A ProspectiveStudy of the Precursors to Problem

Drinkingin YoungAdulthood* LEA PULKKINEN, PH.D.,ANDTUULI PITK•NEN, M.A. Departmentof Psychology, Universityof Jyviiskyld,P.O. Box35, 40351 Jyviiskylii,Finland

ABSTRACT. This study was part of the JyviiskyliiLongitudinal Studyon SocialDevelopment.The subjects(196 males,173females) werestudiedat age8, 14 and26. Threecomponents in drinkinghabits were obtainedat age 26: social,problemand controlleddrinking. Moderateto severeproblemdrinkingwas obtainedfor 26% of the men and 1% of the women, and mild problemdrinkingfor 23% of the men and 15% of the women.Problemdrinking(definedby the CAGE Questionnaire,arrest for alcohol abuseand other indicatorsof

heavydrinking)wasdirectlyaccounted for by poorschoolsuccess at age 14 and,for men,by conductproblemsandlow anxiety.Variables at age 8 that contributedindirectlyto adult problemdrinkingwere

'N HER REVIEW oflongitudinal studies use of ßalcohol, Fillmore (1988) paysattention to on thethe following shortcomingsof longitudinalstudies, among others: (1) the findingsof longitudinalstudies,suchas the relationship between "antisocial behavior" and adult drinking practices,havenot beenreplicatedin contemporarystudies; (2) nonbiologicalstudiesof adolescentshave neglectedto take into seriousaccountthe presenceor absence of parental alcohol problems as precursorsto later alcohol-relatedproblems;and (3) many studiesof unwanteddrinking behaviorsexcludefemales.Fillmore also reveals that subjectshave rarely been youngerthan 13 when the studywas started. The JyviiskyliiLongitudinalStudy on Social Developmentmadeit possibleto examinebehavioralprecursors of adult problemdrinking in a Finnishsampleof men and women. Behavioral measures were taken from early schoolage and adolescence,and data on parentalalcohol problemswere also available. In the presentstudy,these data were used to analyze (1) individual differences,includinggenderdifferences,in a youngadult'sdrinkingbehavior; (2) behavioral precursorsof adult problem drinking; and (3) the effects of parental alcohol-related problemson offsprings'useof alcohol.

Received:October7, 1992. Revision:July 8, 1993. *This studywas originally presentedat a conferenceon "Addictive Behaviour:Determinantsfor Long-termCourse" held in Zurich, Switzerland,September23-25, 1992. Financialsupportwas providedby the Academyof Finlandand the FinnishFoundationfor AlcoholStudies. 578

aggression, low anxiety,low prosociality andpoorschoolsuccess for men, and high anxietyand poor schoolsuccess for women.Women andmendifferedin theeffectof socialanxiety;in men,anxietywas a protectivefactoragainstproblemdrinking;in women,it wasa risk factor.Althoughconductproblemsoften precedesevereproblem drinking,otherrisk factorsmay existamongsonsof alcoholicparents.Parentaldrinkinghada significantthresholdeffecton maleoffsprings'drinking: if parentaldrinking was low, there was less problemdrinkingamongthe male offspringthan if a mild dependenceon alcoholwas observedin the parents.(J. Stud.Alcohol 578-587, 1994)

The prevalenceof drinkingincreases considerably from early to late adolescence(Pulkkinen and Narusk, 1987;

Rahkonenand Ahlstr6m, 1989). The drinkinghabitsof Finnishboysandgirls are rathersimilarat the time of the onset of drinking, but differencesincreasewith age (Rahkonenand Ahlstr6m,1989). Womenare responsible for only one-fifth of the total amount of alcohol consumed.About 28% of men'sdrinkingoccasions resultin drunkenness; for womenthe proportionhasrisensincethe 1960sfrom 6% to 18% (Simpura,1985). Adolescentstudieshaveyieldedfindingson the correlates and predictorsof the use of alcoholand drugsin adolescence.For instance,Jessorand Jessor(1977) dem-

onstratedthat proneness to problembehaviorwas linked with alcohol-relatedproblems, and Shedlet and Block (1990) linked poor impulsecontrolto drug use. Other studiessupportthe ideathatadolescent drinkingis related to undercontrol(e.g., aggressiveness, impulsiveness and lack of concentration)(Anderssonet al., 1989) and to var-

ious"problem"characteristics suchas problembehavior, low schoolperformance and variousformsof delinquent and predelinquentbehavior (Barnes and Welte, 1986; Donovanet al., 1983). The resultsof studiesby Jones (1968), van Kammenet al. (1991), McCord and McCord (1960), Pulkkinen(1982a, 1983)andRobinsandMcEvoy (1990) have also shownthat the early yearsof life influence later alcohol-relatedproblems.Conductproblems andlack of adequatecontrolsover impulsivity,andespecially overaggressive impulses,predictalcoholproblems in men. For women,the findingsare lessconsistent.Robins and McEvoy (1990) state,however,that "girls with a

PULKKINEN AND PITK•NEN

historyof conductproblemswho startsubstance useearly haveas greata risk of later substance abuseas do boys with a similarhistory" (p. 202). Sincethe frequencyof alcohol use in late adolescence correlated with alcohol

abusein youngadulthood (r = 0.68 for menand0.20 for women,p's < .001 and .05, respectively; Pulkkinenand R/3nkfi,1993), it was expectedthat adult alcohol abuse wouldalsoproveto be associated with early indicatorsof lack of adequatecontrolof impulsesand delinquentbe-

havior.Adjustment to schoolwasexpected to be a protective factoragainstdrinking. The influenceof parentaldrinkingon offsprings'alcohol abusehas been studiedfrom differentpointsof view. Bohmanet al. (1981), Cadoretet al. (1980) and Cloninger et al. (1981) have shownthat there is a "genetic factor" operatingbetweenalcoholism in fathersandtheirsons.In contrast,studiesof the psychosocial influenceof an alcoholic parenton the offspring'ssocialdevelopment have

yieldedinconsistent findings.Knopet al. (1984) did not find differencesin the drinking patternsof sonsof alcoholic and nonalcoholic fathers, whereas Barnes and Welte

(1990), Chassinet al. (1991), Cumes-Rayner et al. (1992) and Gleibermanet al. (1991) did find suchdifferences.

