Parenting Style and Smoking-Specific Parenting Practices as Predictors of Adolescent Smoking Onset

Journal of Pediatric Psychology pp. –, doi:./jpepsy/jsi

Journal of Pediatric Psychology © Society of Pediatric Psychology ; all rights reserved.

Parenting Style and Smoking-Specific Parenting Practices as Predictors of Adolescent Smoking Onset

Laurie Chassin,1 PHD, Clark C. Presson,1 PHD, Jennifer Rose,2 PHD, Steven J. Sherman,2 PHD, Matthew J. Davis,3 BA, and Jeremy L. Gonzalez,3 BA1Arizona State University, 2Indiana University, and 3Arizona State University

Objective To test whether parenting style and smoking-specific parenting practices prospec-

tively predicted adolescent smoking. Methods Three hundred eighty-two adolescents (age

10–17 years, initial nonsmokers, 98% non-Hispanic whites) and their parents were interviewed,

with smoking also assessed 1–2 years later. Results Adolescents from disengaged families

(low acceptance and low behavioral control) were most likely to initiate smoking. Adolescents’

reports of parents’ smoking-related discussion was related to lowered smoking risk for adoles-

cents with nonsmoking parents, but unrelated to smoking onset for adolescents with smoking

parents. Smoking-specific parenting practices did not account for the effects of general parent-

ing styles. Conclusions Both parenting style and smoking-specific parenting practices have

unique effects on adolescent smoking, although effects were largely confined to adolescents’

reports; and for smoking-specific parenting practices, effects were confined to families with

nonsmoking parents. Interventions that focus only on smoking-specific parenting practices may

be insufficient to deter adolescent smoking.

Because of the significant mortality and morbidity asso-ciated with cigarette smoking, combined with the difficultyof achieving cessation among smokers, considerablepublic health effort has been directed toward preventingsmoking onset in adolescence and identifying the processesunderlying smoking initiation (Lynch & Bonnie, 1994).In general, parenting and family factors have played arather minor role in these efforts, with greater emphasisbeing placed on peer social influences as well as onlarger social contextual factors such as cigarette adver-tising and taxation. Recently, however, there has beenincreased interest in family-based interventions both todeter adolescent substance use in general and to preventadolescent cigarette smoking specifically (Ashery,Robertson, & Kumpfer, 1998; Ennett, Bauman, Pemberton,et al., 2001).

One rationale for focusing on family influences isthat parenting has consistently been shown to be relatedto adolescent “problem behaviors,” including substanceuse (Baumrind, 1985, 1991). Research has suggested that

there are two key dimensions of parenting: parentalacceptance (nurturance, warmth, attachment) and behav-ioral control (monitoring, consistent discipline). Each ofthese dimensions has independently prospectively pre-dicted a variety of adolescent problem behavior, includingtobacco use (e.g., Ary, Duncan, Duncan, & Hops, 1999;Fleming, Kim, Harachi, & Catalano, 2002). Moreover, acombination of these two dimensions has been used todescribe four parenting “styles”: (1) authoritative (highlevels of acceptance and control), (2) authoritarian (highlevel of control but low level of acceptance), (3) indulgent(high level of acceptance but low level of control), and(4) disengaged or neglectful (low levels of acceptance andcontrol; see Baumrind, 1985; Maccoby & Martin, 1983).Previous studies of these four different parenting stylessuggest that adolescents from disengaged homes exhibit thehighest levels of problem behaviors (Darling & Steinberg,1993; Lamborn, Mounts, Steinberg, & Dornbusch, 1991;Steinberg, Lamborn, Darling, Mounts, & Dornbusch,1994), including cigarette smoking ( Jackson, Bee-Gates,

All correspondence should be sent to Laurie Chassin, Department of Psychology, PO Box 871104, Arizona State University, Tempe, Arizona 85287–1104. E-mail: [email protected].

Journal of Pediatric Psychology Advance Access published February 23, 2005 by guest on June 3, 2013

http://jpepsy.oxfordjournals.org/D

ownloaded from

Chassin et al.

& Henriksen, 1994). Theoretically, disengaged parentingraises a risk for adolescent problem behavior because thecombination of (1) a lack of emotional bonding to parentsand (2) a lack of supervision and consistent behavioralcontrol fails to provide a clear communication of parents’values and also undermines motivation for adolescents toattend and comply, thus weakening adolescents’ internal-ization of parental values and socialization (Baumrind,1991; Grusec & Goodnow, 1994).

However, from the point of view of prevention, anattempt to change broad general parenting styles involvingbehavioral control and affective bonds to children mayrequire an intensity of intervention that may not alwaysbe feasible. For this reason, there has also been interestin the hypothesis that parents’ socialization practicesand communication about a more delimited targetbehavior (such as smoking) may influence adolescents’behaviors. If smoking-specific parenting (i.e., parents’antismoking values, communications, and behaviors)can influence adolescents’ smoking, then these moredelimited aspects of parenting may be easier to modify inprevention programs than are broad, general parentingstyles (Ennett, Bauman, Pemberton, et al., 2001).

Existing data are conflicting about whether parents’smoking-specific parenting practices can deter adolescentsmoking. Mothers’ attitudes about their child’s smokinghave prospectively predicted smoking onset, but onlyamong nonsmoking mothers (Andersen et al., 2002),and the study by Andersen et al. did not assess how suchattitudes might have been translated into antismokingparenting practices. Cross-sectional studies have suggestedthat parental antismoking practices are correlated withadolescent nonsmoking (Chassin, Presson, Todd, Rose,& Sherman, 1998; Chassin, Presson, Rose, Sherman, &Prost, 2002; Jackson & Henricksen, 1997), but thecausal direction of these relations is unclear. In a pro-spective study with a large sample of parent/adolescentpairs recruited for a smoking prevention study, Ennett,Bauman, Foshee, Pemberton, and Hicks (2001) foundno evidence that parent/adolescent communicationabout tobacco or alcohol prospectively predicted drinkingor smoking onset. Moreover, there were marginal effectsin the opposite of the predicted direction in that parent/adolescent communication about rules about tobaccoand alcohol were associated with increases in smokingand drinking among those who had started. These resultscast doubt on the interpretations of cross-sectional rela-tions between smoking-specific parenting practices andadolescent smoking, which might reflect a reverse direc-tion of effect (i.e., an effect of adolescent smoking on thebehaviors of parents, rather than vice versa). The relations

between parenting and adolescent behaviors will likelybe bidirectional, as parents both react to and influencetheir adolescents’ behavior, and adolescents both reactto and influence their parents’ behaviors.

