, vandalism, lying, stealing, and truancy (e.g., “How w rong do you think it is to damage o r mark up public o r private property on purpose?” ). Possible answers w ere based on a 4-point Likert scale in w h ich 1 = not wrong, 2 = a little wrong, 3 = rather wrong, and 4 = very wrong.

Regulatory self-efficacy, that is, perceived capability to resist peer pressure to engage in high-risk activities (12 items; a = .80), was m easured by using a Bandura self-efficacy scale (Bandura, 1997; Caprara, 2001 ). Adolescents w ere asked to state their sense o f efficacy in response to items such as “H ow are you able to resist i f someone asks you to do something forbidden ?” Possible answers w e re based on a 4-point Likert scale in w h ich 1 = not at all, 2 = a little, 3 = enough, and 4 = much.

Extensive psychom etric w ork was conducted on the scales o f the questionnaire, and it is reported in Bonino et al. (2005) and Caprara (2001). These analyses indicated that all the scales in the questionnaire w ere valid and had adequate reliability. Establishing the individual risk index and the friends' risk index

As has been done in previous studies (Cattelino et al., 2011), tw o risk indices w ere established to analyse the involvem ent o f adolescents and their friends in m ultiple risk behaviours. These indices integrate som e behaviours that in adolescence are

Table 1

Mean and standard deviation by gender and age group (one-way ANOVA). Groups M (SD) Individual risk index T1 Individual risk index T2 Friends' risk index Friends' models o f conventional behaviour Positive orientation toward health Positive orientation to school Religiosity Regulatory self-efficacy Intolerance o f deviance Sample N Gender Boys Girls 14 15 16 49.14 (9 .3 3 f 47.66 (7.91)*’ fi;i, 697) = 6.73 p = .010 1)2 = .010 46.01 (6 .3 3 f 48.32 (8.63)*’ 51.39 (10.08)'' F{2. 697) = 22.55 p < .001 1)2 = .061 49.79 (10.89)= 47.83 (7.76)» fi;i, 736) = 9.18 p = .003 49.77 (10.17)= 49.85 (9.87)= fi;i, 739) = 0.13 p = .909 7.50 (2.81)= 7.06 (2.59)*’ f ( l , 789) = 6.76 p = .009 ’ = .012 1)2 = .001 1)2 = .009 29.57 (3.70)= 29.98 (3.27)= fi;i,8 3 3 ) = 2.13 p = .145 13.57 (2.11)= 13.80 (1.94)= fi;i,8 3 4 ) = 2.49 p = .115 9.58 (3.27)= 26.49 (4.74)= 41.84 (8.40)= 485 10.28 (3.07)*’ 26.91 (4.44)= 45.23 (5.30)*’ 360 fi;i, 806) = 11.29 p = .001 f ( l , 751) = 1.61 p = .205 1)2 = .002 fi;i, 812) = 50.91 p < .001 1)2 = .059 1)2 = .003 1)2 = .003 1)2 = .014 47.04 (9.20)= 48.54 (8.76)= 51.16 (9.80)*’ F{2. 736) = 11.54 p < .001 46.78 (9.55)= 50.38 (9.75)» 53.02 (9.81)'^ F{2, 739) = 23.51 7.28 (2.53)= 7.40 (2.76)= 6.96 (2.80)= f(2 , 789) = 1.05 p < .001 p = .349 ’ = .031 1)2 = .060 1)2 = .003 30.29 (3.34)= 29.41 (3.55)» 29.74 (3.42)=» F{2. 833) = 4.09 p = .017 13.97 (2.0)= 13.60 (2.06)=» 13.48 (1.94)» F{2. 834) = 4.70 p = .009 10.46 (2.98)= 10.01 (3.18)= 9.27 (3.32)» F{2, 834) = 9.61 p < .001 27.45 45.00 (4.03)= (6.71)= 26.39 42.95 (4.92)» (7.45)» 26.31 43.50 (4.60)» (6.32)» f(2 , 751) = 4.59 fi;2, 812) = 6.24 p = .010 p = .002 1)2 = .012 1)2 = .015 300 332 213 1)2 = .010 1)2 = .011 1)2 = .023 Tukey post-hoc test between age groups: same letter for not statistically different means.

Strongly related: cigarette smoking, alcohol abuse, marijuana use, drug use, nonuse o f seatbelts in the car. Although these behaviours are v e ry diverse phenotypically, all are related to the transgression o f legal or social norms or to the failure to fulfil normal social role expectations.

individual risk index

The individual risk index assesses overall level o f in volvem ent in five types o f adolescent-reported risk behaviour: (a ) cigarette smoking, based on self-reports o f frequency and am ount o f smoking in the past m onth; (b ) alcohol abuse, based on respondents’ reports o f frequency o f drunkenness and frequency o f high-volum e drinking (fou r or m ore drinks per occasion); (c ) marijuana use; (d ) drug use; and (e ) nonuse o f seatbelts in the car. W ith reference to Jessor et al. (1995) and Costa et al. (2005), measures o f the five com ponents o f the index w ere transform ed into t scores (M = 50, SD = 10) and summed. Cronbach’s alpha o f the individual risk index is .71.

