Journalof Studieson Alcohol,Vol. 52, No. 5, 1991

Marriage and Alcohol Use: A Longitudinal Study of "Maturing Out"* CAROL MILLER-TUTZAUER, M.A., KENNETH E. LEONARD, PH.D., MICHAEL WINDLE, PH.D. ResearchInstituteon Alcoholism,1021Main Street,Buffalo,New York14203

ABSTRACT.Earlierfindingshavesuggested thatmarriagemayprovide protectionfrom a variety of physicaland psychological problems. In particular,numerousstudieshave demonstrated a consistent relationshipbetweenmarital statusand alcohol use. To examinethis relationshipmorecarefully,we performeda longitudinalanalysisof datacollectedon the YouthCohortof the NationalLongitudinalSur-

veyof LaborMarketExperience. Fourmaritaltransition groupswere constructed baseduponmaritalstatusacrossa 3-yearperiod:stably

single,marriedyear 3, marriedyear2 andstablymarried.Resultsof repeated-measures MANCOVAS performed onalcohol-use patternsacross time as a functionof marital-transition groupsupportedthe notion that individualsbeginmoderating their alcoholconsumption prior to their actualtransitionto marriedstatuswith the trendcontinuinginto the first year of marriage. The data further suggested that this decline in alcoholusestabilizesshortlythereafter,apparentlywithin 1 year after marriage.(J. Stud.Alcohol52: 434-440, 1991)

ARRIAGE, although accompanied by numerousologicalstudiesalsodemonstratea lower risk for alcohol-

stresses and challenges,nevertheless providesprotection from a host of physicaland psychological problems. In general,married individualsexhibitlower rates of depression and fewerdepressive symptoms than individualswho are single,divorcedor widowed(Vernonand Roberts, 1982; Weissman,1987; Weissmanand Myers, 1980).Similarly,suicideratesare particularlylow among married men and women(Dublin, 1963). The married individual,maleor female,is at a considerably lowerrisk of beingthe victimof a varietyof crimes,includingrobbery, rape and assault(Flanaganand McGarrell, 1986). Traffic and other accidentsresultin more injuriesamong singleanddivorcedthanmarriedpersons(Collins, 1985). Indeed,mortalityoverallappearsto be lowerfor married thanfor singleindividuals(Kotler and Wingard, 1989). The protectionprovidedby marriageextendsto alcohol problemsand alcoholismas well. For example, Bacon (1944)observed thatmenarrestedfor inebrietyweremore

likely to be single,separated or divorced,andlesslikely to be marriedandlivingtogetherthanwouldbe expected on the basisof age-adjusted populationfigures.Epidemi-

Received:July 17, 1989.Revision:January23, 1990. *The writingof thisarticlewassupported in part by NationalInstitute onAlcoholAbuseandAlcoholism grantAA0786(MichaelWindie,principal investigator). An earlierversionof this studywas presented at the annualmeetingof the Research Societyon Alcoholism,Symposium on Alcoholin the Marriageand Family Context,BeaverCreek, Colorado, June 12, 1989.

Requestsfor reprintsshouldbe sentto KennethE. Leonard,Research Institute on Alcoholism, 1021 Main Street, Buffalo, N.Y. 14203. 434

ism and alcoholproblemsamongmarried individualsrelative to thosewho never married and thoseseparatedor divorced(Bailey et al., 1965; Clark and Midanik, 1982; Cosper and Mozersky, 1968). Even among alcoholics, married alcoholicsappearto experienceless severeproblems than alcoholicsof some other marital status(Bromet

and Moos, 1976), suggesting the possibilitythat marriage may serveto protectthe alcoholicfrom the consequences of excessivedrinking. Marital statusrelates to alcohol consumptionin much the sameway. Married men and womenconsumeconsiderablylessalcoholon the average thansingle,separatedor divorcedpersons(Clark and Midanik, 1982).