Moreover, the data on smoking-specific parentingpractices and adolescent smoking also vary dependingon whose perspective is assessed. Ennett, Bauman, Foshee,et al.’s (2001) results were obtained using parents’reports of parenting practices. Analyses of adolescents’reports (Ennett, Bauman, Pemberton, et al., 2001) foundsignificant relations between smoking-specific parentingand adolescent smoking. In general, studies that assessboth adolescent and parent reports of parenting suggestthat the extent of agreement between reporters is rathermodest (Feinberg, Neiderhiser, Howe, & Hetherington,2001; Pelegrina, Garcia, Linares, & Casanova, 2003;Tein, Roosa, & Michaels, 1994). Stronger prediction ofadolescent behavior from adolescent report than parentreport might be produced either by method variance(reports of parenting and smoking both derived from asingle reporter) or because it is parenting as perceivedby the adolescent that is more important. Accordingly,we assessed both parent and adolescent reports. Wehypothesized that adolescents who received more anti-smoking parenting would be less likely to initiate smok-ing but that effects of smoking-specific parenting (andgeneral parenting style) would be stronger for adoles-cents’ reports of parenting than for parents’ self-reports.

Another important question concerns the interplaybetween general parenting styles and smoking-specificparenting practices. Specific parenting practices maysimply be indicators of general parenting styles and maynot contribute uniquely to the prediction of adolescentbehavior (above and beyond general parenting styles).Alternatively, smoking-specific parenting practices maymediate the effects of general parenting styles. Perhapsdisengaged parents place their children at risk to smokebecause they provide the lowest levels of antismokingsocialization practices. If all of the effects of generalparenting are mediated through smoking-specific social-ization, then prevention programs might target onlysmoking-specific socialization for modification. How-ever, studies in other domains (e.g., child social compe-tence) found that mothers’ specific parenting practiceswere largely unrelated to their general parenting style(failing to support a mediational model). Rather, generaland domain-specific parenting each influenced differentaspects of social competence (Mize & Pettit, 1997).Finally, general parenting style might moderate theeffect of specific parenting practices (Darling & Steinberg,1993; Mize & Pettit, 1997). Perhaps smoking-specific

by guest on June 3, 2013http://jpepsy.oxfordjournals.org/

Dow

nloaded from

Smoking-Specific Parenting

punishment and discussion would be more effective inthe context of parenting styles with high levels of accep-tance and control. For example, adolescents in thoseauthoritative families might be more motivated to complywith parents’ socialization practices (Darling & Steinberg,1993; Grusec & Goodnow, 1994). Thus, we tested media-tional, moderational, and separate unique effects models ofthe effects of general parenting style and smoking-specificparenting on later adolescent smoking.

Another important but unresolved question iswhether antismoking parenting practices are effectiveeven for children whose parents themselves smoke ciga-rettes. In fact, children of smoking parents may not evenperceive their parents as communicating such antismok-ing parenting practices (despite parents’ reports of thesebehaviors; Chassin et al., 1998). Moreover, even if theantismoking parenting is accurately perceived by thechild of a smoking parent, it may not be effectivebecause the discrepancy between the parent’s attitudeand behavior may undermine their perceived legitimacyto regulate the adolescent’s smoking (Chassin et al.,2002). Previous data are conflicting, with some studies(e.g., Andersen et al., 2002) reporting that parents’ anti-smoking attitudes were ineffective in deterring adoles-cent smoking if the parents themselves smoked, andother (cross-sectional) studies reporting that antismok-ing parenting practices reduced the likelihood of adoles-cent smoking even among children of currently smokingparents ( Jackson & Henricksen, 1997). Accordingly, wetested whether smoking-specific parenting practicesreduced risk for later smoking equally among childrenof smoking and nonsmoking parents.

In summary, although there is growing interest inboth general parenting styles and smoking-specificparenting practices as influencing adolescent smoking(with the goal of constructing family-based preventionprograms), important questions remain. It is unclearwhether smoking-specific parenting practices prospec-tively predict adolescent smoking, whether they have aunique effect above and beyond general parenting style,and whether the effects of general parenting style may be(in part) mediated through smoking-specific parentingpractices. Finally, it is unclear whether smoking-specificparenting can influence adolescent smoking in familieswhere the parents themselves are current smokers. Thecurrent study provided a longitudinal, prospective test ofthe relations between general parenting styles and smok-ing-specific parenting practices and adolescent smoking,using both adolescent and parent reports of parenting.First, we hypothesized that smoking increases would behighest among adolescents from disengaged families that

provide low levels of acceptance and behavioral control,and also that smoking increases would be highest amongadolescents from families that provide low levels ofsmoking-specific discussion and punishment (and thatthese effects would be stronger in adolescent reports).Second, we tested whether the effects of general parentingstyle on adolescent smoking were mediated throughsmoking-specific parenting or whether general parentingstyle moderated the effects of smoking-specific parentingpractices. Finally, we hypothesized that the effects of bothgeneral parenting style and smoking-specific parentingpractices would be weakened in families with a currentlysmoking parent.