Friends' risk index

To measure the pattern o f risl<y behaviour o f friends, w e calculated the t-score sum o f five questions (4 -poin t Likert scale) relative to the perceived proportion o f friends involved in the same behaviours considered in order to construct the individual risk index ( “How many o f y o u r friends smoke cigarettes on a regular basis?” “drink alcohol regularly?" “use marijuana?" “use other drugs?" “pay attention to using seatbelts in the car?"). Possible answers w ere based on a 4-point Likert scale in w hich 1 = none, 2 = some, 3 = most, and 4 = all. Measures o f the com ponents o f the index w e re transform ed into t scores (M = 50, SD = 10) and summed. Cronbach’s alpha o f the friends’ risk index is .71.

Multiple regression

The association b etw een the individual risk index, friends’ risk index, and protective factors was analysed using a hier­ archical m ultiple regression m odel in w hich the individual risk index at Tim e 2 was the depen den t variable, controlled by the previous level o f individual risk index at Tim e 1.

Gender (0 = girl, 1 = b oy ) and age w ere introduced in the regression at Step 1 to evaluate the contribution o f protective factors to adolescent risk behaviour once sociodem ographic variables w ere controlled. In the second step, the friends’ risk index was entered to evaluate the im pact o f this risk factor on the adolescents’ risk behaviours. The theoretical pre­ dictors— friends’ m odels o f conventional behaviour and the individual variables selected— w e re entered at Steps 3 and 4, respectively, to exam ine their association w ith the individual risk index. A t Step 5, cross-products o f the friends’ risk index and gender and age on on e hand and all protective factors on the other hand w e re entered to assess w h eth er the influence o f friends’ risk behaviours was the same across gen der and age, and to exam ine w h eth er protective factors m oderated the contribution o f the friends’ risk index. In the last step, the th ree-w ay interactions involvin g friends’ risk index, protective factors and age and those involvin g friends’ risk index, protective factors and gen der w e re entered to evaluate w h eth er the m oderating effect o f protective factors was the same across age and gen d er

According to Holmbecl<’s term in ology (1997,2002), a m oderator specifies the conditions under w h ich a given effect occurs and the conditions under w hich the direction or strength o f an effect varies. A m oderator effect is an interaction effect. The preferred strategy is to use the variables in their continuous form and to use m ultiple regression techniques. Predictors w ere centred and interactions terms w ere com puted from these centred variables (Holm beck, 1997). Significant interactions w ere plotted for high and lo w levels o f the m oderators (1 SD b e lo w and above the m ean) and the simple slope test was perform ed, according to the recom m endations o f Aiken and W est (1991).

The statistical softw are IBM SPSS Statistics, version 21, was used in this study.

Results

A general description reported in Table 1 shows that mean scores o f the individual risk index w ere significantly higher for older adolescents than for youn ger ones both in Tim e 1 and Tim e 2; boys had significantly higher individual risk index scores than did girls in both periods. As far as the friends’ risk index is concerned, the m ean scores w ere significantly higher for older adolescents than for younger ones; there w e re no gen der differences.

Considering the protective factors, girls reported having higher values o f religiosity and intolerance o f deviance, and boys reported a greater proportion o f friends involved in conventional behaviours. Younger adolescents had higher levels o f positive orientation toward health and school and greater regulatory self-efficacy, intolerance o f deviance, and religiosity.

A ll protective factors considered w e re negatively correlated w ith the individual risk index at Tim e 1 and maintained the negative association at Tim e 2 (see Table 2). The same negative relationship was also present am ong protective factors and the friends’ risk index, and it was stronger for regulatory self-efficacy and intolerance o f deviance.

The correlations confirm the sim ilarity b etw een the behaviours o f adolescents and those o f their friends: Teens w ith friends involved in risk behaviours also show ed greater in volvem ent in multiple risl<y behaviours at Tim e 1 and at Tim e 2 (see Table 2).

To evaluate the role o f protective factors in their main effects and their possible effects o f moderation, a m odel o f hier­ archical m ultiple regression was tested that had the individual risk index m easured at Tim e 2 (con trolling for Tim e 1 ) as the dependent variable and all the variables described previously as predictors (see Table 3). The tested regression m odel was significant. The explained variance was 38%. As shown in Table 3, the succession o f steps and then the insertion o f the n ew groups o f variables increased step by step the explained variance, although w ith v e ry lo w values, except w h en the friends’ m odels o f conventional behaviour was entered into the model. As expected, the greater proportion o f variance is explained by the level o f involvem ent in risk at initial recognition.