Althoughmarital statusis related to levels of alcohol consumption andalcoholproblems,the processes underlying thisrelationshiphavenot beenclearlydelineated.Single, married, separatedand divorced individualsdiffer fromoneanotherin numerous ways,not simplyin marital status.For example, separatedand divorcedpersonsare generallyolder than married personswho, in turn, are older than those never married (Mattessich and Hill, 1987). Other sociodemographic factorslinked to marital

status--factorssuch as ethnicity,social statusand the presenceor absenceof children--complicatethe picture further.Becauseprior investigations treatedmaritalstatus as a controlvariableratherthan the principalfocus,these studieshave not assessedthe relationshipbetween marriage and alcoholuse controllingfor relevantsociodemographicfactors. The more difficult, thoughmore interesting,issueinvolves determiningwhether marital status and alcohol

MILLER-TUTZAUER,

LEONARD

AND WINDLE

435

consumptionare causallylinked. That is, doesa changein marital statuslead to an alterationof alcoholconsumption patterns?Focusingon the transitionfrom single to married, Zucker (1979), for example,hypothesizedthat this

n = 3,475). The last group(MMM) consistednot only of personswho married initially in 1983 but includedlonger term marriagesas well. Additionally,we consideredany

transition, with its attendant values, serves to moderate

gle to separatedor divorcedas reflective of a changeto married statusduring the previousyear. The numberof persons reporting such a transition varied somewhat acrossthe four marital groups:The SSS groupcontained no suchcases(0%) sinceby definition all individualsreported havingnever married acrossall 3 years.The SSM groupcontaineda small numberof persons(8%) who reportedbeing singlein 1983 and 1984 and reportedbeing separatedor divorcedin 1985. The SMM groupcontained someindividuals(6%) who reportedbeing singlein 1983 and separatedor divorcedin 1984. In the MMM group, 17% of respondents reportedbeingseparatedor divorced at the first measurement occasion(1983). The larger proportionfor the MMM groupprobablyreflectsthe fact that one must first be married to be separated/divorced, that the group containspersonsmarried for any number of years prior to 1983 and thereforepersonsin the group have simplyhad more opportunityto separateor divorce than personsin the SMM and SSM groups.The sample also includedpersonswho remarried, althoughestimation of the proportionwithin eachgroupwas not possible.In short, once a personreporteda transitionto married status, we consideredthat person"married" for all subsequent yearseven thoughthat personmay have separated, divorcedor remarried. We decidedto includeseparated/ divorcedpersonsin the sampleto avoidbiasingthe analysesbecausethosewith alcoholproblemsmay havebeen morelikely to havemarriagesendingin separationor di-

alcohol consumption.Conversely,certain alcohol consumptionpatternsmay facilitatechangesin marital status. More specifically, with regard to the single-to-married transition,extremelyhigh levels of alcoholconsumption may delay entry into marriage. In the absenceof longitudinal data, however,it has not been possibleto determine which processproducesthe differencesin drinking patterns among individualsof different marital statuses.In fact, both processes may play importantroles. The purposeof the presentstudyis to examinethe relationship betweenalcoholconsumptionand marital statusfrom a longitudinalperspective.This study,a secondaryanalysis of data collectedfor the NationalLongitudinalSurveyof Youth, focusesprimarily on the transitioninto marriage. The transitionout of marriageand its relationshipto alcohol consumption will be the topicof a futurepaper. Method

The data for this study were derived from the Youth Cohort of the National LongitudinalSurvey (NLSY) of Labor Market Experience conducted by the National Opinion ResearchCorporationfor the U.S. Departments of Labor and Defense.This annualsurveybeganin 1979 when respondents were betweenthe agesof 14 and 21. The samplingstrategyfor the NLSY is describedin detail elsewhere(e.g., Bock and Moore, 1986). Briefly, the samplefor the NLSY was selectedthrougha multistage stratifiedareaprobabilitysampleof households andgroup living quarters,with oversamplingof Hispanics,blacks and non-Hispanic/non-black economicallydisadvantaged individuals.Interviewerscontactedhouseholds, screening eachto determinethe compositionof the household,specifically whetherany individualswithin the specifiedage range residedin the household.The samplingstrategy identified 12,686 prospectivecivilian participants.Over the years1979 to 1985, the retentionratesfor this sample haveremainedrelativelystableat approximately 95%. Data for the presentstudyspanthe years 1983, 1984 and 1985, the only yearsfor which alcoholconsumption