The findingsasto therelationships betweenan individual's alcohol abuse and the socioeconomic status of the

familyof originhavelikewisebeencontradictory. Parker and Parker(1980) founda relationshipbetweenproblem drinkingandbeingfroma lowersocioeconomic classfamily; Margulieset al. (1977) foundno relationship; and Martin and Pritchard (1991) found that white males from

highersocioeconomic backgrounds tendedto drink more frequently.Becauseof the relativehomogeneity of the Finnishpopulation in termsof cultureandincomes, socioeconomicstatuswasnot expectedto be a significantcontributorto the onsetof problemdrinking. Method

Subjects

The originalrandomsampleof 196boysand 173girls was drawnin 1968 (Time 1) from second-grade pupils(8-

9 yearsold) in the townof Jyviiskylii,whichhas about 60,000 inhabitants.The sampleconsistedof 12 school classesfrom downtown and suburbanareas. Six years later, whenthe subjectswere 14 yearsold (Time 2), 189

579

ernment register that included information about all offensesthat resultedin imprisonment,and (2) the local, more informal, registerheld by the police. The latter included information about arrests for petty offenses. Alcohol-related

offenses consisted of arrests for drunken-

ness and drunken driving. No arrestsfor alcohol abuse were foundin 64% of the men and 89% of the womenby the age of 26. One offensewas foundin 17% and in 7%, two offenses in 6% and 3% and three or more offenses in

12% and 1% of men and women, respectively. Procedures and variables

Use of alcohol (T3). The mailed questionnairecontainedquestionson (1) the frequencyof the useof alcohol in a certain volume (small volume: I bottle of beer or a glassof wine; moderatevolume: 2 to 4 bottlesof beer or half a bottle of wine; high volume: minimum of 5 bottles of beer or half a bottle of distilled spirits); (2) the frequencyof drinking for certain purposes(medication,relief of depression,meeting friends, dinner); and (3) changesin the use of alcohol(increase,decrease).In the semi-structuredinterview, the subjectswere asked what

kinds of drinks they preferred,how much money they spenton alcoholin a week, whetherdrinking had caused any difficultiesfor them (i.e., fights;breakingone'sword; an accident;problemswith work, a friend or spouse)and whether the subjectshad experienceswith drugs; and for women, also whetherthey had usedalcoholduring pregnancyand breastfeeding. An alcoholismscreeningtest, the CAGE Questionnaire (Mayfield et al., 1974), was also included in the interview: (1) Have you ever felt you shouldcut down on your drinking? (2) Have people annoyedyou by criticizing your drinking?(3) Have you ever felt bad or guilty about your drinking?(4) Have you ever had a drink in the morning to get rid of a hangover? Behavioral characteristics.At the age of 8 (TI), data were collected using peer nomination and teacher rating. The numberof questionswas 33. In peer nomination, subjectswere askedto nameclassmatesof the same

sexby selectingfrom the list of the namesof classmates aboutthreenamesfor eachquestion.Teacherratingswere madefor eachpupil on the samequestionsand five additional questionson a scalefrom 0 to 3 (0 = thosepupils in whom the teacher had never observed the characteris-

boysand167girls(97%of theoriginalsample)werein-

tic in question;3 = characteristicin questionwas very

vestigated in a follow-upstudy.In thelastfollow-upstudy at the age of 26 to 27 (Time 3), 150 men (77% of the sample)and 142women(82% of the sample)wereinterviewed. A mailed questionnaire was completedby 166 men (85% of the sample)and 155 women(90% of the

prominent). Variables constructedfor the analysis of patterns of

sample). Criminal recordswere investigatedfor the whole sam-

ple at the agesof 20 and26 in tworegisters: (1) thegov-

boys'socialadjustment in FinlandandCanada(Pulkkinen and Tremblay, 1992) were used in the analysisof data. The variablesmeasureaggression, hyperactivity,anxiety, inattentivenessand lack of prosocialbehavior, factors whichare generallyidentifiedas components of children's behaviorproblemsand as predictorsof long-termadjust-

580

JOURNAL

OF STUDIES

ON ALCOHOL

ment problems(Pulkkinenand Tremblay, 1992, p. 528): "Physical aggression,"a sum scoreof four standardized items (hurts anotherchild when angry, e.g., by hitting, kicking, or throwingsomething;kicks piecesof furniture or other subjectswhen angry at something;attackssomebody without any reason; teases smaller and weaker peers). Cronbach'salpha for boys and girls, respectively, was 0.93 and 0.89 in peer nomination,and 0.86 and 0.79 in teacherrating. "Anxiety," a sum scoreof three standardizeditems (startscrying easily if otherstreat him/her nastily; is afraid of other children;would cry easily, for example, at the dentist). Cronbach'salpha for boys was 0.83 and for girls was 0.73 in peer nomination,and 0.69 for both boys and girls in teacherrating. "Inattention" (teacherrating) (is unstableand lacksconcentrationin his/ her work and attentiveness)."Hyperactivity" (tends to disobeythe teacher). "Prosociality," a sum scoreof five standardizeditems (tries always to be friendly to others; tries to act reasonablyeven in annoyingsituations;thinks that if one negotiates,everythingwill be better;sideswith smaller and weaker peers; is considereda reliable classmate). Cronbach'salpha for boys and girls, respectively, was 0.89 and 0.90 in peer nomination,and 0.89 and 0.84 in teacherrating. In addition, "school success"was rated by a classteacher. At the age of 14 (T2), peer nominationsand teacher ratingswere madebasedon questionsfocusingon aggression ("Who attacks without reason, teasesothers, says naughtythings?"), anxiety ("Who is fearful, helplessin others' company, a target of teasing, unable to defend himself or herself?."),inattention(lability) ("Who is impulsive, lacks concentration, changes moods?") and prosociality("Who is reliable, keepspromises,doesnot get excited?"). In addition, a sum score for "conduct problems" was composedon the basisof teacherratings on punishmentsat school,truancy,smoking,drinking and contactswith the police. Cronbach'salpha was 0.68 for boysand0.63 for girls. Data on gradepointaverageswere collected

from school archives.

At the age of 26 (T3), self-ratingswere madeon a fourpoint scalefor anxiety ("I am shy, tensein others'company,and try to avoid new situations"), inattention("It is difficult for me to concentrateon one thing, my moods change,and I get nervouseasily"), and prosociality("I am reliableand well-adjusted; I controlmyself"). Aggression was measuredby using a sum scoreof 10 questions presentedin the context of the interview; for instance: "Do you threatenother people?", "Do you hit another person?"and "Do you break things?"Cronbach'salpha was 0.60 for women and 0.67 for men (Pulkkinen and Pitkfinen, 1993).

