Novelty Seeking_Cloninger's Constructs Related to Substance Use Level and Problems in Late Adolescence

8/14/2019 Novelty Seeking_Cloninger's Constructs Related to Substance Use Level and Problems in Late Adolescence

1/13

Experimental andClinicalPsychopharmacology1999,Vol.7, No.2,122-134 Copyright 1999 by the American Psych ological Association, Inc.1Q64-1297/99/S3.00

Cloninger's Constructs RelatedtoSubstanceUseLevelandProblemsinLate Ado lescence:AMediational Model Basedon Self-ControlandC oping MotivesThomasAshbyWillsFerkauf GraduateSchoolofPsychology andAlbertEinstein Collegeof Medicine

James M. SandyAlbert EinsteinCollegeof M edicine

OriShinarFerkauf Graduate Schoolof PsychologyPredictions concerning mediating processes for the effects of C. R. Cloninger's (1987a)constructs were tested; criterion variables were substance use level and substance useproblems. Participants were 1,225 adolescents (M age: 15.5 years). Structural modelingindicated indirect effects for novelty seeking, harm avoidance, and task reward dependence,mediated through self-control; harm avoidance also had an inverse direct path to substance uselevel, and social reward dependence had a positive direct path to coping motives for substanceuse. Good self-control had inverse paths to life events and deviant peer affiliations; poorself-controlhad positive pathstolife eventsand coping motives;andrisk takinghadpositivepathstocopingmotives andpeer affiliations. Coping motiveshad apathtoleveland adirectpath to problems; peeraffiliations had a path only to substance uselevel.

Recent research has provided evidence for a role ofdispositionalconstructsinriskforsubstance use. Longitudi-nal studies have indicated that temperament constructsmeasured in childhood predict sub stance use in adolescenceand adulthood(e.g., Masse & Tremblay, 1997; Pulkkinen &Pitkanen, 1994). These constructs generally show evidenceof heritability (Buss&Plomin,1984; Heath, Cloninger,&Martin, 1994) andmoderate stabilityover time (Hagekull,1989; Pedlow, Sanson,Prior,& Oberklaid, 1993).Severalstudies have implicated constructs fromCloninger's theory(Cloninger, 1987a; Cloninger, Svrakic, & Przybeck, 1993)in risk forsubstance use. This theory outlinesfour disposi-tional constructs: novelty seeking (preference fo r novelsituations),harm avoidance (sensitivity to threat cues),taskreward dependence or persistence (sensitivity to rewardsfrom instrumental achievement), and social reward depen-dence (sensitivity to rewards from social relationships). A

Thomas Ashby Wills, Ferkauf Graduate School of Psychologyand Department of Epidemiology and Social Medicine, AlbertEinstein College of Medicine of Yeshiva University; James M.Sandy, Department of Epidemiology and Social Medicine, AlbertEinstein College ofMedicineof Yeshiva University; Ori Shinar,Ferkauf Graduate Schoo l of Psychology of Yeshiva University.This research was supported by Grant R01-DA08880 andResearch Scientist Development Award K02-DA00252 from theNational Institute o n Drug Abuse . We thank the superintendents oftheschooldistrictsfortheir supportand theparentsand participat-ing students for their cooperation.Correspondence concerning this article should be addressed toThomas Ashby Wills, Health Psychology Training Program, Fer-kauf Graduate School of Psychology, Albert Einstein College ofMedicine, 1300 Morris Park Avenue, Bronx, New York 10461.Electronic mailmay [email protected].

rangeofquestionsabouttheseconstructs h ave been investi-gated (for a detailed review, see Howard, Kivlahan, &Walker, 1997). Cloninger's constructsarerelatedtosmokingand alcohol use in adolescence and adulthood (Cloninger,Sigvardsson, & Bohman, 1988; Wills, Vaccaro, & McNa-mara, 1994; Wills, Windle, & Cleary, 1998), and humanstudies show convergence with animal research on drugself-administration (Bardo, Donohew,&Harrington, 1996).Thus, it is of interest for psychobiological theory to gain abetter understanding of the process through which theseconstructsarerelatedtosubstance use.The presentstudy investigated high-intensity, early-onsetsubstance use, whichislinked with TypeIIalcohol abuseinCloninger's (1987a)theoryand is prognostic for substanceabuse in adulthood (Kandel & Davies, 1992; Robins &Przybeck, 1985). The research was designed to test anepigenetic m odelofsubstanceuseetiology (Tarter, Moss,&Vanyukov,1995;Zucker, 1994) proposing thatdispositionalvariables such as novelty seeking will be related to deriva-tive characteristics, such as poor self-control ability andrisk-taking tendency, that develop throughout adolescence(Maggs, Almeida, & Galambos, 1995). Such behavioralcharacteristicsarepositedtopredispose some individualstobe exposed to variables such as negative events and dev iant peeraffiliations, which are proximal risk factors for adolescentsubstance use (Haw kins, Catalano, & Miller, 1992; Wills, 1990).

The study was also designed to clarify two questionsabout substance use outcomes that could be addressed inlater adolescence, during which rates of substance use can besubstantial (Johnston, O'Malley, & Bachman, 1995). First,there is a question as to whether effects of dispositionalconstructs are mediated through copingfunctions that peopleperceive to be provided through substance use (Cooper,122


2/13

SUBSTANCEUSELEVELAND PROBLEMS 1231994; Wills & Shiftman, 1985). Although coping motivesare strong predictors of substance use (Newcomb, Chou,Bentler, & Huba, 1988; Wills & Cleary, 1995), there isrelatively little research onwhetherdispositionalconstructsshow systematic relationships tocoping motives. Second,thereis anempirical distinction between levelof substanceuse andproblems associated with substanceuse(Newcomb,1992; Smith, McCarthy, &Goldman, 1995), but atpresentthere islittle evidence on whether substance uselevel andproblems may have different predictors. Accordingly, thisstudy tested for mediation of effects of other variablesthrough coping motives and for differential prediction ofsubstanceuselevelandsubstanceuseproblems.Thefollow-ing sections describe the rationale for the design and thehypotheses.

MediationThroughSelf-Control andRiskTakingThe firstpartof thestudy design concerns thehypothesisthat dispositional constructs willbe related to self-controlability andrisk-taking tendency.It is posited that this may

occur through bothsocialandnonsocialmechanisms (Roth-bart& Ahadi, 1994; Tarter, Moss,&Vanyukov,1995).Forexample, young people elevated on novelty seeking maylack opportunityf or reinforcement in learning self-controlskills because of a constant search for new experiences.High novelty seekingm ayalsoaffect thequalityof relation-ships with important socializing agents, suchasparents,andthereby have a systematic effect on self-control throughsocialization processes.Thisprocessispositedto becrystal-lized by later adolescence, when higher level executivefunctions that involve planning, restraint, an d monitoringhave become established (Stuss, 1992; Tarter, Mezzich,Hsieh,&Parks, 1995).Parallelarguments suggest that harm avoidance also will

be inversely related to self-control ability. For example,focusing attention on threat cues may restrict attention toalternativecuesandhence interfere withthedevelopmentofeffective problem-solving strategies (Rothbart, Derryberry,& Posner, 1994). Thus, both harm avoidance andnoveltyseeking are predicted to be related to lower self-controlability, although fo r different reasons. In contrast, theattentional control that underlies task reward dependence,facilitating persistence ontasks,isbelieved tocontributetodevelopmento fgood self-control ability (Rothbart&Ahadi,1994).Predictions about social reward dimensionsaremorequalified; Tarter (1988) has suggested that sociability mayhave no consistent relation tosubstance use risk when itsassociation with other personality dimensionsisconsidered.Epigenetic theory posits that self-controlwillberelatedtoproximal factors fo r substance use, including academiccompetence, lifeevents,andaffiliation with deviance-pronepeers (Chassin, Pillow, Curran, Molina, & Barrera, 1993;Hawkins etal.,1992). Good self-control abilityispredictedtobeinversely related tooccurrenceofnegativelifeeventsbecauseof anindividual's better abilitytoplan activitiesan davoid problematic situations (including deviance-pronepeers);incontrast, poor self-controlispredictedto berelatedto more negative events because of impulsive action and

difficulty restraining oneself in the face of provocation.These arguments suggest indirect effects of self-controlconstructs on substance use, mediated through life eventsanddeviantpeer affiliations.The status of risk-taking tendency inpathways for sub-stance use liability is unclear. Developmental theory hassuggested that a preference fo r high levels of excitement

may belinked to temperament characteristics (Rothbart &Ahadi, 1994),but it hasbeen unclear inprior work whetherrisktaking representsanindependent constructinsubstanceuseliabilityorwhetherit issimplyanaspectofundercontrol(e.g.,seeSher&Trull, 1994;Zuckerman,1994a,chap.5). Inthe present study,w e tested whether risktaking would beempiricallydifferentiated frompoorself-controlandwhetheritrepresents adistinct pathwayformediating the effects ofdispositional constructs.