data were available.Of thosepersonsreportingmarital statusfor all 3 years (n = 10,594), we constructedfour groupson the basisof marital statusduringthose3 years: (1) personswho reportedbeing singleat all three measurementoccasions(designatedSSS, n = 5,908); (2) thosewho reportedbeing single in 1983 and 1984, but who reporteda transitionto married statusin 1985 (SSM, n = 605); (3) thosesinglein 1983, but who reporteda transitionto married statusin 1984 (SMM, n = 606); and (4) thosewho reportedbeing married as of 1983 (MMM,

instances in whichpersonsreporteda transitionfrom sin-

vorce. Had we not included such individuals, means on the alcohol variables

for married

individuals

could have

been artifically deflated comparedto single individuals. Thus, we followeda conservativestrategyin our assignment of individualsto marital categories,while simultaneouslyminimizingthe numberof suchcategories. In eachof the 3 years, informationwas collectedconcerning the respondent'salcohol consumptionin the 30 daysprior to the interview.Specifically,the interviewer askedthe respondent the numberof days s/he drank one drink, two drinks, three drinks, four drinks, five drinks,

and six or more drinks.First, we computedan average daily quantity(ADQ) by multiplyingthe numberof days by the numberof drinksspecifiedin eachquestion,summing acrossthe six differentquantityquestionsand dividing by 30. Second, we assessedfrequencyof heavy drinking(FHD) by summingthe numberof timesduring the past monththe respondentadmitteddrinking five or more drinksper day. In 1984 and 1985 a numberof different questionsconcerningproblems associatedwith drinking were also asked of the respondents. We constructedtwo scalesbasedupon thesequestions:alcoholrelatedproblems(composed of job-relatedand aggression

436 TABLE 1.

JOURNAL

OF STUDIES

ON ALCOHOL

/ SEPTEMBER

1991

Comparisonof drinkingratesfor NLSY and Clark and Midanik (1982) samples,in percenta MALES

Abstainers

1-60 drinks/mo.

> 60 drinks/mo.

Age

1983

1984

1985

Avg.

1979

1983

1984

1985

Avg.

1979

1983

1984

1985

Avg.

1979

group

NLSY

NLSY

NLSY

NLSY

C&M

NLSY

NLSY

NLSY

NLSY

C&M

NLSY

NLSY

NLSY

NLSY

C&M

18-20 21-25

25 20

24 18

23 22

24 20

5 10

65 69

64 69

63 68

64 69

79 54

10 11

13 12

14 11

12 11

17 36

26-28 b

21

19

21

20

20

67

69

71

69

50

13

11

8

11

29

18-20 21-25

40 35

34 33

36 36

37 35

31 15

57 62

63 64

62 62

61 63

64 78

3 3

3 3

2 2

2 3

5 6

26-28 t'

44

41

38

41

30

56

58

60

58

65

0

2

2

1

5

FEMALES

aNLSYdenotes theyouthcohortof theNationalLongitudinal Surveyof LaborMarketExperience; C&M denotes the Clark andMidanik1979trendstudy (1982).

t'rhe agegroupfor theClarkandMidanik(1982)studywasactually26-30, ratherthan26-28. TheNLSY did notcontaindatafor anyindividuals beyondthe age of 28 years.

problems)and dependency symptoms.Internalreliabilities (Cronbach's or) for the scales were .65 and .63 for

alcohol-related problemsand .66 and .67 for dependency symptoms in 1984and 1985,respectively. Ratherthanusing a 30-daywindowas wasdonefor the alcoholconsumption measures,the alcohol-related problemsand dependency symptomsmeasuresreflect alcohol-relateddifficultiesexperiencedin the entire year prior to the survey. In addition to the marital status and alcohol measures,

sociodemographic informationwas collected on gender, age, race (white, black, Hispanic,Native Americanand other), religion(Catholic,Protestant,Jewish,other,none) and educationalstatus. Preliminary examinationof the data indicated that insufficient numbers of Native Ameri-

cansand Jewishrespondents existedin the samplefor analytic purposes.We thereforeincludedNative Americans in the "other" racial categoryand classifiedJewishrespondentsas "other" religiousaffiliation. Becausethe NLSY samplecontainsonly personsaged 14-21in 1979 (and thereforeaged 18-25 in 1983, 19-26 in 1984 and 20-27 in 1985), comparisonto other general populationsamplesis difficult. Any alcohol-related measuresare likely to be higher amongparticipantsin the NLSY thanthosefrom samplesof the populationat large sincedrinkingratesare particularlyhigh amongthis age group. However,Clark and Midanik (1982) reportedgeneral patternsof consumptionbrokendown by age group andtheir figurescan serveas a pointof comparison. Table 1 outlines drinking characteristicsbroken down by genderfor the NLSY samplein contrastto the 1979findings of Clark and Midanik (1982). As can be seen, NLSY