MethodParticipants

Subjects in the study were adolescents ages 10–17 andtheir parents, at least one of whom was also a participantin our ongoing, longitudinal smoking survey (Chassin,Presson, Sherman, Corty, & Olshavsky, 1984; Chassinet al., 1998, 2002). Participants in the larger longitudinalstudy were all 6th to 12th graders in a Midwesterncounty school district between 1980 and 1983, who werepresent on the day of testing and consented to participate.All original survey participants who were successfullyfollowed up in 1999 (72% of the original sample) andwho reported on their survey that they had a child betweenages 10 and 18 were invited by telephone or letter to par-ticipate in this family study (N = 861), and 556 (65%)accepted. Parents who refused were more likely to besmokers (45% refusers vs. 32.7% of participants) and lesslikely to have attended college (21.8% of refusers vs.38.8% of participants, p < .05), suggesting some caution ingeneralization.

For the current analyses, we selected adolescents whowere less than 18 years of age, lived at least part time with aparent or parents (so that they could be exposed to parentalsocialization), were nonsmokers at the initial measurement(i.e., either total abstainers or had “tried once or twice” butwere less than monthly smokers), and provided laterfollow-up information on their smoking status (N = 382,50.5% female, mean age = 12.5, 87% living full time withboth parents). Because the sample was 98% non-Hispanicwhite, ethnic differences were not examined.

Procedure

Families living in the vicinity attended a laboratory ses-sion, and others were interviewed in their homes by aninterviewer from a nearby university. For both parentsand adolescents (independently), the laboratory session

by guest on June 3, 2013http://jpepsy.oxfordjournals.org/

Dow

nloaded from

Chassin et al.

began with a bioassay to validate self-reported smoking(using a MicroCO to measure carbon monoxide in expiredair). Parents completed a questionnaire (in separate rooms).In a different room, an interviewer read the questionnaireitems aloud to the child, who responded on his/her owncopy. All participants were assured of confidentiality.

Follow-up data were collected from adolescents intelephone interviews over a 2-year period (94.6% of ado-lescents who met the current eligibility criteria, i.e.,382 out of 404, provided some follow-up information).

Measures

Parent and Adolescent Smoking StatusAdolescents reported whether they had ever smoked andtheir current smoking frequency, from less than monthlyto daily. Parents indicated their lifetime smoking status as(1) never smoked, not even a single puff, (2) smoked one ortwo, “just to try,” but not in the past month, (3) no longersmoke, but used to be a regular smoker, (4) smoke, butno more than once a month, (5) smoke, but no more thanonce a week, (6) smoke, but no more than once a day, and(7) smoke more than once a day. These items were thesame as those we have used in our longitudinal studysince 1980. Parents and adolescents also reported thetime since their last cigarette, from within the past hourto more than 5 years ago.

The bioassay supported the validity of these self-reports of current smoking. Those who did and did notself-report smoking in the past 4 hours were comparedin their bioassay readings as smokers (CO levels greaterthan 10 parts per million; cf. Wald, Idle, Boreham, &Bailey, 1981). The resulting kappas ranged from .73 foradolescents to .91 for mothers (p < .0001 for all).

For the current analyses, the outcome variable wasany reported increase in smoking by the time of the2-year follow-up among participants who initially wereeither abstainers or triers (20.7% [n = 79] reportedincreased smoking). This incidence is similar to nationalfigures that show a 15.7% rate of smoking onset among asomewhat older (12- to 17-year-old) sample over a 1-yearinterval (Koppstein, 2001). Baseline smoking (absti-nence vs. some trying) was used as a covariate in analyses.

Parent current smoking was coded as a dicho-tomous variable1 (neither parent a current smoker vs.

either parent a current [at least monthly] smoker), with39.4% of families (n = 219) reporting at least one parentwho was a current smoker. Although cell sizes did notpermit distinctions between ex-smokers and nonsmokers,our earlier cross-sectional analyses (Chassin et al., 2002)suggested that families with one ex-smoking parent andone currently smoking parent resemble families withtwo smoking parents in terms of their socialization.Families with one ex-smoking parent and one nonsmokingparent resemble families with two nonsmoking parents.

Family Structure A dichotomous variable was created to indicate whetheror not the child lived full time with both biological par-ents. Over half (56.9%) of the children lived full timewith both biological parents.

Parent EducationParent education consisted of the average of the mother’sand father’s report of their education level. Scores rangedfrom 1 (grade school) to 10 (completed graduate degree)with an average parent education level of 4.82 (SD = 2.26),which corresponds to having some post–high school edu-cation but not a college degree.

Parenting StyleParenting style was assessed at baseline using both ado-lescent reports and parent self-reports of parental behav-ioral control and acceptance. Behavioral control wasassessed with nine items from the Child Report of Par-ent Behavior Inventory (e.g., “I let my child get awaywithout doing work she/he has been given to do”;Schaefer, 1965). Response alternatives were on 5-point Lik-ert scales, from strongly agree to strongly disagree.Acceptance was assessed with seven items from the Net-work of Relationships Inventory (e.g., “How much canyour child rely on you to really care about him/her with-out changing from day to day?”; Furman & Buhrmester,1985). Response alternatives were on 4-point scales,from little or none to the most possible. Abbreviated ver-sions of these measures were used because of batterylength concerns. These abbreviated versions have showngood stabilities over a 1-year period, and good predictivevalidity in predicting adolescents’ externalizing out-comes (Stice, Barrera, & Chassin, 1993). Because ado-lescents’ perceptions of their mothers and fathers werehighly correlated (r = .69 for behavioral control and.48 for acceptance), they were averaged into measuresof perceived parental behavioral control and acceptance(coefficient alphas = .91 for behavioral control and .91for acceptance). This allows a fuller consideration of bothmaternal and paternal influences on adolescent smoking.