CopingMotivesforSubstanceUseResearch with adultsha sindicated individualdifferencesin motives fo r substance use (Cooper, Russell, & George,

1988; Wills &Shiftman, 1985). Across avariety ofstudies,there is similarity in a number of motives reported fo rcigarette smoking and alcohol use (U.S. Department ofHealthandHuman Services, 1988, chap.6;Wills&Cleary,1995).Research with adolescentshasindicated some motivedimensions that involvesocial situations and other dimen-sions that represent coping-related functions of substanceuse,with variants reflecting tension reduction, escape fromworries,andrelief from boredom (Cooper, 1994; Windle&Windle,1996).Although several coping motive dimensionshave been reported, therearesubstantial correlations amongthem,a ndsome analyses have indicated scoringfor asinglehigherorder construct (Newcombetal., 1988).

Therehasbeen relatively littleresearchonvariables thatpredict coping motives. Two studies have found relation-ships of sensationseekinga construct related tonoveltyseekingtoexpectanciesabout alcohol (Barleywine, 1994;Stacy, Newcomb, & Bentler, 1993). In addition, Cooper,Frone, Russell, and Mudar (1995) found that a briefsensation seeking scale wasrelated toalcohol use throughenhancement motives fo r drinking. However, these studiesdid not measure other variables that may be involved inmediatingtheeffect ofdispositional factors.Ourperspectivesuggested thattherelationofdispositional factorstocopingmotives would be mediated through self-control, becauseindividualswho arepronetohigherlevelsofcoping-relatedsubstance use are those who have less ability fo r self-regulation ofcognition and emotion (Wills&Hirky,1996).Hence, we predicted that there wouldbeindirect effects ofdispositional constructsoncoping motives, mediated throughpoor self-control.

SubstanceUseLevelandProblemsThe third part of the design concerned the distinctionbetween level ofsubstance use and problems occurring inassociation with substance use. Prior researchhasprovidedseparate measuresfor thelevelo fadolescents'substanceuse(based on frequency or amount of use and number of


3/13

124 WILLS, SANDY,AND SHINARsubstances used) and the problems associated with substanceuse in various domains, such as legal problems, school-related problems, and interpersonal problems (White &Labouvie, 1989). Studies indicate that substance use levelandproblems are correlated but distinct constructs (Bailey &Rachal, 1993; Smithet al., 1995) andraise the questionofwhether level and problems will be predicted by differentvariables.Wehypothesized that coping motives would be a predic-tor of substance use problems as well as level, because of thegrounding of coping motives in poor self-control. Ourreasoning was that substance use problems may occur in partbecause of poor self-control: less ability to anticipatepotentially problematic situations or to restrain uncontrolledresponses(e.g.,drunk driving) that may be provoked in suchsituations (cf. Stacy, Newcomb, & Rentier, 1991; Stice,Myers, & Brown, 1998).Thissuggests that coping motiveswillpredict substance use level and also will involve a directpath to problems. In contrast, we reasoned that affiliationwith deviant peers is driven more by variables such asnegative life events (Castro, Maddahian, Newcomb, &Bentler, 1987) and social approach tendencies (Windle,1995).Hence, we predicted that coping motives would havepaths to both level and problems, whereas peer affiliationswouldinvolve a path to level of use but not to problems.

MethodParticipants

Th e 1,225participants were lOth-grade students in fourpublichigh schools inlower Westchester County,N ewYork.T heschoolsarelocated in mixed urban-suburban communities thataresocio-economicallyrepresentative of thestate population(U.S.Depart-ment ofCommerce, 1992).T hesamplewas 52%female and 48%male;meanage was15.50years(SD =0.67).Self-reported ethnicidentificationindicated thatthesamplewas 20%AfricanAmerican,30 % Hispanic, 4% Asian American, 34 % Caucasian, 4% mixedethnicity, and 8% other ethnicity. Datao nfamilystructure indicatedthat 32% of participants were currently living in a single-parentfamily, 11%were livingin ablended family(onebiological parentand onestepparent),and 56%were livingin anintact family. Meanparentaleducation level on a 1-6 scale was3.80(SD =1.4;justabovehigh school graduate).Procedure

Aquestionnairew asadministered tostudentsinclassrooms bytrained research staff, who followed a standardized protocol ingiving instructions to students. Students were instructed not towrite their nameon thesurveyandwere assured thattheiranswerswerestrictlyconfidentiala ndwere protected throughacertificateo fconfidentiality from theU.S.Public Health Service. Methodologi-ca l research has shown that when participants are assured ofconfidentiality, self-reports of substance use have good validity(Murray &Perry, 1987).In the consent procedure, parents were sentanoticeby directmail informing themof thepurpose of thestudyand thenatureo fthe questionnaire measures.Theparent could havehis or herchildexcluded from thedata collection,if he or shewished,byreturninga stamped postcard to the experimenter or calling a designatedofficialat theschool. Students received similar written information

at thetimeo fquestionnaire administration an dwere informed thattheycouldrefuse ordiscontinue participation.Thecompletion rate(number of surveys completed divided by total class enrollmentfrom school lists) was 83%,with case loss occurring because ofparental exclusion (1%),student withdrawal (2%),and studentabsenteeism orunavailabilityattimeo ftesting (14 ).

MeasuresMeasuresaredescribedin thefollowingsections.Unless other-wise noted, responses were madeon5-pointLikertscalesrangingfromnot at alltrueof me(1) tovery trueof me(5).Scalestructurewasverifiedwith factor analysis an dinternal consistency analysis;a fewitems were dropped becauseof lowitem-totalcorrelations.All measures were scored such that a higher score indicated ahigher levelof thenamed characteristic.Demographics. Itemsondemographiccharacteristicsincludedage, gender, and ethnic background (eight options).An item onfamily composition asked what adult(s) the participant currentlylived with (eight options, multiple responding, receded to threelevels: living withtwobiological parents, single-parent family,orblended family). Responses toitems on parental education weremadeo n a6-pointscalerangingfromgrade school education(1) topostcollege education(6).Cloninger constructs. We used a 28-item version of theTridimensionalPersonality Questionnaire(TPQ;Cloninger, 1987b;Stallings, Hewitt, Cloninger, Heath, &Eaves, 1996)adapted foradolescents in a previous study (Wills et al., 1998). Noveltyseeking (9items;a = .82)included items about tryingnewthingsforfun orthrills,looking fo rexcitement, an ddoing things basedonfeelings of the moment. Harm avoidance (8 items; a =.84)includeditems about worrying thatterriblethings might happen,feeling tense in unfamiliar situations,an d havingdifficulty adapt-ing to changes. Task reward dependence (6 items; a =.85)included items about wantingto do aswella spossibleontasksandlikingto paycloseattentiontodetails. Socialrewarddependence (5items; a.= .76)included items on feeling better around friendsthan being alone an dbeing able to talk about personal problemswithpeople.Parentalrelationship measures. A 14-iteminventory assessedsupportive versus conflictual aspects offamily relationships (Bar-rera, Chassin, & Rogosch, 1993;Wills, Vaccaro, & McNamara,1992).Parental support measures werea5-itememotional supportscale(e.g., If Ihaveaproblem,myparent will listen ;a =.89)anda6-item instrumental support scale(e.g., If Ineed help withmy school work, I can ask my parent about it ; a =.80).Participants were instructed, inresponding to the items, to thinkabout the oneparent they talked to themost.A3-itemmeasureofparent-adolescent conflict (a = .80) included such items as Ihave a lot ofarguments withmyparent.Family l i f e events. A20-item checklisto fnegativelifeeventswasbased on previous inventories fo r adolescents (Newcomb &Harlow, 1986;Wills et al., 1992). Items were presented with adichotomous(yes-no)response scale, and theparticipant indicatedwhether each event had occurred during the previous year. Theinventory included a 10-item scale for family events (a =.58),those that could involve another family member an d could no tplausibly becausedby theadolescent (e.g., My father ormotherlosthis/herjob ).Parentalsubstanceuse. Questions about regular substanceuse(defined asweekly ormore often) by parents included items fo rcigarette smoking, alcohol use (beer or wine), and use of liquor(whisky, scotch, orrum).Measures for parental smoking, beer-winedrinking, andliquor drinking were coded on3-pointscales(1= noneuse,2 = oneuses,3 = twouse).