estimatesof the proportionof abstainersare higher for both men and womenacrossall age groupswith the exception of men aged 26-28 years. We should caution, however,that direct comparisonof the figuresfor the two samplesis risky sincedifferentquestionswere usedin the two studiesto elicit responses regardingalcoholconsumption. One such difference involves the time windows used

in each study.NLSY used a 30-day window, whereas Clark andMidanik requested average30odayconsumption (numberof drinks)duringthe pastyear. Results

Analytic strategy

To control for relevant sociodemographic factors, we first determined which factors to include in the analysis. After convertingrace and religion into a seriesof dummy variables,we obtaineda correlationmatrix for

the potentialcovariates(age, education,race-white,raceblack,race-,Hispanic, religion-Protestant, religion-Catholic, religion-none)and alcoholmeasures(ADQ, FHD, problems,dependency) for eachyearavailable.We eliminated from consideration any covariatesthat correlatedlessthan ß10 with at leastone of the alcoholmeasures,leavingonly education and race to be included as covariates in the

largeranalysis.Thus, we excludeda covariatefrom considerationonly if it correlatedlessthan . 10 with every dependentvariablefor every year. After isolatingrelevantcovariates,a repeated-measures MANCOrAwas performedwith marital group(SSS, SSM, SMM and MMM) and gender(male, female) as betweensubjectsfactors, time (1983, 1984 and 1985) as the within-subjects factorand educationand race as constant covariates.Althoughthe underlyingdistributions of the dependentvariableswere highly non-normal,we chose not to transformthe data due to the extremelylarge size of our sampleßProvidedthat the numberof observations is sufficientlylarge,evenseverenon-normality in the distributionsof the variablesdoes not distort the resulting probabilitylevelsand confidenceintervalssincethe test statisticsare relativelyrobustto violationsof normality assumptions undersuchcircumstances (Bock, 1985).The largesamplesizein the presentstudy(in excessof 10,000 cases)thereforeallowedus to retain an interpretablemetric for the drinking variables.We conductedseparate

MILLER-TUTZAUER,

LEONARD AND WINDLE

analysesfor each set of repeateddependentmeasures: ADQ (averagedaily quantity) for 1983, 1984 and 1985; FHD (days/month frequentheavydrinking)for 1983, 1984 and 1985; alcohol-relatedproblemsfor 1984 and 1985; and dependency symptomsfor 1984 and 1985. The rationale behindconductingseparateanalysesfor each set of measuresrelatesto the fact that problemand dependency data were availablefor only 2 of the 3 yearswhile ADQ and FHD data were availablefor 3 years.In addition,separate analysesallowed us to examinemean consumption patternsin contrastto patternsof problembehaviorslike frequencyof heavydrinking,alcohol-related problemsand dependencysymptoms.It could be argued, for instance, that transitionalchangesoccuronly amongsocialdrinkers while not amongheavier,problemdrinkers.Finally, comparisonsof the abstinencerates (proportion of persons reporting abstinenceacrossall three measurementoccasions)revealeddifferencesamongmarital groups:16% of those in the SSS and SSM groups,20% of those in the SMM groupand 23% of the MMM groupqualifiedas abstainers.Therefore, we examinedtrendsfor all drinking variablesboth includingand excludingabstainers.Exclusionof abstainershad no effecton any analysesotherthan to increasemean levels of drinking-relatedbehaviorsan equivalent amount across all measurementoccasions.In other words, exclusionof abstainersdid not changethe pattern of means but only shifted the means upwards. Thus, althoughwe performedeach analysistwice, all reported findingsreflect analysesperformedincludingabstainers. However, aside from an occasionalmarginal differencein pairwisecomparisons, no resultsdifferedbetween analyses. Alcoholconsumption The repeatedmeasuresanalysisfor absolutedrinking quantity (ADQ) showed a significant Time x Marital Group interaction (Wilks' lambda = .9965, 6 df, p < .001) as well as significantmain effects for time (Wilks' lambda = .9993, 2 df, p < .05), for gender (F = 464.91, 1/10,507 df, p < .001) and for marital group(F = 44.55, 3/10,507 df, p < .001). There was no significantMarital Group x Gendernor Time x Marital Group x Gender interaction, men and women showing the samepatterns of consumptionover time thoughthe women consistentlydrank less on averagethan did the men. Means for each marital group at each measurement occasionare reportedin Figure 1. Based upon post hoc comparisons (or = .05), the SSS and the SSM groupsdid not differ significantlyat Time 1, nor did the SSM and the SMM groupsdiffer from one another.The stablymarried group (MMM) was significantlydifferent from all other groupsat Time 1. At Time 2, the SSS group differed significantlyfrom the SSM group.The SSM group, in turn, differed significantlyfrom the SMM and MMM