1Adolescent and parent smoking were dichotomized becauseof their distributions. Only 20% of initial adolescent nonsmokerswere nonabstainers at follow-up, suggesting that adolescent smokingwas best represented by abstinence versus any smoking. For parents,fewer than 5% of smokers reported frequencies less than daily,suggesting that adult current smoking was best represented byabstinence versus regular smoking.

by guest on June 3, 2013http://jpepsy.oxfordjournals.org/

Dow

nloaded from

Smoking-Specific Parenting

To compare our findings with those of previous stud-ies and to maintain the largest sample size, mothers’reports were used as the measure of parent report, exceptfor the small number of cases (3.5%) in which onlyfathers’ reports were available (coefficient alphas = .84 forbehavioral control and .77 for acceptance). Multiplereporter aggregates were not created because of low corre-lations between reporters. Correlations for mothers’ andfathers’ acceptance and behavioral control were .18 and.16. Correlations between mother and adolescent reportswere .15 and .22. Mean scores for behavioral control were1.89 for child reports (skewness = .48, kurtosis = −.39) and1.70 for parent reports (skewness = .73, kurtosis = 1.28),with higher scores indicating less behavioral control.Mean scores for acceptance were 3.20 for child reports(skewness = −.79, kurtosis = .64) and 3.54 for parentreports (skewness = −1.08, kurtosis = 1.26), with highscores indicating more acceptance.

As in previous research (e.g., Steinberg et al., 1994),parenting styles were created using median splits onbehavioral control and acceptance. For adolescentreports, 40.3% were categorized as having authoritativeparents (high behavioral control, high acceptance),15.2% as having authoritarian parents (high behavioralcontrol, low acceptance), 16.8% as having indulgentparents (low behavioral control, high acceptance), and27.7% as having disengaged parents (low behavioralcontrol, low acceptance). For mother reports, 40.0%were categorized as being authoritative, 12.0% as beingauthoritarian, 23.8% as being indulgent, and 24.3% asbeing disengaged. As expected from the low cross-reporter correlations, the agreement between classifica-tion based on parent report and adolescent report waslow (kappa = .13).

To examine the construct validity of this classifica-tion, we used the parenting classification to prospec-tively predict both parent-reported and child-reportedrebelliousness (measured by Smith & Fogg’s 1979 scale)and resistance to control (measured by the Resistance toControl subscale from Bates’s [1994] Youth Characteris-tics Questionnaire–Short Form), for a total of eightanalyses of variance. In support of construct validity,there was a significant prospective effect of parentinggroup on both rebelliousness and resistance to controlboth within reporter and across reporters (significant inseven of the eight models, significant F values rangedfrom 2.60 to 20.64, ps < .05). In all cases, children withauthoritative parents had the lowest levels of rebelli-ousness and resistance to control, and children fromdisengaged families had the highest levels of these char-acteristics in seven of the eight models.

Smoking-Specific Parenting Practices: Discussion and Punishment Smoking-specific parenting practices were assessed byadolescent reports and parent self-reports of two dimen-sions taken from our earlier work (Chassin et al., 1998)that asked about parental reactions (actual or hypothet-ical) to the child’s smoking. Smoking-specific discussionwas assessed with four items each about mother andfather (e.g., “If you found out that [name of target child]smoked cigarettes, would you talk with the child aboutreasons why he/she should not smoke?”). Smoking-specific punishment was assessed with five items eachfor mother and father (e.g., “If you found out that [nameof target child] smoked, would you withdraw privi-leges?”). Responses were on 5-point scales, from yes,definitely to no way. In support of the validity of thesemeasures, cross-sectional analyses (Chassin et al., 1998)found significant relations with adolescent smoking, forboth adolescent and maternal reports (all but motherreports of punishment, correlations from .18 to .49), aswell as significant cross-sectional relations between ado-lescent perceptions of parenting practices and mother’ssmoking (Chassin et al., 1998). Again, because adoles-cents’ reports about their mothers and fathers werehighly correlated (r = .63 for discussion and .79 for pun-ishment), these reports were averaged into measures ofperceived parent smoking-related discussion and punish-ment (coefficient alphas = .89 for discussion and .90 forpunishment). For consistency with the parenting stylemeasures, parent reports of smoking-specific punish-ment were mother-report measures unless only fatherreport was available (coefficient alphas = .64 for discussionand .77 for punishment). Again, multiple reporter aggre-gates were not created because of low correlationsbetween reporters. Correlations between mother andadolescent reports were .12 and .22, and correlationsbetween mother and father reports were .21 and .26.

ResultsRelations Between Parenting Style and Smoking Increase

We first tested whether parenting style was associatedwith an increase in smoking at the 2-year follow-up,using hierarchical logistic regression models (separatelyfor mother’s reports and adolescent’s reports). Child age,child baseline smoking, family structure, parent educa-tion, and parent smoking status were entered in the firstblock as covariates. In preliminary analyses, we alsotested the effect of gender, but because it was unrelatedto smoking increase, we did not include it in further

by guest on June 3, 2013http://jpepsy.oxfordjournals.org/

Dow

nloaded from

Chassin et al.

analyses. Parenting style categories were entered in thesecond block, represented by three dummy coded vari-ables, and the pairwise differences are presented inTable I. In the third block, interactions between parent-ing style categories and parent smoking were tested. (Inall models, other Covariate × Predictor interactions weretested but were found to be nonsignificant.)

Results of the logistic regressions are shown inTable I. Among the covariates, there were significanteffects for adolescents’ age and family structure such thatsmoking was more likely to increase among older childrenand among those who did not live full time with both bio-logical parents. Baseline smoking, parent education, andparent smoking were unrelated to an increase in smokingover the 2-year period. In the adolescent-report model, theblock adding the parenting style dummy variables wassignificant over and above the covariates, χ2(3) = 18.92,p < .0001. Adolescents with authoritative, authoritarian,and indulgent parents were significantly less likely toincrease smoking compared with adolescents with disen-gaged parents, and there were no other significant pairwisedifferences. The percentages of adolescents who increasedtheir smoking were 11.0%, 17.2%, 15.6%, and 39.6% forauthoritative, authoritarian, indulgent, and disengagedparenting, respectively. In the mother-report model, the

block adding the parenting style dummy variables was notsignificant over and above the covariates, χ2(3) = 3.82,p = .28. However, adolescents with authoritative motherswere marginally less likely to increase their smokingcompared with adolescents with disengaged mothers(OR = .53, p = .06), and there were no other pairwise dif-ferences. In terms of mothers’ reports, the percentages ofadolescents who increased their smoking paralleled thepattern for the adolescent-report model and were 15.5%,19.6%, 22.2%, and 29.9% for authoritative, authoritarian,indulgent, and disengaged parenting, respectively. Therewere no significant interactions between parenting style(either by adolescents’ or mothers’ reports) and parentsmoking in predicting adolescent smoking increase, andno other Covariate × Predictor interactions.