Social background.Two variables were constructedto describethe subjects'social background.Parentaluse of alcohol was categorized(Pitkfinen, 1990) on the basisof the subjects'semi-structured interviewsat the age of 26

/ SEPTEMBER

1994

by placingspecialemphasison parentaldrinkingat the time when the subjectslived at home. The interviewcov-

ered 12 themes,for instance,maritalstatus,child-rearing comparedto child-rearingin the daysof one'sown childhood,memoriesof homewhenthe subjectwas 14 years old, leisureactivities,anddrinking.Eachinterviewlasted from 2 to 4 hours.Data collectedat the agesof 14 and20 wereusedto validatethe classification. A smallersample of the subjects (77 boysand 78 girls) andoneparentof eachsubjectwere interviewedat the age of 14, and the subjects(68 men and 67 women)againat the age of 20 (Pulkkinen, 1982b). These semi-structuredinterviews in-

cludedquestionson parentaluse of alcohol.In general, the dataobtainedat differentageswereconsistent. It was, however,notedthat at a youngerage the subjectswere morereluctantto speakaboutthe mothers'and the absent

fathers'drinkingthanthey wereduringyoungadulthood. Problemscausedby parentaldrinkingemergedspontaneously in different contextsof the interview.

The main criterionfor the classification of parental drinkingwasthe parent'sdependence on alcoholapproximatingthe categories in DSM-III-R (AmericanPsychiatric Association,1987), which describesdependence in behavioral terms. The classification was based on the sub-

ject'sdescriptions of the frequency andvolumeof parental drinkingand experiences of problemsand difficulties causedby it. No distinctionwas made as to whetheralcohol abuse existed in the father or the mother. In 78% of

homeswith an alcoholicparent,the alcoholicparentwas the father or stepfather;in 5%, the mother;and in 17%, both parents. "Low consumption" of alcohol was found in 165 homes(57% of the subjects).The subjectssaid that their

parentsusedalcoholeithera little or moderately. The parentsdid notdrinkto becomedrunkand,consequently, the parent'suse of alcoholhad not causedany troublefor familylife. Only in a few familieswerebothparentsabstainers. The abstinence of oneparentwasnot considered a criterionin the classification, because the spouse of an alcoholicparentwas often an abstainer. "Mild dependence"was found in 47 homes(16% of

the subjects). The subjects described theirparents'heavy drinking, but made no mention of any problemsthis broughtaboutfor familylife. Severalsubjects (34% of the subjects in thiscategory)described the followingpattern of drinkingas normalbehavior:parentsboughta "Friday

bottle"of Finnishvodkaanddrankit on Fridayevening to becomedrunk.Drinkingoftencontinued on Saturday. Otherdescriptions of parentaldrinkingin thiscategoryincludedregulardrinkingof beer (four or morebottlesin an

evening),heavy drinkingof the absentfather, heavy drinkingat the time whenthe subjectlived at homealthoughafterwardsdrinkingwasreducedfor various(e.g. health)reasons,and a positivedescription of the father's drinking(the fatherwas in a cheerfulmoodwhendrunk).

PULKKINEN AND PITK,•NEN

"Moderate dependence"was foundin 43 homes(15% of the subjects). The subjectsdescribedtheir parents' heavydrinking, as describedabove,and variousproblems that it causedfor family life. Most subjects(71% of the subjectsin this category)describedthe father'slow tolerance for alcohol, negativechangesinducedby drunkennessin the parents'personality(irritated, aggressive) and conflicts between their father and mother concerning drinking.The subjectsalsodescribedparentaldrinkingin very negativeterms, expressing their disgustand disappointment.Parentaldrinkingmighthavehappenedsurreptitiously;only drunkenness was visible. "Severe dependence"(an alcoholicparent) was found in 37 homes(13% of the subjects).The subjectsdescribed one or bothparentsas an "alcoholaddict," "boozer" or "alky." Other descriptionswere usedalso; for instance, "addicted to drink," "often on a binge," "drank for several daysor evenweeks," "more oftendrunkthansober" and "drank to death." Other problemscausedby drinking includedparents'divorce, continuousquarrelsand violence at home and severingthe relationshipwith an alcoholic parent. The rate of divorce dependedon parental

drinking:it was 54% for severedependence, 26% for moderateand mild dependence and 13% for low consump-

TABLEi.

581

Factorsfor drinking

Variables

I

II

III

I bottleof beer or a glassof wine

.78

.00

.21

2-4 bottles of beer or half a bottle

.84

.19

.03

5 bottles of beer or half a bottle

.63

.54

-.12

of spirits Drinking for medication Drinking for depression Drinking when meetingfriends Drinking at dinner CAGE Questionnaire Number of difficulties causedby drinking

.19 .35 .83 .57 .17 .06

.15 .39 .18 .14 .78 .69

- .01 -.03 .12 .28 -. 11 -.03

of wine

Preference for beer Preference for wine

.38 -. 10

Preferencefor spirits Money spenton alcoholweekly Number of times being drunk annually

-. 11 .54 .33

Arrests for alcohol abuse

Eigenvalue % of variance

.14 -.20

-. 14 .92

.04 .58 .76

- .28 -.04 -.05

.04

.52

-. 13

4.95

1.77

0.79

33.0

11.8

5.3

to consumebeer and distilled spirits. If only two factors were extracted,preferencesfor wine loaded negatively (-0.42) on the factor for problem drinking. Otherwise,

tion (}(2= 30.3, 3/290df, p < .001). On average,the

the first two factors remain the same.

rate of divorcein the samplehadreached22% by the time the subjectwas 26 years old. The socioeconomic statusof the family of origin was definedusingthe father'sandmother'soccupational status

Socialand problemdrinkingwere more frequentamong men than women (t's = 3.42 and 7.96, respectively;

as a criterion.

The status classification

was based on

Finnish normative data (Rauhala, 1966). For data analy-

sis, five categorieswere constructed followinga normal distribution.

Results

Problem drinking

A factor analysis(principalfactor axis method)of 15 variables for the use of alcohol and a varimax rotation

yieldedthreefactors(Table1). The factorswereidentified as "social drinking" (I), "problem drinking" (II) and "controlled drinking" (llI). Although the loading of "moneyspenton alcohol"was equalfor socialdrinking andproblemdrinking,problemdrinkingdifferedfrom social drinkingin severalrespects,suchas a higherloading on the CAGE Questionnaire,difficultiescausedby drinking and arrestsfor alcoholabuse.The secondfactorcovered four areas of problem drinking: trouble due to drinking,preoccupied drinking,drinkingfor personaleffects and uncontrolleddrinking (Mulford and Fitzgerald, 1992). Social drinkingwas characterized by drinkingin socialgatherings. The third factorwasspecificallyformed by preferences for wine. Drinkingwine has slowly increasedin Finland;traditionallyit has been more typical

p's < .001), whereascontrolleddrinking was more frequentamongwomen(t = 2.89, p < .01). A large volume of alcohol on a single occasionwas consumedone or more times in a week by 20% of the men and 3% of the women;a moderatevolume by 28% of the men and 10% of the women;and a small volumeby 42% of the men and 31økof the women.Three percentof the men and women had never consumedalcohol, and 3% had stoppedthe use of alcohol. Several women (38% of the women) said that

they had used more alcoholwhen they were younger.In contrast,drinking had increasedamongmen, especially amongmen who had a drinkingproblemindependentof their marital status (Pulkkinen and R6nkii, 1993). Most women(80%) did not use alcoholat all duringpregnancy or breastfeeding. The resultsof the factor analysisconfirmedthe existence of the factor for problem drinking. A scale with three categoriesfor problem drinking was constructed. The main criteria

were documented

arrests for alcohol

abuseand affirmative answersto the CAGE Questionnaire along the guidelinesgiven by McCord (1984). According to Ewing (Ewing, 1984;Rydon et al., 1992), any respondent who scoredpositivelyon two or more of the four items of the CAGE Questionnaireshould be considered alcoholic.