437

0.7

0.6

0.5

0.4

0.3

Groups 0.2

SSS



SSM



SMM •

0.1

i

1983

i

1984

MMM i

1985

Year

FIGURE1. Absolutedrinkingquantityby marital group

groups.Finally, at Time 3, only the SSS groupdiffered from the others.We performedadditionalpost hoc comparisonsof the changeswithin marital groupsfrom Time I to Time 2 and from Time 2 to Time 3. The Time

1 to

Time 2 changein absolutedrinkingquantityfor both the SSM and SMM groupsdifferedsignificantlyfrom the stably single and stably married groupsbut did not differ from eachother.Both transitionalgroupshad lower levels of consumptionthan in the precedingyear, whereasthe nontransitional groupsexhibiteda reversetrend. The Time 2 to Time 3 contrastswere not as conclusive,showing only that the changepatternsof the stably married and the stablysingledifferedfrom one another.However,the changefor the SSM groupwas very close(p = .0518) to being significantlydifferentfrom the MMM group. The analysisof frequencyof heavy drinking among

maritalgroupsparalleledthe resultsfor absolutedrinking quantity.Again, there was a significantTime x Marital Group interaction (Wilks' lambda= .9968, 2 df, p < .001), a significant main effect for time (Wilks' lambda--- .9987, 6 df, p < .01), a significantmain effect for gender(F = 320.56, 1/10,508 df, p < .001) and a significantmain effect for marital group (F-- 30.78 3/ 10,508 df, p < .001). Again, we found no differencesin patternsfor men andwomen.Patternsof changeandpost hoccomparisons werevirtuallyidenticalto thoseobtained for the ADQ measure,with one minorexception:At Time 2, the frequency of heavydrinkingfor the SSM groupdid

438

JOURNAL

OF STUDIES

ON ALCOHOL

/ SEPTEMBER

1991

TABLE2. Percentageof respondents with alcohol problems,dependencysymptoms a

0.20

Problems Group SSS SSM SMM MMM

0.15

Dependency

1984

1985

1984

1985

24 17 18 14

19 15 14 11

26 18 20 14

23 17 17 13

aPercentages reflectpersonsreportingone or more problemsor symptoms and are adjustedfor the covariateseducationand race. 0.10

Alcohol-relatedproblemsand dependency

Groups

0.05

-

SSS



SSM



SMM --

0.00

I

1983

1984

MMM I

1985

Year

FIGURF2.

Proportionof heavydrinkersby maritalgroup

not differ significantlyfrom that of the SMM groupeven thoughthe SSM groupdid differ from the MMM group. Given the high correlationwe obtainedbetweenabsolute drinking quantity and frequency of heavy drinking (r = .88 for 1983, .89 for 1984 and .85 for 1985;

p < .001 in eachcase), sucha correspondence in patterns for the two variable setsis not surprising.However,one could argue that changesmay have occurredonly among thosewith the least problematicdrinking behaviors,such that heavier drinkerssimply continuedtheir premarital consumptionpatterns. To examine this possibility, we computedthe proportionof personswithin each marital group who drank heavily at least four times per month. The resultingproportions,adjustedfor covariates,are displayedin Figure2. We foundthe patternfor proportionof heavy drinkersidenticalto the patternfor meandaysof heavy drinking. Post hoc comparisonsof the Time 1Time 2 changesshowedresultssimilar to thosefor absolute drinking quantity. Both transitionalgroupsdiffered significantlyfrom the nontransitional groupsand did not differ from each other. From Time 2 to Time 3, however,

the SSM groupdid not differ significantlyfrom the stably singlegroupbut did differ from boththe SMM andMMM groups.As can be seenin Figure2, the SSM groupexhibited a decreasein frequencyof heavydrinkingwhile the SMM and stably married groupsshowedlittle change from Time 2 to Time 3.