Relations Between Parenting Style and Smoking-Specific Parenting Practices

We next tested whether parenting style was related tosmoking-specific parenting practices using analyses ofcovariance with separate models for smoking-specificdiscussion and punishment. Covariates included adolescentage, adolescent baseline smoking, family structure, parenteducation, and parent smoking status. In the adolescent-report models, parenting style was significantly relatedboth to smoking-specific discussion, F(3, 389) = 15.09,p < .0001, and to punishment, F(3, 389) = 6.91, p < .0001.Parenting style explained 10% and 5% of the uniquevariance in discussion and punishment, respectively,over and above the covariates. In the mother-reportmodels, parenting style was significantly related tosmoking-specific discussion, F(3, 387) = 4.34, p < .01(3% of unique variance explained), but not to smoking-specific punishment, F(3, 387) = 1.38, p = .25 (1% ofunique variance explained). For both adolescents’ reportsand mothers’ reports, there were no significant interac-tions between parenting style and parent smoking inpredicting smoking-specific parenting practices and noother Covariate × Predictor interactions.

Table II shows mean levels of adolescents’ andmothers’ reports of smoking-specific discussion and punish-ment for each of the four parenting style groups. In themodels using adolescents’ reports, individual contrastsrevealed that adolescents from disengaged families repo-rted the least smoking-specific discussion (significantlydiffering from all the other three groups). Adolescents inauthoritative homes reported the most smoking-specificdiscussion, and those from authoritative and indulgenthomes reported significantly more smoking-specificdiscussion than did those from authoritarian homes.Adolescents from authoritarian and authoritative homes

Table I. Logistic Regression Results Relating Parenting Style to Adolescent Smoking Increase

Note: Interactions between parent smoking and parenting style were not

significant.

*p < .05, **p < .01, ***p < .001, ****p < .10.

� SE OR

Covariate block

Child age 0.45*** .14 1.57

Baseline smoking 0.34 .33 1.40

Family structure 0.78** .28 2.17

Parent education −0.19 .15 0.82

Parent smoking 0.09 .29 1.09

Parenting style block

Child report:

Authoritative vs. disengaged −1.26*** .35 0.29

Authoritarian vs. disengaged −1.12** .42 0.33

Indulgent vs. disengaged −1.26** .42 0.28

Authoritative vs. indulgent 0.004 .46 1.00

Authoritarian vs. indulgent 0.14 .51 1.15

Authoritative vs. authoritarian −0.14 .46 1.00

Mother report:

Authoritative vs. disengaged −0.64**** .34 0.53

Authoritarian vs. disengaged −0.38 .46 0.69

Indulgent vs. disengaged −0.19 .37 0.83

Authoritative vs. indulgent −0.46 .36 0.63

Authoritarian vs. indulgent −0.19 .47 0.83

Authoritative vs. authoritarian −0.27 .46 0.77

by guest on June 3, 2013http://jpepsy.oxfordjournals.org/

Dow

nloaded from

Smoking-Specific Parenting

reported the most smoking-specific punishment, andboth differed significantly from adolescents in disengagedfamilies (who reported the least punishment of smoking).Authoritarian parents were also significantly more likely topunish smoking than were indulgent parents.

For models using mothers’ reports, disengagedmothers were the least likely to discuss smoking and sig-nificantly differed from authoritative and indulgentmothers. Mothers’ self-reported punishment of smokingdid not differ as a function of parenting style.

Relations Between Smoking-Specific Parenting Practices and Adolescent Smoking

Third, we tested the relations between smoking-spe-cific parenting practices and adolescent smokingincrease, with four hierarchical logistic regressions test-ing the effects of smoking-specific discussion and pun-ishment separately (and separately for adolescents’ andmothers’ reports). Results are shown in Table III. Ado-lescents’age, baseline smoking, and family structurewere entered in the first block as covariates. As notedearlier, older adolescents and those who did not live fulltime with both biological parents were significantlymore likely to increase their smoking. In the secondblock, we entered parent smoking and either smoking-specific discussion or smoking-specific punishment; andin the third block, we entered the interaction of eithersmoking-specific discussion or smoking-specific punish-ment with parent smoking. The interaction wasincluded to test whether the relations between smoking-specific parenting and adolescent smoking increase dif-fered as a function of the parents’ own smoking status.In all models, all the other Covariate × Predictor interac-tions were tested but were not significant.

In the models using mothers’ reports, neither themain effects of discussion or punishment nor their inter-actions with parent smoking were significant. For themodels that used adolescents’ reports, there were nosignificant main effects of either discussion or pun-ishment. However, there were significant interactionsbetween parent smoking and both discussion, χ2(1, N =

380) = 8.62, p < .003, and punishment, χ2(1, N = 380) =6.09, p < .02 (see Table III). Probing these interactionsshowed that adolescents were less likely to increasesmoking in families that discussed smoking to a greaterextent, but this effect was present only in families inwhich the parents were nonsmokers (∃ = −.89, OR =.41, p = .004). Among families with a smoking parent,there were no significant effects of smoking-specificdiscussion on adolescent smoking (∃ = −.02, OR = .98,p = .940). For smoking-specific punishment, therewere no significant effects in either subgroup, but theinteraction was caused by differences in the directionof the (nonsignificant) effects (nonsmoking families,∃ = −.24, OR = .79, p = .384; smoking families, ∃ = .33,OR = 1.39, p = .212).