Arrests for alcohol abuse were used as a crite-

rion for lifetime drinking;the recordsof the local police were investigatedat the agesof 20 and 26. Althoughattrition was not high (23% of the men and 18% of the

582

JOURNAL

OF STUDIES ON ALCOHOL

women were not interviewed), it concernedseveral personsknown to be problemdrinkerson the basisof their arrests for alcohol abuse.

A personwas classifiedas falling into Category3 (51 men, 2 women) if he/shehad been arrestedat least three

timesfor publicdrunkenness, or if he or shehad givenat leastthreeaffirmativeanswersto the four questionsin the CAGE Questionnaire.Other variablesfor drinkingbehavior that hadhigh loadingson the factorfor problemdrinking were usedfor ascertaining the severityof drinking problems.Elevenmenandonewomanwereclassifiedin thiscategoryon the basisof policeregistrations on arrests for alcohol abuse.

Category2 was made up of subjects(45 men, 25

/ SEPTEMBER

1994

differencesin the correlationswere significant(p < .01) for peer-nominated anxietyat the agesof 8 and 14. Conduct problemsand poor schoolsuccessat the age of 14 were predictiveof both male and female problemdrinking. The numberof criminal convictionsalso correlated with ScPD. A cross-tabulation revealedthat severeproblem drinking was associatedwith antisocialbehavior(excluding arrests for alcohol abuse) in 41% of men in Category3 (18% of womenin Categories2 and3), if two or more

arrests were

used as a criterion.

Correlations

were systematicallyhigher when calculatedwith ScPD than when calculatedwith the factor scoresfor problem drinking becauseof a higher numberof problemdrinkers involved

in the former.

Male

social and controlled

drink-

womenwere includedin this categoryon the basisof police registrations on arrestsfor alcoholabuse. Category1 was madeup of subjects(68 men and 120 women) for whom the aforementioned criteria were not met. In this category,drinkingwas low or moderate;6%

ing were unrelatedto aggressionand conductproblems, but they alsocorrelatedwith lack of anxiety. A path analysismodel was usedto test the hypothesis that problemdrinkingis accountedfor by indicatorsof a lack of adequatecontrolof behaviorin childhoodand adolescence:aggression,anxiety, hyperactivity,inattention and conduct problems. Prosocialityand school success were expectedto be protective factors againstproblem drinking. The analyseswere basedon peer nominations

of the men and women did not use alcohol at all. The rest

when available because the distributions of variables were

of the sample(32 men and26 women)hadno arrestsfor alcoholabusebut they were excludedfrom the analysis

more normal in peer nominationthan in teacherrating. Causalorderingwas basedon the time of measurement. For variablesmeasuredat the sametime, no causalpriority could be assumed. The analyseswere carried out for each genderby first calculatinga step-wiseregressionanalysiswith all the predictors(T1 and T2 measures;see Table 2). A dependent variable was ScPD. Variables with significantstandardized beta coefficients(p < .05) and correlationcoefficientsbetweenthesevariablesat each measurementpoint are shownin Figures 1 and 2. Since analyseswere made by using standardizedvariables, standardizedand unstandardized beta coefficientsdifferedonly marginally(by 0.02 at most). Standard errors varied from 0.07 to 0.09. The analysesrevealedthat male problemdrinkingwas equallyaccountedfor by low anxiety,poor schoolsuccess and conduct problems at the age of 14 (R--0.43,

women)who had beenconvictedonceor twice for public drunkenness and/orwho gave two affirmative answersin the CAGE Questionnaire.Data from the interview sup-

portedthe ideaof frequentdrinking.Eightmen andfour

because of the lack of interviews.

A high correspondence betweenthe scalefor problem drinking(ScPD)and the factorscoresfor problemdrink-

ing wasrevealed by a cross-tabulation (•2= 135.35, 6/240 df, p = .000); 97% of the subjectsin Category3 belongedto the highestquadrantof the factor scores. ScPD correlatedwith factor scoresfor problemdrinking by 0.80 (p < .001), for social drinking by 0.19 (p < .001), and for controlleddrinkingby -0.18 (p < .01). The validityof ScPDwasanalyzedfor menby comparing the percentages of affirmativeanswers to the CAGE questions given by men in Category3 to thoseobtainedby Mayfield et al. (1974) for alcoholicpatients.In their study,the percentages were as follows(for CAGE questions, see Proceduresand Variables): (1) "cut down,"

87% (vs 77% for Category3 in this study);(2) "annoy," 50% (vs our 68%); (3) "guilty," 76% (vs our 75%); and (4) "hangover,"63% (vs our 68%). For nonalcoholic patients,the percentages variedfrom 4 to 16 in their study. In this study,individualsin Category1 did not give any affirmative

answer.

Behavioralprecursorsof problemdrinking

The Scale for ProblemDrinking correlatedpositively with male aggressionat each point of measurement (age 8, 14 and 26), whereasnegative correlationswere obtainedfor peer-nominated socialanxietyand prosociality in men (Table 2). In females, ScPD was unrelatedto ag-

gression,but it correlatedpositivelywith anxiety.Gender

R2= 0.19, F = 11.21, 3/148df, p = .000). All other beta coefficientswere nonsignificantand smaller than 0.06 exceptfor aggressionat the age of 8 (0.12) and anxiety at the ageof 8 (-0.12). Aggression andinattentionat the age of 14 correlatedwith conductproblemsby 0.44 and0.40, respectively(p's < .001), and with schoolsuccess by-0.30 and-0.34, respectively (p's < .001). Likewise, prosocialityat the age of 14 correlatedwith schoolsuccessand conductproblemsby 0.40 and -0.27, respectively(p's < .001). Strongcorrelationsare likely to induce collinearity when both are entered as predictors into the regressionequation.This may causeunderestimation of the effect sizes.