4/13

SUBSTANCE USE LEVELAND PROBLEMS 125Good self-control measures. Multiple indicators for the con-structofgood self-control (cf. Block&Blo ck, 1980) were de rivedfrom an inventory on generalized control in everyday situations(Kendall&Williams, 1982; Wills,DuHamel,&Vaccaro,1995).Aset ofeight itemson generalized good self-control (a = .73)wasscored fo rsubscales indicated in previous factor analyses (Willse t

al., 1998); these were termed soothability (two items; e.g., I cancalm down easily w henexcited ),dependability (three items; e.g., WhenIpromiseto dosom ething,you cancounton me to do it ),andattentionalcontrol (three items; e.g., Iprefertoco ncentrateononethingat atime ).Afourth indicatorwas ascaleonbehavioralcoping(from Wills,McNamara,Vaccaro,&Hirky, 1996;a = .92);this scale included eight items indicating that when the participanthad aproblem,he or she would think about choices before doinganything and try different w ays to solve the problem. Subscalecorrelations (rs) ranged from .29 (soo thability-dependability) to.51 (dependability-attentional control).Poorself-control measures. Multiple indicato rs were obtainedforthe co nstruct of poor self-control (cf. Sher & Trull, 1994). A setof nine items from the Kendall-Wilcox inventory (Kendall &Wilcox, 1979; Kendall & Williams, 1982) on generalized poorself-control(a = .80)wasscoredfor subscales termed impatience(three items; e.g., I have to have everything right away ) anddistractibility (five items; e.g., I liketo switch from onethingtoanother ). A seven-item impulsiveness scale (a = .86) was de-rived from an inventory developed by Eysenck and Eysenck(1977); this scale included items about doing things withoutstopping to thinkand getting involvedin thingsone wished laterone could get out of. A seven-item anger coping scale (Wills,McNamara, etal., 1996; a = .89) included items indicating thatwhenthe participant had a problem, he or she would take it out onsomeone else and blame and criticize other people. Subscalecorrelations ranged from .41(impatience-anger coping) to .55(distractibility-impulsiveness).Risk taking. A six-item measure of risk-taking tenden cy(a = .90)wasadapted fromtheinventoryofEysenckandEysenck(1977). This scale included items suchas Ienjoy taking risks, Iwould enjoy fast driving, I would do almost anything for adare, I thinklifewith no danger in it would be dull for me, and I'd prefer to do so mething a little risky rather than having tostayquiet for a fewhours.Academic competence. Competence in academic situations, asperceived by the respondent, was indexed with a seven-itemmeasure (a = .72) derived from the Perceived Competence Scalefor Adolescents (Harter, 1985). This scale included positively scoreditems suchas I like school because I do wellinclass and reverseditems such as I have trouble figuring out answers inschool.Adolescent l i f e events. The lifeevents checklist described previ-ously included a set ofevents that could occurin theadolescent's lifeandm ight plausiblybecausedby theadolescent himselfo rherself (e.g., I had a serious accident and I got disciplined or suspended inschool ).Anine-item scale(a = .54)wasused.Substance use motives. Tw o 15-item inventories indexingmotives for cigarette and alcohol use were adapted frompreviousmeasures (Cooper, 1994; Wills &Cleary, 1995); item content wasidentical for the two inventories except that the nam e of the targetsubstancewas changed. Items were introduced with the followingstem: Hereare some things that people have said about smokingcigarettes/drinking beer orwine. Circle anumber (from 1 to 5) toshow whatyouthink. Factor analyses indicated some separationofthe 4itemsonsocial mo tives (alphasof .88 and .89 for tobaccoan d alcohol, respectively) from th e I I items on coping motives(alphasof .94 and .93), but there were substantial intercorrelationsamong the scales; accordingly, each inventory was scored for fourrational subscales.A4-item social motives scale included items

such as Smoking [drinking] helpsyou fit in with other peopleand Smoking makes social gatherings more fun. A 4-itemself-enhancement motives scale included items suchas Smokingmakes youfeelmore self-confident and Smokingmakes you feelmoreenergetic. A 2-item boredom relief motives scale includedthe items Smoking is something to dowhen yo u're bored and You smoke when there's nothing better to do. A 5-item stressreduction motives scale included items such as Smoking helpsyo u calm down when you're feeling tense and nervous and Smoking cheers you up when you're in a badmood. Subscalecorrelations rangedfrom .63 (self-enhancement-bo redom relief) to.80(social motives-self-enhancement).Peer substance use. Three items askedthe participant whetherany of his or her friends smoked cigarettes, drank beer or wine, orsmoked marijuana. Responses were madeon5-point scales rangingfrom none(1) tomore than three(5). The alpha coefficient for thethree-itemscalewas .86.Substance use level. Substance use by the participant wasmeasured with items that asked about the typicalfrequency of hisor her cigarette, alcohol, and marijuana use. Three items wereintroduced toparticipants withth estem How often do you smokecigarettes/drink alcohol/smokemarijuana? Responses were madeon 6-point scales(1 =never used,2 = tried once-twice,3 = used

four-five times, 4 = usually use a few times a month, 5 = usuallyuse a few timesa week, 6 = usuallyus e every day).Afourth itemasked the participant whether there was a time in the previousmonth when he or she had had three or more drinks on oneoccasion; response points were no , happened once, happenedtwice,an dhappened more than twice. A composite score based onthe fouritems had an alphacoefficient of .82.Substance use problems. The measure for problems associatedwith substance use was based on the inventory of White andLabouvie(1989).Items were introduced with the stem Herearesome things that may happen in connection with smoking, alcohol,or drug use. During the past year, have you experienced any ofthese things? Responses were made on a 4-point scale rangingfrom neverhappened (1) tohappened threeormore times(4). The16-itemmeasure(a = .85)did nothaveaclear factor structureandwas scored for four rational subscales. A 3-item indicator forinterpersonal problems included items such as Broke up with aboy/girl friend because of smoking or alcohol use and Friendsavoided you because of smoking or alcohol use. A 6-itemindicator for institutiona l problem s included items such as Got introuble for alcohol or druguse atschool, Got intofightswhenyou were drinking, an d Got into trouble withth epolice becauseof alcohol or drugs. A 3-item indicator for excessive use problemsincluded items such as Used more alcohol or drugs than you hadintended and Peopletold you abo ut things you said o r did whileyouwerehigh,but youcouldn't remember. A 3-itemindicatorforcontrol problems included items such as Told by a friend to cutdown on smoking, drinking ordruguse and Youtried to cut down onsmoking or alcohol use but couldn't cut dow n much. Subscalecorrelationsrangedfrom .33(interpersonalproblems-excessuseprob-lems) to.61(institutionalproblems-excessuse problems).1

1Several other measures w ere obta ined in the study. These w erescales fo r approach tendency an d inhibitory tendency (Windle,1994, 1995), anger proneness (Swaim, Getting, Edwards, &Beauvais, 1989), optimism (Scheier & Carver, 1985), positive andnegative perceived control (Wills, 1994), and externalizing andinternalizing symptomatology (Achenbach, 1991). The motiveinventories included scales on ne gative attitudes toward cigarettesand alcohol. These variables were obtained to address otherquestionsand are not reported indetailhere.