Table II. General Parenting Group Means for Smoking-Specific Discussion and Punishment (N = 404)

Note: Means in the same row that do not share subscripts differ at p < .05. Higher values indicate greater smoking specific discussion and punishment.

Hypothesized Mediator Authoritative Authoritarian Indulgent Disengaged

Child report n = 160 n = 61 n = 66 n = 113

Discussion 4.63a(.49) 4.29b(.67) 4.45ab(.67) 3.99 (.85)

Punishment 4.25a(.72) 4.28ab(.59) 3.96ac(.87) 3.74c(.84)

Parent report n = 160 n = 48 n = 95 n = 96

Discussion 4.86a(.26) 4.83ab(.32) 4.88a(.26) 4.71b(.41)

Punishment 3.80 (.78) 3.60 (.87) 3.55 (.84) 3.65 (.80)

Table III. Logistic Regression Results Relating Parenting Practices to Adolescent Smoking Increase

*p < .05, **p < .01, ***p < .001.

β SE OR

Covariates

Child age 0.44*** .14 1.56

Baseline smoking 0.35 .33 1.42

Family structure 0.79** .27 2.20

Parent education −0.20 .14 0.82

Parenting practice: discussion

Child report:

Parent smoking 0.02 .29 1.02

Discussion −0.29 .19 0.75

Parent Smoking × Discussion 1.08** .38 2.95

Mother report:

Parent smoking 0.05 .29 1.05

Discussion −0.30 .41 0.74

Parent Smoking × Discussion 0.15 .84 1.16

Parenting practice: punishment

Child report:

Parent smoking 0.10 .29 1.11

Punishment 0.11 .18 1.12

Parent Smoking × Punishment 0.84* .35 2.32

Mother report:

Parent smoking 0.16 .29 1.17

Punishment 0.24 .17 1.28

Parent Smoking × Punishment 0.31 .35 1.36

by guest on June 3, 2013http://jpepsy.oxfordjournals.org/

Dow

nloaded from

Chassin et al.

Tests of Mediated Effects

Do Smoking-Specific Parenting Practices Account forthe Effects of Parenting Style on Adolescent Smoking?To test this parameter, we used multiple group path analy-ses with Mplus software (Muthen & Muthen, 1998).These analyses tested the indirect effects of generalparenting style (represented in the model by three dummycoded variables) through smoking-specific punishment ordiscussion (in separate models) on smoking increase infamilies in which neither parent smoked versus familiesthat had at least one smoking parent. We divided the sam-ple by parent smoking using a multiple group analysisbecause of our finding (reported above) that the effects ofsmoking-specific parenting practices varied as a functionof parent smoking. Indeed, as expected based on thesefindings, there was a significant model, χ2(7) = 19.46, p =.007, that indicates that the pathways were not invariantacross parent smoking. Covariates were child age, parenteducation, family structure, and baseline smoking. Thepurpose of these multiple group path analyses was to pro-vide the path estimates required to test the significance ofthe mediated effects. Standard errors for the mediatedeffects were calculated using an extension of the multi-variate delta method (MacKinnon, Lockwood, Hoffman,West, & Sheets, 2002). These models were estimated onlyfor adolescents’ reports because mothers’ reports of smok-ing-specific parenting practices did not predict adolescentsmoking, either as a main effect or as an interaction withparent smoking.

The tests for the significance of mediated effectsindicated that the mediated effects of smoking-specificdiscussion and punishment were generally nonsignificant.Only one marginally significant mediating pathway wasidentified for discussion between the parenting styledummy variable representing indulgent parenting relativeto disengaged parenting and smoking increase for fami-lies in which a parent smoked (path estimate/standarderror = 1.74, p < .10).2 Thus, there was no strong evidencethat smoking-specific parenting practices mediated theeffects of parenting style on adolescent smoking.

Does Parenting Style Moderate the Effect of Smoking-Specific Parenting Practices on Adolescent SmokingIncrease? To test this parameter, we estimated fourlogistic regression models (separately for discussion andpunishment and for adolescents’ and mothers’ reports).In addition to the main effects of the covariates (describedearlier), the main effects of parenting style (three dummycoded variables), and the main effects of either discussionor punishment (in separate models), we also tested theinteraction of each dummy coded parenting style variablewith discussion or punishment (in separate models).There were no significant interactions in any of themodels. Thus, there was no evidence that parenting stylemoderated the relation between smoking-specific parent-ing practices and adolescent smoking increase.

DiscussionThe current study tested the roles of general parentingstyles and smoking-specific parenting practices in pro-spectively predicting smoking among adolescent non-smokers. The first finding of note was that generalparenting did significantly predict smoking. As hypothe-sized, and consistent with the broader literature on ado-lescent substance use in general (Hawkins, Catalano, &Miller, 1992) and adolescent smoking in particular( Jackson et al., 1994), adolescents who received lowlevels of parental behavioral control and acceptance(i.e., whose families were disengaged) showed the great-est prevalence of smoking onset. It is noteworthy thatthese findings were produced above and beyond theeffects of parental smoking, parental education, familystructure, and initial baseline smoking, providing astringent test of parenting influences. However, also aspredicted, these relations were stronger when adoles-cents’ perceptions of parenting were considered. Whenmothers’ reports were considered, the pattern was in thesame direction, but only one pairwise difference wasmarginally significant. This difference across reporters isalso consistent with previous research. For example,similar to our findings, Pelegrina et al. (2003) reportedlow agreement between adolescents and their parentsabout parenting characteristics, with reports producingstronger prediction of academic competence. Interest-ingly, this was true even when academic competencewas rated by teachers, suggesting that the finding couldnot be attributed to method variance. In our data, thefact that adolescents’ reports predicted their smokingmore strongly than did parents’ reports might indicatethat parenting style is effective only to the extent that itis perceived by the adolescent, or it might indicate thatthe effects of parenting style are weak and are partially