Each contributingvariable measuredat the age of 14 was, in turn, treated as a dependentvariable, and step-

PULKKINEN AND PITK,•NEN

583

TABLE2. Components of drinkingbehaviorin relationto early behavioralcharacteristics (productmomentcorrelations)

Behavioral

Problem drinking •

characteristics

TI: Age 8 Aggression PN TR Social anxiety

Social drinking t'

M

F

M

F

24* 24*

07 13

-03 05

03 10

Controlled drinking t' M

F

00 -05

-02 -05

- 13

- 18'

-02

-07

- 16'

- 10

-02 -05

-02 - 11

PN

- 15'

24*

-20*

TR

l0

17'

00

PN TR Inattention

12 05

09 13

- 15' 01

00 - 11

TR Prosociality PN TR

13'

13

-23*

- 14

00

- 10

-21' - 17'

-23' - 16'

14 09

-05 -02

23* 11

12 -02

- 11

- 27*

17'

11

21'

20*

24* 17'

04 20*

04 - 19'

13 16'

15' 16'

-07 -22*

Hyperactivity

School success

TR T2: Age 14 Aggression PN

TR Social anxiety PN TR

-25* -05

- 16' -06

-09 -08

-09 -05

- 16' - 15'

-01 -05

Inattention

PN

18'

20*

- 10

15

- 15

-21 *

TR

15'

08

- 17'

12

-09

-05

Prosociality

PN TR Conductproblems TR

- 13 -26*

-22* - 19'

04 05

- 18' - 12

36*

23*

04

20*

Gradepointaverage

- 31'

- 31'

-09

- 10

15

27*

05 04 -08

18'

04 04 -06

11

T3: Age 26 Aggression

SR

26*

10

-03

-01

SR Inattention

16'

01

-05

-11

-11

SR Prosociality SR Crime excl. alcohol-

23*

20*

-04

14

02

- 10

-40* 49*

-03 46*

-04 - 17'

04 04

09 03

20* -09

Social anxiety -11

related offenses

Notes:M = males;F -- females;PN = peer nomination;TR = teacherrating;SR -- self-rating.

•rhe Scalefor ProblemDrinking(ScPD), classifiedinto threecategories(n = 164 for men, 147for women). øScores for FactorsI andIII (seeTable 1) (n = 120for men, 122for women).

*p < .05,*p < .01,½p< .001(onetailed).

wise regressionanalyseswere calculatedwith the variables measuredat the age of 8 as predictors.For men, conductproblemsat the age of 14 were accountedfor by aggression, low anxietyandlow prosocialityat the ageof

8

(R = 0.50, R2 = 0.25, F= 16.28, 3/148 df,

of 14 (R = 0.36, R2-- 0.13, F = 9.21, 2/125 df, p = .0002). All other beta coefficientswere nonsignificant and smallerthan 0.06 exceptfor prosocialityat the ageof 8 (-0.12), and prosociality(-0.09), conductproblems (0.09) and aggression(0.09) at the age of 14. Cor-

p = .000); schoolsuccess at the ageof 14 by prosociality

relations between school success and other variables

andschool success at theageof 8 (R = 0.59,R2 = 0.36,

the age of 14 were as follows:conductproblems,-0.45; aggression, - 0.25; inattention,- 0.35; prosociality, 0.31; and anxiety, -0.17 (p's < .001, except for anxiety at p < .05). Schoolsuccessat the age of 14 was accounted for by schoolsuccessand low anxiety at the age of 8

F = 26.28, 2/148 df, p = .000); and anxietyat the ageof

14 by anxietyat the ageof 8 (R = 0.49, R2= 0.24, F = 47.44, 1/148 df, p = .000).

For women,problemdrinkingwasequallyexplainedby anxietyat the age of 8 and poor schoolsuccess at the age

at

(R = 0.51,R2 = 0.26,F = 22.41,1/125df,p = .000).

584

JOURNAL

OF STUDIES

Time2

Time 1

ON ALCOHOL

Time3

Time 1

00 PROBLEM

I/

DRINKING

[

SUCCESS

FIGUREI. Path diagramsfor a causalmodelfor malesaged8 to 26 (a standardized regression coefficient,beta, aboveor belowa unidirectional

Time 2

Time 3

-.44'** PROBLEM

SCHOOL ,v. ß •5UHUULß

-.21

16•.•

1994

ANXIETY • .22'

/(•iGoG•'• cONDUcT / **PROBLEMS]" ß

/ SEPTEMBER

DRINKING

SUCCESS lSUCCESS

FIGURE 2. Pathdiagrams for a causalmodelfor femalesaged8 to 26 (a standardizedregressioncoefficient,beta, aboveor below a unidirectional arrow indicatesthe direct causaleffect of one variable on another;twodirectionalarrows indicate correlation coefficients;asterisksshow the

significance of a betaor correlationcoefficient:*p < .05, ***p < .001)

arrow indicates the direct causal effect of one variable on another; twodirectional arrows indicate correlation coefficients; asterisks show the

significanceof a beta or correlationcoefficient:*p < .05, **p < .01, ***p < .001)

As expected,male problemdrinkingwas accountedfor by conductproblemsand aggression; the latter indirectly (0.30 x 0.20 - 0.06). Protectivefactorsagainstproblem drinkingwereanxiety,schoolsuccess andprosociality; the latter indirectly (-0.30 x 0.20 = 0.06). School success was also a protectivefactor for female drinking, but in contrastto men, anxiety was a risk factor for female drinkingboth directly and indirectly. One-way ANOVASof behavioralcharacteristics,classified into threecategoriesby usingthe 25th and 75th percentilesas cut-off points, confirmedthe findings that aggressionat the agesof 8 and 14 was a risk factorfor male problemdrinking, whereasanxiety at the age of 8 was a risk factorfor femaleproblemdrinking(Table 3). Significanttwo-way interactionswere obtainedfor gender and anxietyat the agesof 8 and 14 (F = 3.95, 2/319 df, p = .020; F = 5.39, 2/294 df, p = .005; respectively); genderand aggression at the age of 14 (F = 5.80, 2/294 df, p = .003); gender and inattentionat the age of 14 (F = 3.70, 2/294 df, p = 0.26); and genderand conduct problems at the age of 14 (F = 4.81, 2/295 df, p = .009).

andthe ParentalDrinkingx Genderinteractionweresignificant(F = 3.62, 3/291 df, p = .014; F = 3.08, 3/291 df, p = .028; respectively). The latter was causedby a threshold effectfoundfor menbut notfor women(Figure 3)--the problemdrinkingof menwhoseparents'alcohol consumptionwas low was significantlylower (p < .05, Scheff6'stest) than the problemdrinkingof men whose parentshad a mild dependence on alcohol.Other group differenceswere not significant.There was a trend that this thresholdeffect was strongerif the socioeconomic statusof the family was higherthan if it was middleor lower(three-wayANOVA: F = 1.69, 12/273df, p = .070). However,in a few casesin whichparentsbelonging to the higheststatusgrouphad a severedependence on alcohol, the offspring'sproblemdrinkingwasvery low. Two-wayinteractioneffectsof parentaldrinkingand behavioralcharacteristics on problemdrinkingwere not

Problem

drinking 2 --

1.5

I

Low

Social background

MEN

WOMEN

I

Mild

I

Modemto

consumption dependence dependence

I

Severe

dependence

Parentalddnking

The correlationof parentaldrinkingwith the offspring's problemdrinkingwas 0.21 (p < .01) for boysand 0.08 (Ns) for girls. Both the main effect of parentaldrinking

FIGURE 3. Mean scores of young adults' problem drinking (1--Category 1; 3 = Category3) in differentgroupsof parental drinking

PULKKINEN AND PITK,•NEN TABLE

3.