5/13

126 WILLS, SANDY,AND SHINARResults

Prevalence rates for the substance use indexes are pre-sentedin Table 1.There was a range of usage levels, with aproportion of the sample engaging in high-intensity use. Forexample, 13%of theparticipants reported cigarette smokingweekly or more often, and 9% reported daily smoking. Interms of the heavy drinking index, 22% of the participantsreported heavy drinking once or twice in the previousmonth,and 8% reported heavy drinking three times or moreduring theprevious month. Rates fo rregular marijuanau sewere lower, with 5% of the participants reporting monthlymarijuana use and 5% reporting weekly or daily use. Theseprevalence ratesare generally comparable to those reportedbyJohnston et al. (1995).The various indexes of substance use were substantiallycorrelated. Correlations (rs) ranged from .47 (smoking-heavy drinking) to .62 (cigarette smoking-marijuana use).The level of intercorrelation among different types ofsubstance use was comparable to data previously reportedfor adolescent samples (Hays,Widaman,DiMatteo, & Stacy,1987; Wills & Cleary, 1995); hence, the tobacco, alcohol,andmarijuana use indexes were combined for analysis in acomposite scoreorlatent construct, consistent with previousresearch (Needle, Su, &Lavee,1989; Newcomb & Bender,1988).AnalysesofTPQ Constructs

Descriptive statistics for the TPQ constructs indicated thatthey were normally distributed (skewness: -0.07 to0.30).Most scale intercorrelations were low (.00 to .09), theexceptions being those between novelty seeking and harmavoidance (.34) and between novelty seeking and socialreward dependence (.26). Although novelty seeking andharm avoidance have been shown to have slight inverserelationships in adult samples (Cloninger, Przybeck, &Svrakic, 1991), they have been positively correlated amongadolescents (Wills et al., 1994).Preliminary analyses indicated that correlations for theCloninger constructs were quite similar for indexes oftobacco, alcohol, and marijuana use and heavy drinking;

thus, analyses were performed for the composite substanceuse score. Zero-order correlations of the TPQ constructswith the substance use score were significant with oneexception (harm avoidance-substance use; r = .03, ns).Multiple regression with the constructs entered simulta-neously indicated significant main effects for allco nstructs,with positive coefficients fo r novelty seeking { J = .30,p


6/13

SUBSTANCE USELEVELAND PROBLEMS 127Table2Substance Use byLevelso f Novelty Seeking, Harm Avoidance (HA),an dReward Dependence

Noveltyseeking

LevelLo w High

LowH A High HA Low HA HighH A

LowHigh

LowHigh

7.866.71

6.967.78

Taskreward dependence6.286.32

Social reward dependence5.557.17

10.548.87

8.3710.82

10.308.46

8.889.56Note. Criterion variableiscomposite substanceusescore. Valuesareregression-estimated levelsofsubstanceuse forcasesatM 1SDo nrespective predictor variables.

greaterat ahigh levelofsocial reward dependence, (b) thepositive relation between social reward dependence andsubstance use wasgreaterat a lowlevelofharm avoidance,and (c) the inverse relation between harm avoidance andsubstanceuse wasgreaterat a lowlevelofnovelty seeking.Resultsinvolving task reward dependence were consistentinform with previous research (Willset al., 1994, 1998), butthe analyses involving social reward dependence werenottotally consistent with previous findings.Confirmatory Model

Wefirstanalyzed aco nfirmatorymodel based on specifi-cationof themeasurementof theconstructs.Forexogenousvariables, parental supportwasspecifiedas alatent constructmeasured by two indicators, emotional support an d instru-mental support. Parental substanceuse was alatent constructmeasured by two indicators, parental beer-wine use andparental liquoruse;parentalsmokingwas droppedbecausepreliminary analyses indicated thatit did notload highlyonthe latent construct.The other exogenous measures (TPQconstructs, parent-child conflict, and family events) weremanifestvariables measuredby asingle indicator.In themeasurement model fo rendogenous variables, theconstruct for good control was based on three indicators(dependability, attentional control, and behavioral coping);soothabilitywasdropped becauseof alower loadingon thelatent construct. The latent construct fo r poor control hadfour indicators (impatience, distractibility, impulsiveness,and anger coping),and the risk-taking constructhad threeindicators parceled from the scale items. Measures fo rsmoking motives and alcohol motives were combinedbecause they showed similar correlations withthecriterionvariables. Social motives were excluded from the analysison theoretical grounds, so the latent construct fo r copingmotiveshadthree indicators (enhancement, boredom relief,andstress reduction).T helatent constructforsubstanceuseproblems was based on three indicators fo r institutional,excessive use, and control problems; the measure fo rinterpersonal problems was dropped because of a lowloading on the construct. Other constructs, consistent with

prior research (Wills,DuHamel, &Vaccaro,1995; Willsetal.,1998), werealatent constructforpeersubstanceusewiththree indicators (friends' cigarette, alcohol, andmarijuanause) and alatent constructfo radolescents'levelo fsubstanceusebasedonfourindicators (frequencyofsmoking, alcoholuse, marijuana use,andheavy drinking).The other endog-enous variable, adolescent life events, was a manifestvariable measuredby asingle indicator.Aconfirmatory model was analyzed with this measure-ment model; covariances were specified among all of theconstructs.Foranalysis, those personswho hadnever usedtobacco or alcohol (n = 194; 16% of the sample) wereexcluded.Themodel withnocorrelated errors providedanadequate fit to the data, \2(414, N =977) = 1,502.96,comparative fit index (CFI) = .93. Addition of correlatederror terms resulted in a good fit, X2(397, N =977) =813.83,CFI = .97.2Correlations amongthestudy constructsare presented in Table 3. Novelty seeking was correlatedwith poor self-control and risktaking, whereas harm avoid-ance was correlated with poor self-control but was notrelated to risk taking. Task reward dependence was corre-lated with good self-controlbut was notsubstantially relatedtoany of the riskfactorvariables. Social reward dependenceshowedadifferentpattern, being somewhat correlated withboth risk factors and protective factors. Thus, each of theTPQ constructs showeda distinctive pattern of correlationwith other variables. Correlations for the other study con-structs were consistent with prediction: Good self-controlwa sinversely correlated withtheproximalriskfactors (lifeevents, coping motives, and friends' substance use), andpoor self-controland risktaking were positively correlatedwith these factors.The proximal risk factors were moder-

2W eused conservative criteria fo rmodel modification: Corre-lated errors were introducedonlyiftheyh ad modification indexesgreater than 15 (p


7/13

128 WILLS, SANDY, AND SHINARTable3RelationshipsAmong ConstructsinConfirmatory Model

Construct 1 10 11 12 13 14 151. Novelty seeking2. Harm avoidance3. TaskRD4. Social RD5. Parent support6. Parent conflict7. Parent substance use8. Family events9. Good control10. Poor control11. Risk taking12. Adolescent events13.Coping motives14. Peersubstance use15. Substance use level16. Substance use problems

.35.01.21-.14.34.13.13-.16.66.56.25.39.28.36.32

.09.09-.15.26.14.23-.04.57.07.21.24.10.05.15

.06

.21-.14.06.02.73-.13-.08-.16-.14-.13-.21-.13

.06.10.00.00.18.14.14.09.19.12.18.14

-.49-.13-.14.43-.30-.16-.26-.20-.19-.22-.15

.15.16-.27.49.26.27.28.26.29.22

.15-.06.22.24.18.29.15.24.26

-.06.24.02.33.13.13.10.18

-.44-.28-.26-.18-.25-.32-.24

.54.45.57.39.49.53

.25.43.35.48.40

.30.29.37.41

.35.76.68.61 .39 .78

Note. Coefficientsarestandardized covariances(i.e.,correlations). RD = reward dependence.

ately to strongly correlated with substance use level andsubstance use problems.Test of theMediational Model

Predictionsaboutthe mediational processweretested in astructural model with substance use level and problems ascriterion variables. The structural model was specified withthe four TPQ measures as exogenous variables (those notcausedby any prior variable in the model). Parental support,parent-adolescentconflict, parental substance use,andfam-ilylife events were included as exogenous variables so thateffects for TPQ constructs would be independent of anyassociations with familial characteristics. Endogenous vari-ables (those specified as being caused by a prior variable inthe model) were specified with good self-control, poorself-control, and risk-taking tendency subsequent to theexogenous variables. These three constructs were specifiedwith covariancesfreelyestimated among theirerror terms.The adolescentlifeevents measure was specified subsequentto the self-control and risk-taking constructs, with regressioneffects allowed from prior constructs. Coping motivesandpeer substance use were specified subsequent to theseconstructs, with a covariance specified between their errorterms. The construct of substance use level was specifiedprior to the construct of substance use problems (cf.Newcomb,1992).