Results for alcohol-relatedproblemsand dependency are not as clear as thosefor the alcoholconsumptionmeasures,due in part to the lack of data for the year 1983. However,we found a significantmain effect of marital group both for alcohol-relatedproblems (F = 25.22, 3/ 10,229 df, p < .001) and for dependencysymptoms (F = 24.86, 3/10,486 df, p < .001). Similarly, we obtained a significantmain effect of gender for alcoholrelatedproblems(F = 112.25, 3/10,229 df, p < .001) as well as for dependency symptoms(F = 163.78, 3/10,486 df, p < .001). The effect for time was also significant, both in the case of alcohol-relatedproblems(F = 28.11 1/10,233 df, p < .001) and dependency symptoms (F = 7.79, 1/10,490 df, p < .01). The Time x Marital Group interactionwas not significant.Posthoc comparisonsfor meanproblemsshowedthat at both Time 2 and Time 3 the SSS groupdiffered significantlyfrom all others, the SMM and SSM groupsdid not differ from one anotherand the SSM groupdid not differ from the MMM group.Similarresultswere obtainedfor meandependency symptoms,exceptthat the mean for the SSM group was higher(thoughnot significantlyso) than that of the SMM group at Time 3. The SSM groupdid not differ significantlyfrom the MMM groupat eitherTime 2 or Time 3 eventhoughit differedfrom the SMM groupon bothmeasurementoccasions.We also examinedthe proportionof personswithin each marital group who reportedexperiencingone or more instancesof the behavior,and found the samepatternsas thoseexhibitedby the mean measures.Table 2 outlinesthe proportionsafter adjustingfor covariates.

Discussion

Due to the availabilityof alcoholconsumption data for 3 yearsratherthanjust 2, the absolutedrinkingquantity and frequencyof heavy drinking measuresprovidedthe clearestinformationon the impact of a changein marital statuson alcohol-relatedbehaviors.As shownin Figure2,

the stably single groups(SSS) exhibiteda consistently higher level of alcoholconsumptionthan did the stably

MILLER-TUTZAUER,

LEONARD AND WINDLE

married group (MMM). Of more interest, however,are the transitiongroups(SSM and SMM). The SSS and SSM groupswere not significantlydifferentat Time 1 (2 years prior to the transition to married status for the SSM group), whereasthe SMM group exhibitedsignificantly lower alcohol consumptionthan the stably single group during the first measurement occasion(1 year prior to marriage), suggesting that there is a diminutionof drinking prior to marriage. The pattern of means at Time 2 corroborated this. At Time 2, the alcoholconsumption of the SSM group(in the year prior to marriage)was significantly lower than that of the SSS group. This declinein drinking continuedinto the first year of marriage, as shownby the fact that groupsmaking the transitionto marriagestill differed significantlyfrom longer married groupsbut did not differ significantlyfrom othermarried groupsin the year in which the other groupsindicated they married. Similarly,the resultsfor frequencyof heavy drinking demonstratedthat personspreparingfor marriage, though still single, drank heavily less often than stably single individuals1 year prior thoughnot 2 years prior to marriage.As canbe seenin Figure2, thisprocess occurredprior to marriage and continuedinto the first year of marriage,but did not appearto continuebeyond that time. Thus, by the time a personenteredher/hissecond year of marriage,s/hewasunlikelyto differ from persons married for longer periods of time. Finally, the similar results obtained for both mean days of heavy drinkingand proportionof heavy drinkersare important and demonstratethat the changesextend beyond light drinkersaloneto thosemorelikely to exhibitmoreserious drinkingbehaviors.In otherwords,the reductionin drinking appearsto spanthe full spectrumof drinkers--from light to heavy--and does not reflect a change among lighter drinkersalone. Becausethe NLSY survey does not contain data on problemsand dependency in 1983, changescouldonly be assessedfor the 1984-85 time interval. As such, we could