585

Problem drinkingin relationto earlybehavioral characteristics" (means of problem drinking •' andA•OVA)

Behavioral

characteristics and age

Low I

Medium 2

High 3

F

Scheff6

M

1.73

1.80

2.24

4.90*

3 > 1,2

F

1.11

1.21

1.26

1.28

2.00 1.09

1.94 1.16

1.71 1.37

1.44 4.56*

1.82

1.76

2.15

3.10'

1.14

1.30

1.23

2.17

M

2.18

1.88

1.68

3.71'

F

1.39

1.15

1.08

6.46*

1.98 1.32

1.95 1.19

1.75 1.10

1.19 3.28*

M

1.58

1.98

2.24

8.14'

F

1.19

1.23

1.19

0.12

TI : Age 8 Aggression

Social anxiety M F Inattention M F

3 >

1

Prosocialtiy I > 3

I > 2,3

School success

M F

I > 3

T2: Age 14 Aggression

Social anxiety M

2.09

1.79

1.59

4.89*

1.16

1.25

1.26

0.82

M

1.57

2.07

2.07

6.64*

F

1.15

1.16

1.32

1.82

M

2.00

1.90

1.31

1.22

1.76 1.06

0.84

F

F Inattent ion

2,3 > 1 I > 3

2,3 > 1

Prosociality 3.22*

Conductproblems

M F School success

M F

1.51

1.92

2.25

8.85*

1.14

1.19

1.30

1.49

2.18 1.46

1.96 1.14

1.51 1.07

6.29* 8.79'

2,3 > 1 1,2 > 3 I > 2,3

Note: M = males, F = females.

"Classifiedinto threecategoriesby usingthe 25th and 75th percentiles as cut-offpoints.

t'Problem drinking codedon a scalefromI to 3. *p < .05, *p < .01, *p < .001.

significant. One three-way interaction effect (Parental Drinking x Genderx Conduct Problems) on problem drinking was, however,significant(F = 2.95, 6/283 df, p -- .008). Generally,the higherthe male problemdrinking, the moreconductproblemsa boyexhibitedin adolescence. Similarly, 50% of the daughtersof alcoholic parentshadhigh scoresfor conductproblemsandproblem drinking, supportingthe findingsof Robinsand McEvoy (1990). However, 42% of the sonsof severelydependent parentshad low conductproblemsbut high problem drinking,suggesting a differenttype of alcoholic. The correlation of socioeconomic status (SES) of the

family of origin with the problemdrinkingof adultchildren was -0.17 (p < .05) for men and -0.10 (ns) for women. The main effect of SES and the SES x Gender

interactioneffect on problemdrinking were not significant. Nevertheless,SES had significanteffectson problem drinkingin the interactionwith prosocialityat the age of 8 and anxiety at the age of 14 (F = 2.13, 8/283 df,

p = .034; F = 3.32, 8/267 df, p = .001; respectively). Low prosocialitywas mostpredictiveof problemdrinking in the lowerstatusgroups,whereashighanxietywasmost protective against male problem drinking in the middle and higher statusgroups. Discussion

Three components of drinkinghabitsand social, problem and controlleddrinkingemergedin youngadulthood. Both male drinking types (social and problemdrinking) involvedthe consumption of alcoholin large quantities, but indicatorsof dependenceon alcoholand problems causedby drinkingwere characteristicof problemdrinking only. Among the most discriminative variables for problemdrinking was the numberof affirmative answers to the CAGE Questionnaire.Its high diagnosticaccuracy for the identification

of alcoholics has also been noted in

other studies(Rydon et al., 1992).

586

JOURNAL

OF STUDIES

ON ALCOHOL

The target of the presentstudywas problemdrinking. Heavy drinking is frequentamong Finnish men and it givesrise to manyproblemsfor their marriagesand children. The reasonsfor drinkingare manifold,as is shown by the variouspsychologicaltheoriesof drinking (Blane and Leonard, 1987); but in the presentstudy,a developmental perspectiveon drinking was adopted.Within the frameworkof the longitudinalstudy,it waspossibleto determine whethera child's early behavioralcharacteristics were predictiveof problemdrinking. Supportingthe findings of other studies(Anderssonet al., 1989; Barnesand Welte, 1986; Donovan et al., 1983; van Kammen et al.,

1991; McCord and McCord, 1960; Robinsand McEvoy, 1990), indicatorsof weak controlof a boy's behavior(aggression,conductproblems)predictedproblemdrinking. Social drinking did not correlatewith theseindicators;it supportsthe findingsby Kandelet al. (1984). The hypothesis that drinking is related to delinquentbehaviorwas only confirmedfor problemdrinking, and eventhen, only less than half the men had been convicted for nonalcohol offenses.

According to a model of impulse control (Pulkkinen, 1982b), anxiety is also an indicatorof lack of adequate controlof behavior.Althoughsocialanxiety at the age of 8 was relatedto femaledrinking, it was a protectivefactor againstproblemdrinkingin men. McCord (1984) hasalso foundfor problemdrinkersthat they were lesslikely than othersin their schoolsto have appearedto be insecureor dependent.Likewise, anxietyhas beenshownto be a pro-

/ SEPTEMBER

port, in the early upbringingof problemdrinkers(Chassin et al., 1993; Martin and Pritchard, 1991; Mercer and Kohn, 1980; Pulkkinen, 1983; Wechsler and Thum,

1973). As expected,the relationshipbetweensocioeconomicstatusand problemdrinkingwas ratherlow. The resultsconfirmedthe hypothesis/that schoolsuccessand prosocialityare protectivefactorsagainstproblem drinking.Low prosociality in the presentstudymeans poorinterpersonal socialskillsandcopingstrategies. Poor socialskills may causepoor adjustment to schoolwork which, in turn, may causereluctance to continuestudies after the obligatoryschoolperiod, and unemployment. Poor conflict-resolution

skills and low work-status ac-

countedfor excessive maledrinkingin the studyby Frank et al. (1990).