2Because these analyses considered mothers’ reports on them-selves but adolescents’ reports on both parents (aggregated), wealso repeated the analyses using the adolescents’ report of just themothers’ parenting. The general pattern of results was the same,although the effects were somewhat weaker (as would be expectedwhen the influence of fathers’ parenting was not considered). Inone notable exception, in the Mplus analyses of the mediatingeffects of smoking-specific discussion, there was an additionalsignificant mediated effect among families with a smoking parent.That is, smoking-specific discussion also significantly mediatedthe effect of general parenting on adolescent smoking in author-itarian families (versus disengaged families, z = 2.05, p = .04).

by guest on June 3, 2013http://jpepsy.oxfordjournals.org/

Dow

nloaded from

Smoking-Specific Parenting

inflated by the effects of method variance. However,given Pelegrina et al.’s findings, method variance aloneis not likely to explain the stronger predictive power ofadolescents’ perceptions.

These findings also support the rationale for family-based intervention as a way of deterring adolescentsmoking. If interventions can succeed in increasingparental acceptance and behavioral control, rates of ado-lescent smoking onset might be lowered. Indeed, signi-ficant effects of parenting programs on subsequentadolescent substance use have been reported, even whensubstance use was not an outcome that was targeted forintervention (e.g., Wolchik et al., 2002). Moreover,because the effects of general parenting style did notsignificantly vary with parental smoking status, thesebenefits were not limited to nonsmoking families. Thus,family-based interventions that affect general parentingcould prove beneficial even in homes where parents them-selves smoke.

General parenting style was also related to smoking-specific parenting practices such that children in theleast optimal parenting environment (disengaged fami-lies) also generally received the lowest levels of smoking-specific discussion and punishment (in both mother andadolescent reports of discussion and in adolescentreports of punishment). These relations might beexpected because parents who do not generally providehigh levels of nurturance or behavioral control to theiradolescents should also be less likely to provide eitherpositive or punitive socialization about any one behaviorin particular. However, although general parentingstyles were related to smoking-specific parenting prac-tices, the magnitude of the association was relativelymodest, and smoking-specific parenting practices werenot simple reflections of broader parenting styles. A sim-ilar finding of the relative independence of broad parent-ing styles and specific parenting practices has beenreported for other domains of child outcomes (Mize &Pettit, 1997). Indeed, in the broader social psychologicalliterature, available evidence suggests that relationsbetween general and specific attitudes and behaviors arealso modest (Ajzen & Fishbein, 1980). Thus, althoughparents may be generally accepting or generally control-ling, this does not mean that they will exhibit the sameextent of acceptance or control in the particular domainof their adolescent’s cigarette smoking.

Given this modest relation between general parent-ing style and smoking-specific parenting practices, it isnot surprising that our mediational models showed thatthe effects of general parenting style on adolescentsmoking could not be explained by their associated levels

of smoking-specific discussion and punishment. Rather,both general parenting style and smoking-specificparenting practices were unique (though correlated)prospective predictors of adolescent smoking. Adoles-cents in disengaged families are at risk for smoking (andother problem behaviors) not just because their parentsprovide low levels of antismoking socialization, but per-haps for several other reasons. For example, the low levelsof parental monitoring and behavioral control in disen-gaged families may allow adolescents to be exposed topeer contexts that support smoking (and other problembehaviors). Moreover, low levels of parental acceptancein disengaged families might make adolescents lessmotivated to comply with parental values and rules(Grusec & Goodnow, 1994).

The fact that smoking-specific parenting practiceswere unique predictors of adolescent smoking (aboveand beyond general parenting style) might suggest thatthey are useful targets for preventive intervention.Indeed, because smoking-specific parenting practicesmay be easier to influence than are more general parent-ing styles, it is tempting to conclude that interventionsshould focus only on these more specific aspects ofparenting behavior. However, other aspects of our find-ings suggest that such a recommendation would be pre-mature. First, smoking-specific parenting practices couldnot account for the effects of general parenting style onadolescent smoking. As others have argued (Darling &Steinberg, 1993; Jaccard, Dittus, & Gordon, 1998), thegeneral family environment and overall quality of therelationship between the adolescent and the parent must beconsidered, not just the parent/adolescent communicationabout specific problem behaviors. Second, the effects ofsmoking-specific parenting practices were detectableonly in adolescents’ perceptions of parenting. This repli-cates previous findings for other forms of problembehavior ( Jaccard et al., 1998) and suggests that parents’efforts at providing smoking-specific punishment anddiscussion may not necessarily be translated into aneffective deterrent for adolescent smoking unless theyare also perceived by the adolescent him or herself.Alternatively, these effects of smoking-specific parentingpractices in adolescents’ reports could be inflated bymethod variance. Finally, unlike the effects of generalparenting style, the effects of smoking-specific parentingpractices were qualified by parent smoking status, suchthat smoking-specific discussion was effective only whenprovided by nonsmoking parents. These findings suggestthat interventions that are focused solely on smoking-specific parenting practices could be limited in theirsuccess to a subgroup of families and that the families

by guest on June 3, 2013http://jpepsy.oxfordjournals.org/

Dow

nloaded from

Chassin et al.

who are most at risk (because they contain a smokingparent) would derive the least benefit. The finding thatsmoking-specific punishment showed a paradoxical ten-dency to escalate smoking among adolescents in familieswith a smoking parent replicates Ennett, Bauman, Foshee,et al.’s (2001) finding. Although this relation was not sta-tistically significant in either Ennett et al.’s study or in thecurrent data, it does suggest caution in recommendingpunitive strategies to parents who themselves smoke.