The structural model was analyzed withLISREL 8.12viathe maximum-likelihood method (Joreskog & Sorbom,1988). An initial model was estimated using a minimal set ofpaths predicted from theory and prior research (MacCallum,Roznowski, & Necowitz, 1992).3 Additional coefficientswere introduced on the basis of modification indexes (seeFootnote 2). The final model is presented in Figure 1;valuesfor measurement modelparametersand covariances amongexogenous constructs (included in the model but not repre-sentedin the figure) arepresented separatelyinTables3 and4. The model had a reasonable fit to the data, x2 (453,N =977) = 960.45, CFI = .97, X2/df ratio = 2.12. The

model accounted for 72% of the variance in substance uselevel and61 %of the variance in problems. This model wasreplicated with demographic controls, including indexesforgender,ethnicity, family structure, and parental education asexogenous variables. Results for the model with demo-graphic controls were quite similar to those shown in Figure1.4Effects for the TPQ constructs were consistent withpredictions (p


8/13

SUBSTANCEUSE LEVELAND PROBLEMS 9

T SK R E W R DD E PE ND E NCE l .is

Figure 1. Structural model forTridimensionalPersonality Questionnaire constructs and parentalvariables, self-control and risk-taking constructs, adolescent (ADOL) life events, peer substance(SUB) use, and motives for substance use, with criterion constructs of substance use level andsubstance use problems. All values are standardizedco efficients; all coefficients are significant. Fo rmeasurementmodel parameters an dcovarianceso fexog enous co nstructs, includedin themodelbutno t represented here, see Tables 3 and 4. Double-headed arrows indicate correlations betweenresidual terms. Values incirclesare squared multiple co rrelations, indic ating variance acco unted forin each construct.SOC = social.

should be noted that all reported results represent indepen-dent effects, and results for TPQ constructs controlled fo rany association with family characteristics.) Other resultsindicated that harm avoidance had an inverse path to risktaking and an inverse direct path to substance use level. Taskreward dependence had a positive path to good control andan inverse pathto poor control. Social reward dependenceevidenced somewhat paradoxical results, with a positivepath to good control as well as a positive direct path tocoping motives (p


9/13

130 WILLS, SANDY,AND SHINARTable4Measurement Model Parameters

Constructand indicators LoadingParentalsupportEmotionalInstrumentalParental alcoholu se

Beer-wineLiquorGood controlDependabilityAttentionalcontrol3BehavioralcopingPoor controlImpatienceDistractibilityaImpulsiveness3Anger copingRisk takingIndicator1Indicator2Indicator3Coping motivesSelf-enhancementBoredom reliefStressreductionPeer substanceuseFriends smokingFriends drinkingbeer-wineFriends using marijuanaSubstance uselevelSmokingAlcoholuseMarijuanauseHeavydrinkingSubstanceuseproblemsInstitutionalExcessControl

.89.80

.75.71

.57.58.72

.60.72.74.71

.90.89.86

.79

.77.87

.83.83.79

.68.65.64.60

.68.85.69Note. Valuesarestandardizedcoefficients (i.e., factor loadings). Indicatorwithoneitem dropped (see footnote3).

DiscussionThis study tested a theoretical modelof the relationshipbetween Cloninger's (1987a) constructs and adolescent

substance usewith data froma largerepresentativesample.Thebasicfindings were consistent with previous research:Novelty seeking and social reward dependence were posi-tively relatedtosubstance use; task reward dependence wasinverselyrelated tosubstance use;a ndharm avoidance hadcomplex effects (Masse & Tremblay, 1997; Sher, Wood,Crews,&Vandiver,1995; Willset al., 1994, 1998). Noveltyseekingwas thestrongest predictorofsubstance use,aresultthat converges with other findings from human research(Howard etal., 1997)andresearch ondrug self-administra-tion in animals (Bardo et al., 1996; Piazza Le Moal,1996).Thesimilarfindingsobtainedfortobaccoandalcoholuse, consistent with recentpsychopharmacologicalresearch(e.g., Koob& Le Moal, 1997;Pontieri, Tanda, Orzi,& DiChiara, 1996), suggest that Cloninger's theory is alsoapplicabletootheraddictivesubstances.Epigenetic theory predicts that relationships of Clon-inger's constructs willbeindirect (Tarter, Moss, Vanyukov,1995), and the results were generally consistent with thisprediction. Effects fo rnovelty seeking andharm avoidancewere primarily mediated through poor self-control, whereaseffects fo rtask reward dependence were primarily mediatedthrough good self-control. The self-control constructs, inturn, were related to proximal risk factors fo r adolescentsubstance use, including negative life events and deviantpeer affiliations. These findings are consistent with recenttheory from clinical and psychopharmacological researchsuggesting that self-regulation ofcognition andemotion isan important underlying concept in vulnerability or resis-tance to substance use(Koob & LeMoal, 1997; Miller&Brown,1991; Newman& Wallace, 1993). Family variableswere also important predictors, consistent with theory

Results for thecriterion constructs indicated that adoles-centlifeeventshad apathtodeviant peer affiliations aswellas direct paths to substance use level and substance useproblems (cf. Castro et al., 1987; Wills, McNamara, &Vaccaro, 1995).5The hypothesis about differential predic-tion received support; coping motives had a path to sub-stance use level ((3= .66) and also had a direct path toproblems (P = .29), whereas peer affiliation had a signifi-cant path only to level of use ((3 = .36). Neither goodself-control nor poor self-control had any direct paths tosubstanceu selevelorproblems; therelation ofself-controlconstructs to criterion variables was indirect, mediatedthrough life events, peer affiliations, and coping motives.Thus,th eresults showedthatahigher levelo fsubstanceus ewaspredicted (inorder ofmagnitude)b ythree riskfactors:coping motives, deviant peer affiliations, andnegative lifeevents. In addition, substanceu seproblems were predictedbyadirecteffect fromcoping motives. Thus, coping motiveshaddualeffects onproblems,inpart through contributingtoahigher levelof use and inpart through adirect effect (cf.Cooperetal.,1995).6

5Thepath betweenlifeeventsand problems((3= .16,p


10/13

SUBSTANCEUSELEVELAND PROBLEMS 131positing an interplay between heritable characteristics andenvironmental factors (Scarr, 1992; Tarter, Moss, &Vanyukov, 1995).Theconstruct of harm avoidancehad complex effects inthe multivariate structural model. Harm avoidance wasrelated to poor self-control, but, at the same time, it wasinversely relatedto risktakingand had aninverse direct pathtosubstanceuselevel.Webelieve that these effects derivefrom twodifferentaspectsof theharm avoidance dimension.One is that harm avoidance theoretically derives from thebehavioral inhibition system (Cloninger, 1987a;Gray, 1991).Thus,individuals high on harm avoidance, havinggreatersensitivitytocuesofthreat, wouldbeless likelytoengageinactivities perceived as dangerous; this accounts for theinverse paths to risk taking and substance use. However,developmental theory also suggests that high levels ofanxiety will interfere with the development of self-regulation skills(Rothbart & Ahadi,1994); thisisconsistentwith theobserved path topoor self-control. These differentaspects become evident only when relationships of harmavoidance witharangeofother constructsa reincludedin amultivariate analysis.Asimilar paradoxical effect wasnotedfor social reward dependence, which was related to goodself-control(possiblybecauseo fbetter socialization throughsocial engagement) bu t also wasrelated to coping motivesfor substance use (possibly because gregarious individualsare more likely to become involved with peers who aresmokinganddrinking; Wills,Mariani, &Filer, 1996). Thus,theoryshould consider thepossibility thatagiven personal-itydimensionm ayhave complexeffects onsubstance use.Theresults showed that risk-takingtendency representsadistinct pathway in substanceuse etiology, independentofpoor control. The relationship of novelty seeking to risk-taking tendency suggests adispositional component tothisattribute (cf. Cloninger et al., 1993; Zuckerman, 1994b).Risk takingw asrelated tocoping motivesforsubstance use,possibly reflecting a greater physiological sensitivity toeffects of alcohol and other substances (cf. Earleywine,1994; Nagoshi, Wilson, & Rodriguez, 1991). Risk takingwa sindependently relatedtodeviant peeraffiliations, consis-tent with theoretical suggestions that dispositional factorsactthrough a systematic influence on choices of compan-ions,with some individuals gravitating into groupso fpeerswho share an inclination toward exciting, bu t possiblyillegal, activities (Scarr & McCartney, 1983). Thus, indi-vidual-level and social-level processes are not mutuallyexclusive and may act in concert to influence liability fo rsubstance use (Tarter, Moss, & Vanyukov, 1995; Wills,DuHamel,&Vaccaro, 1995).Althoughmaineffects fo r Cloninger's (1987a) constructswereconsistent with previous research,we did notreplicateall of the interaction effects observed in previous studies(Wills et al., 1994, 1998) with younger adolescents. Thedifferingresultsm ayderivefrom thefactthat participantsinprevious studies were approximately 13 years of age,whereas participants in the present study were approxi-mately 16yearsofage; these representdifferentstagesin theonsetandescalationo fsubstance use,and it ispossible thatdifferent aspects of the constructs (e.g., dysregulation vs.