The beta values show that the friends’ risk index plays a significant m ain effect on ly in the initial m odel, w h ile in the final m odel it shows no significant effect. Only religiosity played a protective role against in volvem ent in problem behaviours 1 year later.

Three significant tw o -w a y interaction effects w ere found b etw een the friends’ risk index and age, b etw een the friends’ risk index and the friends’ m odels o f conventional behaviour, and b etw een the friends’ risk index and the positive orientation to health, whereas all the th ree-w ay interaction effects w ere not significant. For the simple slope analysis the m odel has been re- estim ated excluding the block o f variables entered in step 6, since th ey w e re all not significant. The three significant tw o -w a y interactions are shown in Fig. 1, w h ere the relation b etw een friends’ risk index and individual risk index at Tim e 2 (sim ple slope) is plotted for high and lo w levels (1 SD above and b e lo w the m ean) o f the three moderators. The relation was positive and statistically significant w ith in older adolescents [b = .53; £(435) = 2.35, p = .019], w ith in adolescents w ith less friends involved in conventional behaviours and activities [b = .57; t(435) = 2.75, p = .006], and w ith in those w ith lo w positive orientation to health [b = .54; t(435) = 2.29, p = .022] and it was w eakly negative and not statistically significant w ith in youn ger adolescents [b = -.0 5 ; £(435) = -.2 2 , p = .82], adolescents w ith m ore friends involved in conventional behaviours [b = -.0 9 ; £(435) = - 3 9 , p = .70], and w ith in those w ith high positive orientation to health [b = -.0 6 ; £(435) = -.27,p = .78]. Thus, the relationship b etw een friends’ risk index and individual risk index was strong and positive in older adolescents and in those w ith ou t the protective role o f friends’ conventional behaviours and o f positive orientation toward health. Alterna­ tively, this relationship is much w eaker and no significant in youn ger adolescents and in presence o f high friends’ m odels o f conventional behaviour and high positive orientation tow ard health.

Table 2

Correlations among individual risk index at T1 and T2, individual risk index and protective factors at Tl.

1 2 3 4 5 6 7 8 9

1. Individual risk index T l 1

2. Individual risk index T2 .52- 1

3. Friends' risk index .54- .42- 1

4. Friends' models o f conventional behaviour -.25* -.15* -.19* 1

5. Positive orientation toward health -.20* -.16* -.16* .21* 1

6. Positive orientation to school -.27* -.18* -.24* .20* .32* 1

7. Religiosity -.25* -.23* -.16* .16* .21- .26- 1

8. Regulatory self-efficacy -.52* -.36* -.42* .18* .24- .33- .22- 1

9. Intolerance o f deviance -.52* -.37* -.47* .21- .22- .35- .23- .56- 1 •p < 001.

Table 3

Hierarchical regression model: Dependent variable individual risk index at time 2.

Step Predictors (time 1) SeB Initial I AR2

Individual risk index (IRl) T1 Gender

Age

Friends’risk index (FRI)

Friends’models o f conventional behaviour Positive orientation toward health Positive orientation to school Religiosity

Perceived regulatory self-efficacy Intolerance o f deviance FRI X Gender FRI X Age

FRI X Friends' Models o f Conventional Behaviour FRI X Positive Orientation Toward Health FRI X Positive Orientation to School FRI X Religiosity

FRI X Perceived Regulatory Self-Efficacy FRI X Intolerance o f Deviance

FRI X Friends' M. o f Conv. Behaviour x Gender FRI X Positive Orientation Toward Health x Gender FRI X Positive Orientation to School x Gender FRI X Religiosity x Gender

FRI X Perceived Regulatory Self-Efficacy x Gender FRI X Intolerance o f Deviance x Gender FRI X Friends' M. o f Conv. Behaviour x Age FRI X Positive Orientation Toward Health x Age FRI X Positive Orientation to School x Age FRI X Religiosity x Age

FRI X Perceived Regulatory Self-Efficacy x Age FRI X Intolerance o f Deviance x Age

.538 .041 .527=^=^ .310=^=^ .28^ 1.19 .702 .068 .071 .o r 1.038 .447 .095" .099=^ .417 .151 .130=^ .085 .o r -.013 .138 -.004 -.006 .00 -.131 .112 -.050 -.054 .02^ .189 .198 .042 .048 -.301 .114 -.113=^ -.134=^ -.079 .098 -.043 -.066 -.073 .074 -.053 -.044 .304 .269 .060 .072 .04^ .365 .159 .0 9 r .102=^ -.129 .040 -.134=^ -.150=^ -.093 .042 -.093=^ -.140=^ -.012 .069 -.007 .015 -.051 .041 -.051 -.038 -.049 .033 -.076 -.046 .033 .025 .069 .067 -.039 .095 -.031 .02 .087 .088 .061 -.148 .155 -.065 -.022 .087 -.014 -.048 .071 -.051 .041 .054 .064 .084 .062 .076 .040 .055 .033 -.095 .091 -.051 .094 .055 .075 .059 .043 .077 -.058 .034 -.097 = .38; Adj = .34. < .05, < .001. Discussion