Accordingto Goodwin (1984), 20-25% of the sonsof

alcoholicsbecomealcoholic.Correspondingly, a parent's heavy drinkingwas a risk factor for a son'sproblem drinkingin thepresentstudy.Whenparentaldrinkingwas associated with the son'sconductproblems,the risk was highest.However,there were sonsof alcoholicparents who had not exhibitedconductproblems,but who nonethelesshad severeproblemdrinking.Althoughantisocial behavioris oftenassociated with problemdrinking,there may be other risk factors(e.g., congenitaldiathesisfor alcoholismas suggested by Hill, 1992). Consequently, there may be different typesof alcoholics.On the other

hand, one-thirdof the sonsof alcoholicparentshad no problemswith alcohol.

tective factor for antisocial behavior (alumstein et al., 1985). Shedlet and Block (1990) have found that adolescents who used drugs frequently were maladjustedand

showedpoor impulsecontrol, whereasadolescents who, by the age of 18, had neverexperimentedwith any drug were anxious,emotionallyconstrictedand lackingin social skills.High risk-takingwas associated with substance use in the studyby Alexanderet al. (1990). It is possible, as Anderssonet al. (1989) remark, that a multiproblem backgroundis of greaterimportancefor alcoholismdevelopedat an early adult age than it is for alcoholabusedevelopedat a later age. A later onsetof alcoholismmay be more stronglyrelatedto intrapsychological characteristics (Hesselbrocket al., 1984; Gomberg, 1982). At a later state of the longitudinalstudy, when most subjectsare pastthe age of risk for alcohol-related problems,it might be possibleto see whether more passivemen who start drinkingat an olderage, andwomenwho havea typically later onset of problem drinking, also develop problem drinking at an older age. The use of alcoholbeginsamongboyswho exhibit extravertedand impulsive-uninhibited behaviorearlier than amongintrovertedboys. An interactionof a child's temperamentalcharacteristics and child rearing in the developmentof problemdrinking is obvious.Several studies have showninadequacy,especiallylack of parentalsup-

1994

References ALEXANDER, C.S., KIM, Y.J., ENSMINGER, M. ANDJOHNSON, K.E. A

measure of risktakingfor youngadolescents: Reliabilityandvalidity assessments. J. Youth Adolesc. 19: 559-569, 1990.

AMERICANPSYCHIATRIC ASSOCIATION. Diagnosticand StatisticalMan-

ual of MentalDisorders (DSM-III-R), Washington, D.C., 1987. ANDERSSON, T., BERGMAN,L.R. ANDMAGNUSSON, D. Patternsof ad-

justmentproblems andalcoholabusein earlyadulthood: A prospective longitudinal study.Devel. Psychopathol. 1: ! 19-131,1989. BARNES, G.M. ANDWELTE,J.W. Patternsandpredictors of alcoholuse among7-12thgradestudentsin New York State. J. Stud. Alcohol47: 53-62, 1986.

BARNES, G.M. ANDWELTE,J.W. Predictionof adults'drinkingpatterns from the drinkingof their parents.J. Stud. Alcohol51: 523527, 1990.

BEANE,H.T. AND LEONARD,K.E. (Eds.) Psychological Theoriesof Drinking and Alcoholism,New York: Guilford Press,1987. aLUMSTEIN,A., FARRINGTON, D.P. AND MOITRA,S. Delinquentcareers:Innocents,desisters,and persisters.In: TONRY,M. ANDMOR-

RIS,N. (Eds.)CrimeandJustice: An AnnualReview,Chicago: Univ. of ChicagoPress,1985,pp. 187-219. BOHMAN,M., SIGVARDSSON, S. ANDCLONINGER,c.e.

Maternal inher-

itanceof alcoholabuse:Cross-fostering analysisof adoptedwomen. Arch. Gen. Psychiat.38: 965-969, 1981.

CADORET, R.J. CAIN,C.A. ANDGROVE,W.M. Development of alcoholismin adoptees raisedapartfrom alcoholicbiologicrelatives. Arch. Gen. Psychiat.37: 561-563, 1980.

PULKKINEN AND PITK•NEN CHASSIN, L., PILLOW, D.R., CURRAN, P.J., MOLINA, B.S.G. AND BAR-

RERA,M., JR. Relationof parentalalcoholismto early adolescent substanceuse: A test of three mediatingmechanisms.J. Abnormal Psychol.102: 3-19, 1993. eliASSIN,L., ROGOSCH, F. ANDBARRERA, M. Substance useand symptomatologyamong adolescentchildren of alcoholics.J. Abnormal Psychol.100: 449-463, 1991. CLONINGER,C.R., BOHMAN, M. AND SlGVARDSSON, S. Inheritance of

alcoholabuse:Cross-fostering analysisof adoptedmen. Arch. Gen. Psychiat.38: 861-868, 1981. CUMES-RAYNER,D.P., LUCKE, J.C., SlNGH, B., ADLER, B., LEWlN,

m., DUNNE,M. ANDRAPHAEL,B. A high-riskcommunitystudyof paternalalcoholconsumption and adolescents' psychosocial characteristics. J. Stud. Alcohol 53: 626-635, 1992.

44: 109-137, 1983.

EWING, J.A. Detectingalcoholism:The CAGE questionnaire.JAMA 252: 1905-1907, 1984. Alcohol Use acrossthe Life Course: A Critical Review

of 70 Yearsof InternationalLongitudinalResearch,Toronto:Addiction Research Foundation, 1988.

780, 1990.

GLEIBERMAN,L., BARBURG,E., DI FRANCEISCO,W. AND SCHORK,A.

Familialtransmission of alcoholuse:IV. A seventeen-year follow-up on the relationshipsbetweenparentand adult offspringalcoholuse; Tecumseh,Michigan.Int. J. Epidemiol.20: 441-447, 1991. GOMBERG, E.S.L. The youngmalealcoholic:A pilot study.J. Stud.Alcohol 43: 683-701, 1982.

and aversive transmission. J. Stud. Alcohol 43:

497-516, 1982.

HESSELBROCK, V.M.,

BABOR,T.F., STABENAU,

J.R., MEYER,R.E. ANDWEIDENMAN, M. Antisocialbehavior,psychopathology andproblemdrinkingin the naturalhistoryof alcoholism. In: GOODWIN, D.W.,

VAN-DUSEN, K.T. AND MEDNICK, A.S.

(Eds.) LongitudinalResearchin Alcoholism,Boston:Kluwer-Nijhoff Publishing,1984, pp. 197-214. HILL, S.Y. Absenceof paternalsociopathy in the etiologyof severealcoholism:Is there a Type III alcoholism?J. Stud. Alcohol 53: 161169, 1992.

demic Press, Inc., 1977.