inhibition) predominate at different ages. Further researchwith age-stratified samples would be desirable to clarifythese issues.Some aspects of this study could be noted as possiblelimitations. Themeasures ofsubstanceusewerenotidenti-cal to diagnoses of substance abuse or dependence, andstudies with more intensive assessmentof substance abusewould be desirable. The data were obtained through self-report, and although adolescent self-reports fo r such vari-ables have been corroboratedbybiochemical measuresandindependent teacher ratings (Wills & Cleary, 1997; Wills,DuHamel, & Vaccaro, 1995), research with data frommultiple sources is warranted. Finally, the present studyfocused on testing a theoretical model in cross-sectionaldata; longitudinal research issuggested totestfo rtemporalrelationships and possible reciprocal effects among self-controlconstructsandother variables. f f e c t s for CopingMotives

This study provided a strong test of the role of copingmotives because negative life events an d deviant peeraffiliations, previously demonstrated to be important riskfactors fo radolescent substanceuse (Hawkins etal., 1992;Wills, 1990), were also included. Results confirmed thesignificance ofcoping motives, which had a strong path tosubstance use level as well as problems; in comparativeterms, this effect was greater than that fo r affiliation withpeersubstance users, whichhas previously been suggestedasth eprimary proximal factorfo radolescent substance use.The present study was conducted with older adolescents,and those with higher levels of substance use have anextensivehistory ofexperience with tobaccoandalcohol.Itispossible that peeraffiliations aremore importantatearlierages, and the significanceof coping motives emerges onlywhen individuals have had multiple opportunities to usesubstances in a manner that makes the coping-relatedfunctionsof usesalient (Mosbach & Leventhal,1988; Wills&Cleary, 1995).Thedirect path fromcoping motivestoproblems associ-atedwithsubstanceusederives,wethink,from theground-ing ofcoping motivesinpoor self-control, combined withaneffect from risk-taking tendency. Among individuals ele-vated on these dimensions, substance use may lead toproblems because they have less restraint about enteringsituationsthat turnout to beproblematicand aremore likelyto act in animpulsiveorangry mannerinthose situations,assuggested by clinical literature indicating that difficulty indealing with anger is a common problem among drugabusers (cf. Khantzian, 1990; Miller & Brown, 1991).Althoughdeviant peeraffiliations are asignificantriskfactorfor levelofuse, thepresent data suggest that substanceuseproblemsreflectmore thananunwise choiceofcompanions.Psychopharmacology research has generally focused onrewarding effects of drugs of abuse, whereas the presentmotive measures focused more on functions that involveaffect regulation an dstress reduction.W enote that adoles-cents with high levels of substance use also have poorself-regulation and high levels of life stress(Wills, 1990;


11/13

132 WILLS, SANDY,AN D SHINARWills, McNamara, & Vaccaro, 1995), and hence copingfunctionsof substance use should be particularly salient forthis part of the population (cf. Koob & Le Moal, 1997).Another issue is that mood enhancement effects of drugsmay act tooffset the impact of negative experiences (Cooperet al., 1995; Wills, Sandy, Shinar, & Yaeger, in press), soboth positive affectenhancement and negative affect reduc-tionmay beoperativefor producing highlevelsof substanceuse.Implicationsfor Prevention

Epigenetic approaches emphasize that although liabilityto substance use may have a dispositional component, theeffects ofdispositional constructsareshapedbyenvironmen-tal factors (Rothbart & Ahadi, 1994; Tarter, Moss, &Vanyukov, 1995). Moreover, these theories specificallydelineate aspects of self-control that are modifiable, such asfocusingattention on tasks, delaying gratification of a desire,or using verbal cues to restrain impulsive behaviors in apotential problem situation. Thus, epigenetic theory givesconcrete guidance for prevention programs designed forhigh-risk individuals so that existing risk factors such asnovelty seeking will have less impact on the onset ofproblem behavior.

The effects found for coping motives indicate that atten-tion to beliefs about functions of substance use should beconsidered as a component in prevention programs, inaddition to peer influences (Stacy et al., 1991, 1993;Sussman & Johnson, 1996; Wills & Cleary, 1995). Theseresults suggest that treatment programs for substance abus-ers might include self-control training that targets responsepatterns such as anger proneness and impulsive responding.Inthis way, programs might reduce the harm associated withsubstance abuse among individuals who do not succeed inachieving complete abstinence (Miller & Brown, 1991).

ReferencesAchenbach, T. M. (1991). Manualfor the Youth Self-Report and1991profile. Burlington: UniversityofVermont.Bailey, S. L., & Rachal, J. V. (1993). Dimensions of adolescentproblem drinking.Journalof Studies on Alcohol, 54,555-565.Bardo, M. T., Donohew, R. L., & Harrington, N. G. (1996).Psychobiology of novelty seeking and drug seeking behavior.Behavioural Brain Research, 77,23 43.Barrera,M.,Jr.,Chassin,L. ,&Rogosch,F.(1993).Effects ofsocialsupportand conflict onadolescentchildren.Journal o f Personal-ity andSocial Psychology,64 ,602-612.Bentler, P. M. (1992). EQ S structural equations program manual.

Los Angeles: BMDP Statistical Software.Block, J. H., & Block, J. (1980). The role of ego-control andego-resiliency in the organizationofbehavior.InW . A.Collins(Ed.), Minnesota Symposium on Child Psychology (Vol.13, pp.39-101).Hillsdale, NJ:Erlbaum.Buss, A., & Plomin, R. (1984). Temperament: Early developingpersonality traits.Hillsdale,NJ: Erlbaum.Castro, F.G., Maddahian, E.,Newcomb, M. D., & Bentler, P. M.(1987). A multivariate model of the determinants of cigarettesmoking among adolescents. Journal of Health and SocialBehavior,28,273-289.

Chassin, L. A., Pillow, D. R.,Curran,P. J., Molina, B., & Barrera,M. (1993). Relation of parental alcoholism to early adolescentsubstanceuse:Atest ofthree mediating mechanisms.JournalofAbnormalPsychology, 102,3-19.Chou, C.-P., & Bentler, P. M. (1995). Estimates and tests instructural equation modeling. In R. H. Hoyle (Ed.), Structuralequation modeling: Concepts, issues, and applications (pp.37-54).ThousandOaks,CA:Sage.Chou, C.-P., Bentler, P. M. , & Satorra, A. (1991). Scaled teststatistics and robust standard errors fo r non-normal data incovariance structure analysis: A Monte Carlo study. BritishJournal of Mathematical and Statistical Psychology, 44 , 347-357.Cloninger, C. R. (1987a).Neurogenetic adaptive mechanisms inalcoholism. Science,236,410 416.Cloninger, C. R. (1987b). A systematic method for clinicaldescription and classificationofpersonality variants.Archives ofGeneral Psychiatry,44,573-588.Cloninger, C. R., Przybeck, T. R., & Svrakic, D. M. (1991). TheTridimensional Personality Questionnaire: U.S.normative data.Psychological Reports,6 9,1047-1057.Cloninger, C. R.,Sigvardsson, S., &Bohman,M. (1988). Child-hood personality predicts alcohol abuse in young adults.Alcohol-ism: ClinicalandExperimental Research,12 ,494-505.Cloninger, C. R., Svrakic, D. M., &Przybeck,T. R. (1993).Apsychobiological model o f temperament and character.Archiveso f General Psychiatry, 50,975-990.Cooper, M. L. (1994). Motivations for alcohol use among adoles-cents: Development and validation of a four-factor model.Psychological Assessment, 6,117-128.Cooper, M. L., Frone, M. R., Russell, M., & Mudar, P. (1995).Drinking to regulate positiveand negative emotions: A motiva-tional model of alcohol use.Journal of Personality and SocialPsychology, 69 ,990-1005.Cooper, M. L., Russell, M. , & George, W. H. (1988). Coping,expectancies, and alcohol abuse.Journal of Abnormal Psychol-ogy, 97 ,218-230.