The prim ary aim o f this longitudinal study was to exam ine if friends’ m odels o f conventional behaviour, positive orien ­ tation to health, positive orientation to school, religiosity, intolerant attitudes toward deviance, and regulatory self-efficacy m ight m oderate the link b etw een deviant peers and adolescent risk behaviours in a sample o f Italian adolescents. W e tested the hypothesis that these variables maintain a protective effect in a longitudinal m odel and that the same explanatory m odel accounts for risk behaviour in volvem ent across gender and age.

Results revealed that religiosity is a protective factor and that age, friends’ m odels for conventional behaviours, and positive attitude about health can m itigate the influence o f deviant friends on adolescent risk behaviour 1 year later, even after controlling for prior levels o f risk behaviour.

- Younger h - Older

b = + .5 3 *

b=-.05

-1 SD M +1 SD

Friends’ Risk Index

L o w Friends' M odels o f Conventional Behaviour I— High Friends' M odels o f Conventional Behaviow

b=^-.57♦*

b=-.09

-1 SD ^ +1 SD

Friends’ Risk Index

•3

Friends’ Risk Index

p<.05; ** p<.01

Fig. 1. Regression lines for relations between friends' risk index and individual risk index (IRI Time 2) as moderated by age, friends' models of conventional

behavior and positive orientation toward health (2-way interactions). The continuouslines with a diamond symbol represent low values (Mean - 1 DS) and the dashedlines with a square symbol represent high values (Mean + 1 DS).

M ore in depth, as shown in the final m odel o f the regression analysis, controlling for the previous level o f im plication and for other protective factors, the m odel o f deviant peers tends to disappear, but, as shown by the post-hoc probing o f the interaction effect (test o f sim ple slopes), not for older adolescents w h o m ay becom e m ore deviant w h en th ey associate w ith deviant peers. These findings require further investigation, because the results found in other studies often contradict each other. For exam ple, Patterson, Reid, and Dishion (1992) docum ented a lon g-term im pact exerted by the presence o f deviant peers and numerous protective factors; on the other hand, Tremblay, Masse, Vitaro, and Dobkin (1995) did not find a lon­ gitudinal effect exerted by the presence o f deviant peers. The results o f our study seem to indicate that, on ly after ages 15—16 years, having deviant friends in the past predicts the behaviour o f boys and girls 1 year later. This finding is consistent w ith findings from earlier studies, w hich have revealed the im portance o f peers’ m odels for adolescent behaviour. A t the same time, however, it is a caution against em phasizing the effect o f the peers’ deviant m odel over tim e (Arnett, 2007) and suggests that w e consider differences also am ong those w h o are younger (fro m 14 to 16 years) w ith respect to the role played b y the friends’ model.

W ith respect to protective factors, the individual factors that w ere considered, despite having a protective effect w h en exam ined individually, lost strength in the tested m odel, and religiosity had the on ly significant effect. The hypothesis that all the factors w ere significant was not confirm ed. On a cautionary note, other studies have found that although individual attributes o f the adolescent play a significant role in the d evelopm en t o f internalising problems, th ey appear to be less effective for curtailing the d evelopm en t o f externalizing problem s (Dekovic, 1999). W ith regard to religiosity, it has been shown that an orientation toward religion can serve m ultiple and diverse functions for an individual, from providing meaning to on e’s life, to yieldin g a sense o f personal fulfilment, to securing access to social contacts and interpersonal relationships, to offerin g a set o f standards against w hich to ju d ge and guide on e’s actions. Our results show that during m iddle adolescence religiosity has an im pact on daily, secular life, and it exerts protection against transgression. Perceiving religion as important, being active in religious w orship and activities, and referring to religious values to guide on e’s choices all constitute a deterrent w ith respect to risk. This finding is consistent w ith findings from other studies (Sinha, Cnaan, & Gelles, 2007) in w h ich religiosity was found to be consistently associated w ith reduced risk behaviours. In Italy, however, the m ajority o f youths adhere to Roman Catholic values, w h ich leaves the question o f effects o f other religions open.

Contrary to w hat is com m only assumed, positive orientation to school and regulatory self-efficacy did not appear to be protective factors. This finding is not consistent w ith findings from previous studies that have indicated that youn g people w ith a strong sense o f regulatory self-efficacy are better equipped to cope w ith the transitional stressors o f adolescence and to resist peer pressure to engage in risl<y or antisocial conduct (Bandura, 1997; Caprara, Barbaranelli, Pastorelli, & Cervone, 2004). It is likely that w h en one considers several intercorrelated predictors, as was done in our study, the odds that sig­ nificant effects are found for any o f those predictors decrease due to the covariance th ey share w ith each other and w ith the dependent variable.