not makethe sameprecisecomparisons as we madewith absolutedrinkingquantityand frequencyof heavydrinking. Although we found a significanteffect for marital group, examination of the means and proportions for alcohol-related problemsanddependency symptoms led to no clear interpretation.Surprisingly,the meanconsumption of the SMM group, married at Time 2, was higher thanthat of the SSM group,who were singleat Time 2. However,the meansof thesetwo transitiongroups,SSM and SMM, failedto differ significantlyat eithermeasurement occasion. Therefore, conclusionscould not be drawn

baseduponthe relativerankingof their means.Evenconvertingthe measures to proportions of personsexperiencing one or more problemsand proportionsof persons exhibitingone or moredependency symptomsdid not result in the emergence of any clear patterns.The inconclu-

439

sivenessof the problemand dependencydata is probably attributable

to the fact that measurements

were available

for only 2 of 3 years and the fact that base rates were particularly low given the sporadicnature of drinking problemsand dependency. Althoughdata concerningalcohol-relatedproblemsand dependencysymptomsfailed to show any consistentpattern, results relating to alcohol consumptionvariables providedevidenceof a periodof valuechangestartingapproximately 1 year prior to marriage and continuing throughthe first year of marriage. After that time, individualsmarriedfor only 1 year did not differ from those marriedlonger.This reductionin drinkingmay occuras a result of transitionto a new role--one that requiresbehaviorsdifferentfrom thoseassociatedwith being single, behaviorsnecessary to hold downa steadyjob, maintaina relationshipand so forth. Becauseextensiveuseof alcohol is inconsistentwith this new role and may actually interferewith the demandsimplicit in that role, individuals may find themselvesreducingtheir overall consumption and may insulatethemselvesfrom the opportunities for heavydrinkingthat were so prevalentprior to the decisionto marry. Sincethe drop in alcoholuseoccurredin the year prior to marriageratherthan in the marital transition year itself, we believethat the changesmore likely reflect a role transitionphase prior to marriage rather than the constraints that marriageitself can placeon opportunitiesto drink. Unfortunatelyinformationon such lifestyle characteristicsas presenceand number of children was not consistentlycollectedacrossall yearsfor all participants.However,any suchdifferenceswould likely producetheir effect in the yearssubsequent to the transition to marital statusrather than prior. The patternsof changewe obtainedare limited to the year prior to marriage and do not appearto continuelong after marriage, with all marital groupsexhibiting similar consumption patternsin the yearsfollowingthe marital transitionyear. Although the results point to a definite change in alcohol-related behaviors associated with the transition to

married status, we believe it is also important to note what the resultsdo not suggest.In particular,with respect to the alcohol measuresexaminedin our study,the find-

ingsdo not suggestpreexistingdifferences betweenthose who marry and thosewho do not. Only in the stagespreparatoryto marriagedid individualsbeginto differ from others who continuedbeing single. Individuals 2 years prior to marriagedid not differ from thosein the stably singlegroup. Whether the changesare part of a single processor representdistinctphasesin the marital transition is lessclear. It is possiblethat the changesprior to marriageare part of the whole processof courtshipand marriage,wherebychangesbeginduringthe early stages of courtshipand culminateat the time a personmarries. On the other hand, the changesmay representdiscrete

440

JOURNAL

OF STUDIES

ON ALCOHOL

phases.First, the personmay experiencea shift in values that includeshis/herbeliefsconcerningmarriageand the responsibilities marriageentails.A secondphasemay occur early in marriagewhenthe marital relationshipsignificantly alters the individual'senvironmentand therefore his/heropportunitiesto drink. Finally, it is also possible that a declinein drinkingmay facilitatean individual's moveinto marriage. Additionalresearchinto the relationshipbetweenmarriageand alcoholusewill haveto be performedto explicate the underlyingintra- and interpersonal processes in greaterdetail. Particularlyimportantwould be an examinationof thosepersonswhosepatternsof consumption do not changeas a resultof marriage. Questionscould be raisedas to why certainindividuals continueto engagein problembehaviorsin spiteof the manychangesthat generallyoccurduringthe transitionintomarriage.Similarly, baseduponavailabledata, it is uncertainwhat alcohol-use patternslook like in later years.As data collectionefforts continue,it will be possibleto look at trendsseveralyears beyondthe transitionto marriage.Also, it wouldbe interestingto examinemorecloselythosepersonswho divorce. Perhapsthosewho divorceare the sameindividualswho fail to changeduringthe transitionto marriage.Future efforts, by focusingon the marital relationshipspecifically, ratherthan relegatingit to an incidentalphenomenon, will no doubtclarify muchof our thinkingon the subjectof maturationaltrendsin alcoholbehaviors. References