JONES,M.C. Personalitycorrelatesand antecedents of drinkingpatterns in adult males.J. Cons.Clin. Psychol.32: 2-12, 1968. KANDEL,D.B., RAVElS,V.H. ANDKANDELP.I. Continuityin discontinuities:Adjustmentin youngadulthoodof formerschoolabsentees. Youth and Soc. 15: 325-352, 1984.

TEASDALE, T.W.,

the elderly?J. Stud. Alcohol 53: 601-610, 1992.

PARKER, D.A. ANDPARKER, E.S. Statusand statusinconsistency of parentson alcoholconsumption of teenagechildren.lat. J. Addict. PITK•NEN,T. Alkoholiperheiden lastenkehityksenseurantakouluik•iisistiinuoriksiaikuisiksi(Development of Childrenof AlcoholicParems from SchoolAge to YoungAdulthood.In Finnish.), master's thesis,Universityof Helsinki, 1990.

PULKKINEN, L. Nuortentupakointija alkoholink/iytt6 pitkitt/iistutkimuksenvalossa(Youthfulsmokinganddrinking:A longitudinal study.In Finnish.), Health Education, Whole Nr. 2, Finland: National Board

PULKKINEN,L. Self-controland continuityfrom childhoodto late ado-

lescence. In: BALTES, P.B. ANDBRIM,O.G., JR. (Eds.)Life-span Development and Behavior,Vol. 4, San Diego, Calif.: Academic Press,Inc., 1982b,pp. 63-105.

PULKKINEN, L. Youthfulsmokinganddrinkingin a longitudinal perspective.J. YouthAdoIPsc.12: 253-283, 1983.

PULKKINEN, L. ANDNARUSK, A. Functions of adolescent drinkingin Finlandandthe SovietUnion. Europ.J. Psychol.Educ.2:311-326, PULKKINEN, L. ANDPrrK•NEN,T. Continuitiesin aggressive behavior fromchildhoodto adulthood.Aggress.Behav.19: 249-263, 1993. PULKKINEN, L. ANDRONK•, A. Ongelmajuominen nuoressa aikuisiiissii ja senkehitystaustaa. (Problemdrinkingin youngadulthood andits developmental background. In Finnish.)Alkoholipolitiikka 58: 90102, 1993.

PULKKINEN, L. ANDTREMBLAY, R. Patterns of boys'socialadjustment in twocultures andat differentages:A longitudinal perspective. Int. J. Behav. Devel. 15: 527-553, 1992.

RAHKONEN, O. ANDAHLSTROM, S. Trendsin drinkinghabitsamong Finnishyouth from 1973 to 1987. Brit. J. Addict. 84: 1075-1083, 1989.

RAUHALA,U. Suomalaisenyhteiskunnansosiaalinenkerrostuneisuus WSOY, 1966.

ROBINS, L.N. ANDMcEvoY,L. Conduct problems aspredictors of substanceabuse.In: ROBINS, L.N. ANDRUTTER,M. (Eds.)Straightand DeviousPathwaysfrom Childhoodto Adulthood,New York: Cambridge Univ. Press, 1990. RYDON,P., REDMAN,S., SANSON-FISHER, R.W. ANDREID, A.L.A. De-

tectionof alcohol-related problemsin generalpractice.J. Stud.Alcohol 53: 197-202, 1992.

MIKKELSEN, U. AND

SCHULSINGER, F. A Danish prospectivestudy of young males at high risk for alcoholism.In: GOODWIN,D.W., VAN-DUSEN,K.T. AND MEDN!CK,S.A. (Eds.) LongitudinalResearchin Alcoholism, Boston:Kluwer-NijhoffPublishing,1984,pp. 107-124. McCORD,J. Drunkendriversin longitudinalperspective.J. Stud. Alcohol 45: 316-320, 1984.

McCORD,W. ANDMCCORD,J. Originsof Alcoholism,Stanford,Calif.: Stanford Univ. Press, 1960.

MARGULIES, R.Z., KESSLER, R.C. AND KANDEL,D.B. A longitudinal studyof onsetof drinkingamonghigh-schoolstudents. J. Stud. Alcohol 38: 897-912, 1977.

MULFORD, H.A. ANDFITZGERALD, J.-L. Elderlyversusyounger problemdrinkerprofiles:Do theyindicatea needfor specialprograms for

(Social structureof the Finnishsociety.In Finnish.),Helsinki:

JESSOR, R. ANDJESSOR, S.L. ProblemBehaviorandPsychosocial Development:A LongitudinalStudy of Youth, San Diego, Calif.: Aca-

KNOP, J., GOODWIN, D.,

chiat. 131: 1121-1123, 1974.

MERCER,G.W. ANDKOHN,P.M. Child-rearingfactors,authoritarianism,druguseattitudes,andadolescent druguse:A model.J. Genet. Psychol.136: 159-171, 1980.

1987.

GOODWIN,D.W. Studiesof familial alcoholism:A review.J. Clin. Psychiat. 45 (No. 12, Sect. 2): 12-17, 1984. HARBURG, n., DAVIS,D.R. ANDCAPLAN,R. Parentandoffspringalco-

HESSELBROCK, M.N.,

MAYFIELD, D., MCLEOD,G. ANDHALL,P. The GAGE-questionnaire: Validationof a newalcoholism screening instrument. Amer.J. Psy-

of Health, 1982a.

FRANK,S.J., JACOBSON, S. ANDTUER, M. Psychological predictorsof youngadults' drinking behaviors.J. Pers.Social Psychol.59: 770-

hol use: Imitative

Factors associated with alcohol

use in later adolescence.J. Stud. Alcohol 52: 5-9, 1991.

15: 1233-1239, 1980.

DONOVAN,J.E., JESSOR, R. ANDJESSOR, L. Problemdrinkingin adolescenceand youngadulthood:A follow-upstudy.J. Stud. Alcohol

FILLMORE, K.M.

MARTIN, M.J. AND PRITCHARD, M.E.

587

SnEDLER, J. ANDBLOCK,J. Adolescent drug use and psychological health:A longitudinal inquiry.Amer.Psychologist 45:612-630,1990. SIMPURA, J. Muuttuvatko suomalaisten juomatavat? Vuosien1968, 1976 ja 1984juomatapatutkimusten vertailua.(Are Finnishdrinkinghabits changing? A comparison of resultsof surveys on drinkinghabitsin 1968, 1976and 1984. In Finnish.)Alkoholipolitiikka 50: 294-302, 1985.

VAN KAMMEN,W.B., LOEBER,R. ANDSTOUTHAMMER-LOEBER, M. Sub-

stanceuseandits relationship to conduct problems anddelinquency in youngboys.J. YouthAdoIPsc.20: 399-413, 1991.

WECHSLER, H. ANDTHUM,D. Teen-age drinking,druguse,andsocial correlates.Q. J. Stud. Alcohol 34: 1220-1227, 1973.