O f greater interest, theoretically, are the findings about interaction effects and post-hoc probing o f these interaction effects that heighten the ability to establish m oderator effects on risk and to prom ote changes in risk behaviours. T w o factors w ere found to play an im portant role: positive orientation toward health and friends’ m odels o f conventional behaviours. A positive orientation toward health m oderates the im pact o f deviant peers because it represents a com m itm ent to behaviours that prom ote healthful outcom es and that should serve to discourage health-com prom ising behaviours. These tw o -w a y in­ teractions hold for both males and fem ales and both am ong youn ger and older adolescents: none o f the th ree-w ay in­ teractions w ith gen der and age was significant.

A social n etw ork com posed o f friends w h o adopt conventional behaviour appears to be particularly im portant for m oderating the im pact o f peers w h o have adopted risk behaviours. This finding is consistent w ith findings from other recent studies (Lösel, & Farrington, 2012; Mathys, Hyde, Shaw, & Born, 2013) and highlights the im portance o f a social netw ork that com prises various individuals w ith a variety o f behaviours and actions. A variety o f alternative behavioural patterns on the on e hand seems to favour m ore extensive testing o f conventional and risk behaviours and on the other hand seems to prom ote the ability to critique and choose. The im portance o f a m ixed social network, referring to risk behaviours and conventional behaviours, has been recently pointed out by Mathys et al. (2013,2014), w h o experim entally tested the effects o f group com position on deviant talk interaction processes am ong hom ogeneous and heterogeneous groups (m ixed group condition). Results showed less antisocial talk w ith in the m ixed group. W ith regard to nondelinquent adolescents from the purely norm ative condition, it was surprising to observe no significant differences b etw een these adolescents and those from the purely delinquent condition, w ith the exception o f reinforcem ent o f norm ative talk during the last collective session. This finding suggests that m ixed netw orks are not on ly better than a deviant peer network, but are better than any other type o f network, including a netw ork w ith on ly w ell-adapted peers. Results from our study seem to confirm these previous results in a m ore ecological context and are consistent w ith other findings that pointed out that n ew friendships and changes in friendship netw orks m ight provide opportunities and stimuli for behavioural change (Poulin, Kiesner, Pedersen, & Dishion, 2011 ), and these changes are m ore likely to occur during school transitions.

In conclusion, the main results o f this study, in continuity w ith Jessor’s w orks (Jessor et al., 2003; Jessor, & Turbin, 2014), indicate that (a ) the im pact o f the m odel o f deviant peers is stronger for 15- to 16-year-old adolescents than for younger adolescents, (b ) religiosity is an im portant protective factor against risk behaviours in m iddle adolescence, and (c ) health and healthy living and having friends involved in conventional behaviours prom ote opposition to risl<y behaviours in the presence o f deviant peers, even after controlling for prior levels o f risk behaviour. A substantial account o f the variation in individual risk behaviours was provided by these protective factors.

Limitations, strengths, and implications

Shortcomings in our study lim it the inferences that can be drawn and m ay constrain applicability o f the findings. First, it is relevant to acknow ledge that other factors, both individual (e.g., beliefs and norms, susceptibility to externalizing behaviours) and contextual (e.g., parental control and supervision, peer culture, leisure tim e spent w ith deviant peers, in volvem ent in school) could have played a main protective role or a m oderating role. Second, the fact that on ly self-report measures w ere used is a strength because evaluation o f subjective “perception” is preferable to other, apparently m ore objective measures, such as direct observation or assessment by parents, peers, and teachers, because perceived environm ent has a greater in­ fluence on the behaviour o f adolescents (Aunola, Stattin, & Nurmi, 2000). On the other hand, objective measures are needed to guide future interventions designed to target external factors that influence adolescents’ risk behaviours. In particular, the independent variables related to friends’ risk and conventional behaviours represent the perceived deviant and conventional behaviour o f adolescents’ friends. This approach does not negate the concern that peers are known to report that their friends’ behaviour is m ore similar to their o w n than it is in reality, thereby resulting in an inflated estim ate o f the relationship b e­ tw een deviant peers and adolescent behaviours (Otten, Engels, & Prinstein, 2009).

Despite its limitations, the study also has several strengths. First, it was conducted in the understudied Italian context. It was also based on a solid theory, and it had a longitudinal design. In addition, although the extant literature suggests that association w ith deviant peers contributes to the persistence and exacerbation o f adolescents’ problem behaviours, fe w studies, especially in the Italian context, have investigated processes through w h ich this relation develops and changes.