BACON,S.D. Inebriety,socialintegration,andmarriage.Q. J. Stud.Alcohol 5: 86-125, 1944.

BAILEY,M.B., HABERMAN, P.Wo ANDALKSNE,H. The epidemiology of alcoholismin an urban residentialarea. Q. J. Stud. Alcohol 26: 19-40, 1965.

/ SEPTEMBER

1991

BOCK, R.D. Multivariate Statistical Methods in Behavioral Research,

2d Edition, Mooresville, Ind.: Scientific Software, Inc., 1985.

BOCK,R.D. ANDMOORE,E.G.J. Advantageand Disadvantage: A Profile of American Youth, Hillsdale, N.J.: Lawrence Erlbaum Assocs., Inc., 1986. BROMET,E. AND MOOs, R. Sex and marital status in relation to the characteristicsof alcoholics.J. Stud. Alcohol 37: 1302-1312, 1976.

CLARK,W.B. AND MIDANIK, L. Alcohol use and alcohol problems amongU.S. adults:Resultsof the 1979 National Survey.In: NATIONAL INSTITUTE ON ALCOHOL ABUSE AND ALCOHOLISM. Alcohol

Consumptionand RelatedProblems.Alcohol and Health Monograph No. 1, DHHS PublicationNo. (ADM) 82-1190, Washington: GovernmentPrintingOffice, 1982,pp. 3-52.

COLLINS,J.G. PersonsInjured and DisabilityDays Due to Injuries, United States, 1980-1981. Vital and Health Statistics, Series 10, No. 149, DHHS PublicationNo. (PHS) 85-1577, Washington:Govern-

mentPrintingOffice, 1985. COSPER, R. ANDMOZERSKY, K. Socialcorrelatesof drinkinganddriving. Q. J. Stud.Alcohol,Supplement No. 4, pp. 58-117, 1968. DUBLIN,L.I. Suicide:A Sociological and StatisticalStudy,New York: Ronald Press, 1963.

FLANAGAN, m.J. ANDMCGARRELL, E.F. (Eds.) Sourcebookof Criminal

JusticeStatistics--1985.U.S. Department of Justice,Bureauof Justice Statistics,Washington: Government PrintingOffice, 1986. KOTLER,P. AND WINGARD,D.L. The effect of occupational,marital andparentalroleson mortality:The AlamedaCountystudy.Amer.J. publ. Hlth 79: 607-612, 1989. MATTESSICH, P. ANDHILL, R. Life cycle and family development.In: SUSSMAN,M.B. AND STEINMETZ,S.K. (Eds.) Handbook of Mar-

riage andthe Family,New York:PlenumPress,1987,pp. 437-469. VERNON, S.W. AND ROBERTS,R.E. Use of the SADS-RDC in a tri-

ethniccommunitysurvey.Arch. gen. Psychiat.39: 47-52, 1982. WEISSMAN, M.M. Advancesin psychiatricepidemiology:Rates and risksfor majordepression. Amer.J. publ. Hlth 77:445-451, 1987. WEISSMAN, M.M. AND MYERS,J.K. Psychiatricdisordersin a U.S. community:The applicationof ResearchDiagnosticCriteria to a resurveyedcommunitysample.Acta psychiatr.scand.62: 99-111, 1980.

ZUCKER,R.A. Developmentalaspectsof drinking throughthe young adultyears.In: BLANE,H.T. ANDCI-IAFETZ, M.E. (Eds.)Youth,Alcohol, and SocialPolicy,New York: PlenumPress,1979, pp. 91146.

Marriage and alcohol use: a longitudinal study of "maturing out".

Earlier findings have suggested that marriage may provide protection from a variety of physical and psychological problems. In particular, numerous st...
822KB Sizes 0 Downloads 0 Views