Earleywine, M .(1994).Personality riskforalcoholism and alcoholexpectancies.Addictive Behaviors,1 9,577-582.Eysenck, S. B. G., & Eysenck, H. J. (1977). The place ofimpulsiveness in a dimensional system of personality descrip-tion. British Journal of Social and Clinical Psychology, 16 ,57-68.Gray, J. A. (1991). The neuropsychology of temperament. In J.Strelau & A. Angleitner (Eds.), Explorations in temperament:Perspectives on theory and measurement (pp.105-128). NewYork:Plenum.Hagekull,B .(1989).Longitudinal stability of temperament withina behavioral style framewo rk. In G. A.Kohnstamm,J. E. Bates,& M. K. Rothbart (Eds.), Temperament in childhood (pp.283-297).N ew Y ork: Wiley.Harter, S. (1985). Manual for the Self-Perception Profile forChildren.Denver,CO:UniversityofDenver.Hawkins, J. D., Catalano, R. F., &Miller, J. Y.(1992). Riskandprotective factors for drug problems in adolescence and earlyadulthood.Psychological Bulletin,111,64-105.Hays, R. D., Widaman, K. F.,DiMatteo, M. R., & Stacy,A. W.(1987).Structural equation models of current druguse.JournalofPersonality and Social Psychology, 52,134-144.Heath,A. C., Cloninger, C. R., & Martin, N. G. (1994). Testing amodel for the genetic structureofpersonality:Acomparisonofthe personality systems of Cloninger and Eysenck. Journal ofPersonality and Social Psychology,66 ,762-775.


12/13

SUBSTANCE USE LEVEL AND PROBLEMS 133Howard, M. O.,Kivlahan,D., & Walker, R. D. (1997). Cloninger'stridimensional theory of personality andpsychopathology:Appli-cations to substance usedisorders.JournalofStudiesonAlcohol,58,48-66.Johnson,V.,&Pandina,R. J.(1993).Alongitudinal examinationofthe relationships among stress, coping strategies, and problemsassociated withalcoholuse.Alcoholism: Clinicaland Experimen-

talResearch, 17,696-702.Johnston, L. D., O'Malley, P. M. , & Bachman, J. G. (1995).National survey results on drug use from theMonitoring theFuture study, 1975-1994: Vol.1. Secondary school students.Rockville, MD:National InstituteonDrug Abuse.Joreskog, K.G., &Sorbom, D. (1988).LISREL 7: Aguide to theprogramandapplications.Chicago: SPSS.Kandel, D., & Davies, M. (1992). Progression to regular mariju anainvolvement: Phenomenology and risk factors for near-daily use.In M. Glantz & R. Pickens (Eds.), Vulnerability to drug abuse(pp.211-253).Washington,DC :American Psychological Asso-ciation.Kendall, P. C., & Wilcox, L. E. (1979). Self-control in children:Development of a rating scale. Journal of Consulting andClinicalPsychology, 47, 1020-1029.Kendall, P.C.,& Williams, C. L. (1982). Assessing the cognitiveandbehavioral componentso fchildren's self-management. In P.Karoly & F. H. Kanfer (Eds.), Self-management and behaviorchange(pp. 240-284).NewY ork: PergamonPress.Khantzian,E. J.(1990).Self-regulation and self-medication factorsin alcoholism and the addictions. In M. Galanter (Ed.), Recentdevelopments in alcoholism(Vol.8, pp.255-271).NewYork:Plenum.Koob, G. F., & Le Moal, M. (1997). Drug abuse: Hedonichomeostatic dysregulation.Science,278,52-58.MacCallum, R. C., Roznowski, M. , & Necowitz, L. B. (1992).Model m odifications incovariancestructure analysis.P sychologi-calBulletin, 111,490-504.Maggs, J. L.,Almeida, D.M., &Galambos, N. L.(1995).Riskybusiness: The paradoxical meaning of problem behavior foryoungadolescents. Journal of Early Adolescence, 15 ,344-362.Masse, L. C., & Tremblay, R. E. (1997). Behavior of boys inkindergarten and the onset of substance use during adolescence.Archives of General Psychiatry,54,62-68.Miller, W. R., & Brown, J. M. (1991). Self-regulation as aconceptual basis for the prevention of addictive behaviours. InN. Heather, W. R. Miller, & J. Greeley (Eds.), Self-control andthe addictive behaviours(pp. 3-79). Sydney, New S outh Wales,Australia: MaxwellM acmillan.Mosbach, P., & Leventhal, H. (1988). Peer group identification andsmoking.Journal o fAbnormal Psychology,97,238-245.Murray, D. M., & Perry, C. L. (1987). Measurement of substanceuse among adolescents: When is the bogus pipeline methodneeded1 Addictive Behaviors,12 ,225-233.Nagoshi, C. T, Wilson, J. R., & Rodriguez, L. A. (1991).Impulsivity, sensation seeking, and behavioral andemotional

responses to alcohol. Alcoholism: Clinical and ExperimentalResearch,15,661-667.Needle, R., Su, S., & Lavee, Y. (1989). A comparison of theempirical utility of three composite measures of adolescent druginvolvement. Addictive Behaviors, 14,429-441.Newcomb, M. D. (1992). Understanding the multidimensionalnature of drug use and abuse. In M . Glantz & R. Pickens (Eds.),Vulnerability to drug abuse (pp. 255-297). Washington, DC:American Psychological Association.Newcomb, M. D., & Bentler, P. M. (1988). Consequences ofadolescentdruguse. NewburyPark, CA:Sage.Newcomb, M.D .,Chou,C .,Bentler, P. M., & Hu ba, G. J. (1988).

Cognitive motivations for drug use am ong adolescents. Journalo f Counseling Psychology,35,426 -38.Newcomb, M. D., & Harlow, L. L. (1986). Life events andsubstance use among adolescents. Journal of Personality andSocial Psychology,51,564-577.Newman, J. P., & Wallace, J. F. (1993). Diverse pathways todeficientself-regulation: Implications fordisinhibitory psychopa-thology in children. Clinical Psychology Review, 13 ,699-720.Pedlow,R .,Sanson,A. ,Prior, M., &Oberklaid,F. (1993). Stab ilityof maternally reported temperament from infancy to 8 years.Developmental Psychology,29 ,998-1007.Piazza, P. V, & Le Moal, M. (1996). Pathophysiological basis ofvulnerability to drug abuse: Role of an interaction betweenstress, glucocorticoids, and dopaminergic neurons. Annual Re -viewof Pharmacology and Toxicology, 36 ,359-378.Pontieri,F.E.,Tanda,G .,Orzi,R, & DiChiara,G .(1996). Effectsofnicotine on the nucleus accumbens and similarity to those ofaddictive drugs.Nature,382,255-257.Pulkkinen, L., & Pitkanen,T. (1994). A prospective study of theprecursors toproblem drinking in young adulthood. Journal ofStudiesonAlcohol, 55,578-587.Robins,L .N .,&Przybeck,T. R. (1985).Age ofonset ofdruguseas a factor in drug and other disorders. In C. L. Jones & R. J.Battjes (Eds.),E tiology of drug abuse(pp. 178-192).Rockville,MD: Natio nal Institute on Drug Abuse.Rothbart, M. K. , & Ahadi, S. A. (1994). Temperament and thedevelopment of personality. Journal of Abnormal Psychology,103,55-66.Rothbart, M. K., Derryberry, D., & Posner, M. J. (1994). Apsychobiological approach to thedevelopment of temperament.In J. E. Bates & T. D. Wachs (Eds.), Temperament: Individualdifferences at the interface o f biology andbehavior(pp. 83-116).Washington, DC: American Psychological Association.Scarr, S. (1992). Developmental theories for the 1990s: Develop-ment and individual differences. ChildDevelopment,6 3, 1-19.Scarr, S., & McCartney, K. (1983). How people make their ownenvironments: A theory of genotype environment effects.ChildDevelopment,54,424-435.Scheier, M. F, & Carver, C. S. (1985). Optimism, coping, andhealth: Assessment and implications of generalized outcomeexpectancies.H ealth Psychology, 4,219-247.Sher, K. J., & Trull, T. (1994). Personality and disinhibitorypsychopathology: Alcoholism and antisocial personality disor-der.Journalof Abnormal Psychology,103,92-102.Sher,K. J., Wood, M.D. ,Crews, T. M., & Vandiver, P. A. (1995).Tridimensional Personality Questionnaire: Reliability/validitystudies. Psychological Assessment, 7,195-208.Smith, G. T, McCarthy, D. M. , & Goldman, M. S. (1995).Self-reported drinking and alcohol-related problems amongearlyadolescents: Dimensionality and validity over 24 months.Journal of Studieson Alcohol,56,383-394.Stacy,A .,New com b, M. D., & B entler, P. M. (1991). Personality,problem drinking,andd runk driving.JournalofPersonality andSocial Psychology, 60,795-811.Stacy, A., Newcomb, M. D., & Bentler,P. M. (1993). Cognitivemotivations and sensation seeking as long-term predictors ofdrinking problems. Journal of Social and Clinical Psychology,12, 1-24.Stallings, M.C., Hewitt, J. K. ,Cloninger, C.R. , Heath,A.C.,&Eaves, L. J. (1996).Genetic and environmental structure of theTridimensional Personality Questionnaire: Three orfour tempera-ment dimensions? Journal of Personality and Social Psychol-ogy, 70 ,127-140.