W ith respect to possible im plications o f this study, in Italy the increasing precociousness o f various risk behaviours, especially substance use, suggests the im portance o f early prevention. H owever, results from our study suggest that m iddle adolescence is a period during w h ich preventive interventions based on individual and on social factors should continue to be available. In addition, recognizing that adolescence is a tim e w h en individuals becom e m ore autonomous, m ore open to reflection about the future and about their values, and at the same tim e m ore oriented to the p eer group and its influences, these findings m ight point to n ew com ponents to enhance existing interventions. Study findings have potential im plications for prevention and treatm ent interventions that can enhance individual factors, such as religiosity and positive orientation toward health, and that prom ote social interactions am ong m ixed groups. In fact, g ro w in g eviden ce suggests these factors m ay m oderate the links b etw een deviant peer affiliations and on e’s o w n problem behaviours. Transitions in friendships, w hich in Italy are often tied to school transitions, could be vie w e d as salient opportunities to change individual behaviours. These changes could be encouraged and prom oted, for exam ple, by offerin g a variety o f structured activities (such as expressive ones or playing gam es and sports) and contexts equipped and freely usable during w hich the adolescent m ay spend tim e in a constructive manner, expand the circle o f com panions and friends, and interact w ith others w h o exhibit various patterns o f behaviour.

References

Aiken, L, & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage.

Arnett, J. J. (2007). The myth o f peer influence in adolescent smoking initiation. Health Education Behavior, 34(4), 594-607. http://dx.doi.org/10.1177/ 1090198105285330.

Aunola, K., Stattin, H., S: Nurmi, J. E. (2000). Parenting styles and adolescents' achievement strategies. Jouma/ o f Adolescence, 2 3 ,205—222. http://dx.doi.org/ 10.1006/jado.2000.0308.

Bandura, A. (1997). Self-efficacy: The exercise o f control. New York, NY: W.H. Freeman and Company.

Bonino, S., & Cattelino, E. (2012). Italy. InJ. J. Arnett {Ed.), Adolescent psychology around the world (pp. 290-305). New York, NY: Psychology Press, ISBN 978- 1-84872-889-9.

Bonino, S., Cattelino, E., & Ciairano, S. (2005). Adolescents and risk. Berlin, Germany: Springer.

Caprara, G. V. (Ed.). (2001). La valutazione dell’autoefftcacia [Self-efftcacy assessment]. Trento, Italy: Erickson.

Caprara, G. V., Barbaranelli, C., Pastorelli, C., & Cervone, D. (2004). The contribution o f self-efficaQ^ beliefs to psychosocial outcomes in adolescence: pre­ dicting beyond global dispositional tendencies. Personality and Individual Differences, 37, 751-763. http://dx.doi.Org/10.1016/j.paid.2003.ll.003. Caprara, G. V., Scabini, E., Steca, P., & Schwartz, S. H. (Eds.). (2011 ). I valori nell’Italia contemporanea [values in contemporany ItalyJ. Milano, Italy: ErancoAngeli. Cattelino, E. (Ed.). (2010). Rischi in adolescenza [Risks in adolescence]. Roma, Italy: Carocci.

Cattelino, E., Begotti, T., & Bonino, S. (1999). Questionnaire: lo, la scuola e il mio stile di vita [I, the school and my life-stylej. Torino, Italy: Department of Psychology.

Cattelino, E., Calandri, E., Bina, M., & Graziano, R (2011). Fattori di protezione nei confronti di pari devianti in adolescenza [Protective factors from deviant peers in adolescence]. Giornale di Psicologia dello sviluppo/Joumal o f Developmental Psychology, 9 8 ,104—114.

Costa, F. M., Jessor, R., Turbin, M. S., Dong, Q., Zhang, H., & Wang, C. (2005). The role o f social contexts in adolescence: context protection and context risk in the United States and China. Applied Development Science, 9(2), 67-85. http://dx.doi.org/10.1207/sl532480xads0902_3.

Dekovic, M. (1999). Risk and protective factors in the development o f problem behavior during adolescence. Journal o f Youth and Adolescence, 28(6), 667-685. http://dx.doi.Org/10.1023/A:1021635516758.

Dishion, T. J., Dodge, K. A., & Lansford, J. E. (2008). Deviant by design: risks associated with aggregating deviant peers into group prevention and treatment programs. The Prevention Researcher, 15(1), 8—11.

Dishion, T. J., Ha, T., & Véronneau, M. H. (2012). An ecological analysis o f the effects o f deviant peer clustering on sexual promiscuity, problem behavior, and childbearing from early adolescence to adulthood: an enhancement o f the life history framework. Developmental Psychology, 48(3), 703—717. http://dx. doi.org/10.1037/a0027304.

Dodge, K. A., Coie, J. D., & Lynam, D. (2006). Aggression and antisocial behavior in youth. In W. Damon, & M. R. Lerner (Eds.) (6th ed.)Soda/, emotional, and personality development: Vol. 3. Handbook o f child psychology (pp. 719—788). New York, NY: Wiley.

Dodge, K. A., Dishion, T. J., & Lansford, J. E. (Eds.). (2006). Deviant peer influences in programs fo r youth: Problems and solutions. New York, NY: Guilford Press. Engels, R. C. M. E., Knibbe, R. A., Drop, M. J., & de Haan, Y. T. (1997). Homogeneity o f cigarette smoking within peer groups: Influence or selection? Health

Education Æ Behavior, 24, 801-811.

Ennett, S. T., & Bauman, K. E. (1994). The contribution o f influence and selection to adolescent peer group homogeneity: the case o f adolescent cigarette smoking. Jouma/ o f Personality and Social Psychology, 67, 653—663. http://dx.doi.Org/10.1037/0022-3514.67.4.653.

Felix-Ortiz, M., & Newcomb, M. D. (1992). Risk and protective factors for drug use among Latino and W hite adolescents. Hispanic Journal o f Behavioral Sciences, U , 291-309. http://dx.doi.org/10.1177/07399863920143001.

Fergusson, D. M., Vitaro, P., Wanner, B., & Brendgen, M. (2007). Protective and compensatory factors mitigating the influence o f deviant friends on de­ linquent behaviours during early adolescence. Joumai o f Adolescence, 30, 33-50. http://dx.doi.Org/10.1016/j.adolescence.2005.05.007

Hartup, W. W , & Stevens, N. (1997). Priendships and adaptation in the life course. Psychological Bulletin, 121,355-370. http://dx.doi.org/10.1037/0033-2909. 121.3.355.

Holmbeck, G. N. (1997). Toward terminological, conceptual, and statistical clarity in the study o f mediators and moderators: examples from the child- clinical and pediatric psychology literatures. Journal of Consulting and Clinical Psychology, 65(4), 599—610. http://dx.doi.Org/10.1037/0022-006X.65.4. 599.

Holmbeck, G. N. (2002). Post-hoc probing o f significant moderational and mediational effects in studies o f pediatric populations. Journal o f Pediatric Psychology 2 7 (i), 87-96.

ISTAT [Italian National Institute o f Statistics], (2013). L'ltalia in cifre [Statistics ofltalyj. Roma: ISTAT. http://www.istat.it/. http://www.istat.it/. http://www. istat.it.

Jessor, R. (1992). Health Behavior Questionnaire. USA: Institute o f Behavioral Science, University o f Colorado.

Jessor, R., & Turbin, M. S. (2014). Parsing protection and risk for problem behavior versus pro-social behavior among US and Chinese adolescents. Joumd of Youth and Adolescence, 43(7), 1037-1051. http://dx.doi.Org/10.1007/sl0964-014-0130-y

Jessor, R., Turbin, M. S., & Costa, P. M. (1998). Protective factors in adolescent health behavior Joumd o f Personality and Social Psychology, 75(3), 788-800. http://dx.doi.Org/10.1037/0022-3514.75.3.788.

Jessor, R., Turbin, M. S., Costa, P. M., Dong, Q,, Zhang, H., & Wang, C. (2003). Adolescent problem behavior in China and the United States: a cross-national study o f psychosocial protective factors. Journal o f Research on Adolescence, ¡3(3), 329-360. http://dx.doi.org/10.llll/1532-7795.1303004.

Jessor, R., Van Den Bos, J., Vanderryn, J., Costa, P., & Turbin, M. S. (1995). Protective factors in adolescent problem behaviour: moderator effects and developmental change. Developmental Psychology, 31, 923-933. http://dx.doi.Org/10.1037/0012-1649.31.6.923.

Kandel, D. B. (1978). Homophily, selection, and socialization in adolescent friendships. American Journal o f Sociology, 84(2), 427—436. http://dx.doi.org/10. 1086/226792.

Keijsers, L, Branje, S., Hawk, T. S., Schwartz, S. J., Prijns, T, Koot, H. M., et al. (2012). Porbidden friends as forbidden fruit: parental supervision o f friendships, contact with deviant peers, and adolescent delinquency. Child Development, 83(2), 651—666.

Larsen, H., Pngels, R C. M. P., Souren, P. M., Granic, L, & Overbeek, G. (2009). Peer influence in a micro-perspective: Imitation o f alcoholic and non-alcoholic beverages. Addictive Behaviors, 35(1), 49—52. http://dx.doi.Org/10.1016/j.addbeh.2009.08.002.

de Leeuw, R. N. H., Pngels, R. C. M. P., Vermulst, A. A., & Schölte, R. H. J. (2009). Relative risks o f exposure to different smoking models on the development of nicotine dependence during adolescence: a five-wave longitudinal study. Journal o f Adolescent Health, 45(2), 171-178. http://dx.doi.0rg/lO.lOl6/j. jadohealth.2008.12.013.