13/13

134 WILLS, SANDY,A NDSHINARSlice, E., Myers, M. G., & Brown, S. A. (1998). Relations ofdelinquency to adolescent substance use and problem use.Psychology ofAddictiveBehaviors,1 2 ,136-146.Stuss,D. T.(1992). Biological andpsychological developmento fexecutive functions.BrainandCognition,20,8-23.Sussman, S., & Johnson, C. A. (Eds.). (1996). Drug abuseprevention: Program and research recommendations [Specialissue}.American Behavioral Scientist,39,787-942.Swaim, R. C. , Getting, E. R., Edwards, R. W., & Beauvais,F.(1989). Links from distress to adolescent druguse.Journal ofConsultinga ndClinicalPsychology, 57 ,227-231.Tarter, R. E. (1988). Are there inherited behavioral traits thatpredispose to substance abuse? Journal of Consulting an dClinical Psychology, 56,189-196.Tarter,R.E.,Mezzich,A.C. ,Hsieh, Y-C.,&Parks,S. M. (1995).Cognitive capacityinfemale adolescent substance abusers.DrugandAlcohol Dependence,39,15-21.Tarter,R.E.,Moss,H. B., &Vanyukov,M. M.(1995). Behaviorgenetic perspectiveofalcoholism etiology.In H.Begleiter & B.

Kissin (Eds.),Alcohol and alcoholism (Vol.1, pp . 294-326).New York: Oxford UniversityPress.Tzelgov,J., &Henik,A.(1991).Suppression situationsinpsycho-logical research: Definitions, implications, and applications.PsychologicalBulletin,109,524-536.U.S. Department of Commerce. (1992). 7990 census of thepopulation: General population characteristics (New York).Washington,DC:U.S.Government Printing Office.U.S. Department of Health and Human Services. (1988). Thehealth consequencesofsmoking: Nicotine addiction.A report ofthe surgeon general (DHHS Publication No. CDC 88-8406).Washington, DC:U.S.Government Printing Office.West,S.G.,Finch,J. E, &Curran,P. J.(1995). Structural equationmodels with nonnormalvariables. In R. H.Hoyle(Ed.),Struc-turalequation modeling: Concepts, issues,andapplications(pp.56-75).Thousand Oaks,CA:Sage.White,H.R.,&Labouvie,E. W.(1989).Towardstheassessmentofadolescent problem drinking.Journalo fStudieson Alcohol, 50 ,30-37.Wills,T. A.(1990). Stressandcoping factorsin theepidemiologyof substance use. In L. T. Kozlowski, H. M. Annis, H. D.Cappell, F. B. Glaser, M. S. Goodstadt, Y.Israel, H.Kalant,E. M. Sellers, & E. R. Vinglis (Eds.), Research advances inalcohol and drug problems(Vol.10, pp. 215-250).N ewYork:Plenum.Wills,T. A.(1994). Self-esteemandperceived controlinadolescentsubstanceuse:Comparative testsinconcurrentandprospectiveanalyses.PsychologyofAddictive Behaviors,8,223-234.Wills, T. A., & Cleary, S. D. (1995). Stress-coping model fo ralcohol-tobacco interactions in adolescence. In J. B. Fertig &J. P.Allen (Eds.),Alcohol and tobacco: From basic science toclinicalpractice (pp. 107-128).Bethesda,MD:National Insti-tuteonAlcohol AbuseandAlcoholism.Wills,T.A.,&Cleary,S. D.(1997).T hevalidityo fself-reportsofsmoking: Analysesbyethnicityin asampleofurban adolescents.American JournalofPublic Health,87 ,56-61.Wills, T.A., DuHamel, K., & Vaccaro, D. (1995). Activity andmood temperament as predictors of adolescent substanceuse:Test of aself-regulationmediationalmodel.JournalofPersonal-ity an dSocialPsychology, 68 ,901-916.Wills,T.A.,&Hirky,A. E.(1996). Copingandsubstance abuse.InM.Zeidner& N. S.Endler (Eds.),Handbooko fcoping: Theory,research,an dapplications (pp. 279-302).NewYork: Wiley.Wills,T.A.,Mariani,J.,&Filer,M.(1996).Theroleoffamilyandpeer relationships inadolescent substanceuse.In G. R.Pierce,

B. R. Sarason, & I. G. Sarason (Eds.), Handbook of socialsupport and thefamily (pp. 521-549).N ewYork: Plenum.Wills, T. A., McNamara, G., & Vaccaro, D. (1995). Parentaleducation relatedtoadolescent stress-copingandsubstanceuse.Health Psychology, 14,464-478.Wills,T. A.,McNamara,G.,Vaccaro,D., &Hirky,A. E. (1996).Escalated substanceuse:Alongitudinal grouping analysis fromearly to middle adolescence. Journal ofAbnormal Psychology,105, 166-180.Wills, T. A., Sandy, J. M., Shinar, O., & Yaeger, A. (in press).Contributions of positive and negative affect to adolescentsubstanceuse:Test of abidimensional model in alongitudinalstudy.PsychologyofAddictiveBehaviors.Wills,T.A., & Shiffman, S.(1985). Coping andsubstanceuse: Aconceptual framework. In S. Shiffman & T. A. Wills (Eds.),Coping and substanceuse(pp. 3-24). Orlando,FL:AcademicPress.Wills,T.A.,Vaccaro,D .,&McNamara,G.(1992).Theroleof lifeevents,family support, an dcompetence inadolescent substanceuse: A test of vulnerability and protective factors. AmericanJournal ofCommunity Psychology, 20, 349-374.Wills, T. A., Vaccaro, D., & McNamara, G. (1994). Noveltyseeking, risk taking, and related constructs as predictors ofadolescent substanceuse:Anapplication ofCloninger's theory.Journal ofSubstance Abuse,6 , 1-20.Wills,T.A.,Windle,M.,&Cleary,S. D.(1998). Temperamentandnovelty-seeking in adolescent substance use:Convergence ofdimensions of temperament with constructs from Cloninger'stheory. Journal of Personality and Social Psychology, 74 ,387^06.Windle, M. (1994). Inhibition and activation: Hormonal andpsychosocial correlates and associated psychiatric disorders.Personality andIndividualDifferences, 17 ,61-70.Windle, M. (1995). The approach-withdrawal concept: Associa-tionswith salient constructsincontemporary theorieso ftempera-mentandpersonality development.In K.Hood,G.Greenberg,&E.Tobach(Eds.),Behavioraldevelopment: Concepts of approach-withdrawal and integrative levels (pp. 329-370). New York:Garland Press.Windle, M.,&Windle,R. C.(1996). Coping strategies, drinkingmotives, an d stressful life events among middle adolescents:Associations with emotional an d behavioral problems an dacademic functioning.Journal of Abnormal Psychology,105,551-560.Zucker,R. A.(1994). Pathwaystoalcohol problems:Adevelopmen-ta laccountof theevidence fo rcontextual contributions torisk.In R. A. Zucker, G. M. Boyd & J. Howard (Eds.), Thedevelopment ofalcohol problems(pp. 255-289).Rockville, MD:National InstituteonAlcohol Abusean dAlcoholism.

Zuckerman, M. (1994a). Behavioral expressions and biosocialbases of sensation seeking. New York: Cambridge UniversityPress.Zuckerman, M.(1994b). Impulsive unsocialized sensation seeking.In J. E.Bates& T. D.Wachs (Eds.),Temperament: Individualdifferences at the interface of biology and behavior (pp.219255). Washington,DC:American Psychological Association.

ReceivedApril28,1998Revisionreceived September17,1998AcceptedSeptember17,1998

Novelty Seeking_Cloninger's Constructs Related to Substance Use Level and Problems in Late Adolescence

Documents