Lösel, P, & Parrington, D. P. (2012). Direct protective and buffering protective factors in the development o f youth violence. American Journal o f Preventive Medicine, 43(2), S8-S23. http://dx.doi.Org/10.1016/j.amepre.2012.04.029.

Mathys, C., Hyde, L. W., Shaw, D. S., & Born, M. (2013). Deviancy and normative training processes in experimental groups o f delinquent and nondelinquent male adolescents. Aggressive Behavior, 39, 30-44. http://dx.doi.org/10.1002/ab.21456.

Mathys, C., Vitaro, P., & Born, M. (2014). Comment comprendre l'influence au sein de groups d'adolescents: conflit sociocognitif ou apprentissage social? Canadian Journal o f Behavioural SciencejRevue Canadienne de Sciences du Comportement, 46(2), 289-300.

Moffitt, T. P. (1993). Adolescence-limited and life-course-persistent antisocial behavior: a developmental taxonomy. Psychological Review, 100, 674—701. http://dx.doi.Org/10.1037/0033-295X.100.4.674.

Monahan, K. C, Steinberg, L, & Cauffman, P. (2009). Affiliation w ith antisocial peers, susceptibility to peer influence, and antisocial behavior during the transition to adulthood. Developmental Psychology, 45(6), 1520—1530. http://dx.doi.org/10.1037/a0017417.

Otten, R., Pngels, R. C, & Prinstein, M. J. (2009). A prospective study o f perception in adolescent smoking. Joumai o f Adolescent Health, 44(5), 478-484. http://dx.doi.Org/10.1016/j.jadohealth.2008.09.004.

Patterson, G. R., Reid, J. B., & Dishion, T. J. (1992). Antisocial boys. Pugene, OR: Castalia.

Pauwels, L J., Vettenburg, N., Gavray, C, & Brondeel, R. (2011). Societal vulnerability and troublesome youth group involvement. The mediating role of violent values and low self-control. International Criminal Justice Review, 21(3), 283—296.

Poulin, P, Kiesner.J., Pedersen, S., & Dishion, T. J. (2011). A short-term longitudinal analysis o f friendship selection on early adolescent substance use. Journal o f Adolescence, 34, 249—256. http://dx.doi.Org/10.1016/j.adolescence.2010.05.006.

Rubin, K. H., Bukowski, W , & Parker, J. G. (2006). Peer interactions, relationships, and groups. In W. Damon, & M. R. Lerner (Pds.) (4th ed.)Social, emotional, and personality development: Vol. 3. Handbook o f child psychology (pp. 571-645). New York, NY: Wiley.

Rutter, M. (1987). Psychosocial resilience and protective mechanisms. American Journal o f Orthopsychiatry, 57, 316—331. http://dx.doi.org/10.1017/ CB09780511752872.013.

Schwartz, S. H. (1992). Universals in the content and structure o f values: theoretical advances and empirical tests in 20 countries. In M. P. Zanna (Pd.), Vol. 25. Advances in experimental social psychology (pp. 1—65). New York, NY: Academic Press.

Silbereisen, R. K., Pyferth, K., & Rudinger, P. (1986). Development as action in context. Problem behaviour and normal youth development. Berlin, Germany: Springer-Verlag.

Sinha, J. W , Cnaan, R. A., & Gelles, R. J. (2007). Adolescent risk behaviors and religion: findings from a national study. Journal o f Adolescence, 30, 231-249. http://dx.doi.0rg/lO.lOl6/j.adolescence.2OO6.O2.OO5 .

Tremblay, R. P., Masse, L C, Vitaro, P., & Dobkin, P. L. (1995). The impact o f friends' deviant behavior on early onset o f deliquency: longitudinal data from 6 to 13 years o f age. Development and Psychopathology, 7(4), 649-667. http://dx.doi.Org/10.1017/S0954579400006763.

Turbin, M. S., Jessor, R., Costa, P. M., Dong, Q,, Zhang, H., & Wang, C. (2006). Protective and risk factors in health-enhancing behavior among adolescents in China and the United States: does social context matter? Health Psychology, 25(4), 445-454. http://dx.doi.Org/10.1037/0278-6133.25.4.445. Urberg, A., Luo, Q,, Pilgrim, C., & Degirmencioglu, M. (2003). A two-stage model o f peer influence in adolescent substance use: individual and relationship-

specific differences in susceptibility to influence. Addictive Behaviors, 2 8 ,1243—1256. http://dx.doi.org/10.1016/S0306-4603(02)00256-3.

Wanner, B., Vitaro, P, Carbonneau, R., & Tremblay, R. P. (2009). Cross-lagged links among gambling, substance use, and delinquency from midadolescence to young adulthood: additive and moderating effects o f common risk factors. Psychology o f Addictive Behaviors, 23(1), 91—104. http://dx.doi.org/10.1037/ a0013182.