Finally, we found no evidence of interactionsbetween general parenting styles and smoking-specificparenting practices in prospectively predicting adoles-cent smoking. Thus, our data provided no evidence thatgeneral parenting style moderated the effects of smok-ing-specific parenting practices on adolescent smoking.However, given the relatively low power of such interactiontests, it would be premature to dismiss the possibilitythat general parenting styles could serve as moderatorsof the effects of smoking-specific parenting practices.

Although the current study contributed to the liter-ature by using a longitudinal design and multiplereports of both general and smoking-specific parenting,it is also important to consider some of its limitations.First, the relatively small number of participants whoincreased their smoking precluded the possibility of dis-aggregating transitions to first cigarette from transitionsto experimental smoking and to regular smoking, andstage models hypothesize that predictors can have differ-ent effects at different stages of the smoking acquisitionprocess (Mayhew, Flay, & Mott, 2000). This constraintalso limited our ability to model developmental differ-ences (or gender differences) in the smoking onset pro-cess. Second, the size of the sample also precludeddisaggregating ex-smoking from never-smoking parents,and the socialization of ex-smoking parents may havedifferent effects. Third, the sample was predominantlynon-Hispanic white, so that generalization to other eth-nic groups is not possible. Fourth, although the size ofthe sample provides ample statistical power for detectingmain effects and interactions of moderate magnitude, itis not optimal for detecting small interaction effects orthe multiple mediational chains that might occur if bothmaternal and paternal smoking-specific socializationwere tested as mediators. Finally, no observational mea-sures of parenting were included, and these measuresmight produce different findings and provide a betterunderstanding of the discrepancies that we observedbetween adolescent and parent reports.

In short, the current study tested the role of generalparenting styles and smoking-specific parenting prac-tices in prospectively predicting adolescent smoking.

Results suggested that disengaged parenting, which pro-vides low levels of acceptance and behavioral control,increases risk for adolescent smoking onset. Moreover,although general parenting was related to smoking-specificpunishment and discussion, its effects on adolescentsmoking could not be fully explained by smoking-specificparenting practices. Rather, both general parentingstyles and smoking-specific parenting practices hadunique effects on adolescent smoking, although theeffects of smoking-specific parenting were statisticallyreliable only in adolescents’ reports, and the effects ofsmoking-specific parenting practices were significantonly in families with nonsmoking parents. These find-ings suggest that interventions targeted solely at smok-ing-specific parenting practices may not be sufficient todeter adolescent smoking and that attempts to changemore general parent acceptance and behavioral controlmay be warranted, even though they may be more diffi-cult to achieve.

Acknowledgments

This research was supported by grant DA13555 from theNational Institute on Drug Abuse. Thanks go to AlmaTaubensee for help in data collection.

Received September 8, 2003; revisions received November13, 2003; accepted July 6, 2004

References

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall.

Andersen, M., Leroux, B., Marek, P., Peterson, A., Kealey, K., Bricker, J., et al. (2002). Mothers’ atti-tudes and concerns about their children smoking: Do they influence kids? Preventive Medicine, 34, 198–206.

Ary, D., Duncan, T., Duncan, S., & Hops, H. (1999). Adolescent problem behavior: The influence of parents and peers. Behavior Research and Therapy, 37, 217−230.

Ashery, R., Robertson, E., & Kumpfer, K. L. (Eds.). (1998). Drug abuse prevention through familyintervention (NIDA Research Monograph 177). Rockville, MD: National Institute on Drug Abuse.

Bates, J. E. (1994). Youth Characteristics Questionnaire (YCQ) (Short Form) scoring and psychometric

by guest on June 3, 2013http://jpepsy.oxfordjournals.org/

Dow

nloaded from

Smoking-Specific Parenting

information. Unpublished manuscript, Indiana University, Bloomington.

Baumrind, D. (1985). Familial antecedents of adolescent drug use: A developmental perspective. In C. L. Jones & R. L. Battjes (Eds.), Etiology of drug abuse: Implications for prevention (NIDA Research Monograph 56, pp. 3−14). Rockville, MD: National Institute on Drug Abuse.

Baumrind, D. (1991). The influences of parenting style on adolescent competence and substance use.Journal of Early Adolescence, 11, 56−95.

Chassin, L., Presson, C. C., Rose, J., Sherman, S. J., & Prost, J. (2002). Parental smoking cessation and adolescent smoking. Journal of Pediatric Psychology, 27, 485−496.

Chassin, L., Presson, C. C., Sherman, S. J., Corty, E., & Olshavsky, R. (1984). Predicting the onset of smoking in adolescents: A longitudinal study. Journal of Applied Social Psychology, 14, 224−243.

Chassin, L., Presson, C. C., Todd, M., Rose, J., & Sherman, S. J. (1998). Maternal socialization of adolescent smoking: The intergenerational trans-mission of parenting and smoking. Developmental Psychology, 34, 1189−1201.

Darling N., & Steinberg, L. (1993). Parenting style as context: An integrative model. Psychological Bulle-tin, 113, 487−496.

Ennett., S. T., Bauman, K. E., Foshee, V. A., Pemberton, M., & Hicks, K. (2001). Parent−child communication about adolescent tobacco and alco-hol use: What do parents say and does it affect youth behavior? Journal of Marriage and the Family, 63, 48−62.

Ennett, S. T., Bauman, K. E., Pemberton, M., Foshee,V. A., Chuang, Y.-C., King, T., et al. (2001). Media-tion in a family-directed program for prevention of adolescent tobacco and alcohol use. Preventive Medicine, 33, 333–346.

Feinberg, M., Neiderhiser, J., Howe, G., & Hetherington, E. (2001). Adolescent, parent, and observer perceptions of parenting: Genetic and environmental influences on shared and distinct perceptions. Child Development, 72, 1266–1284.

Fleming, C., Kim, H., Harachi, T., & Catalano, R. (2002). Family processes for children in early elementary school as predictors of smoking initiation. Journal of Adolescent Health, 30, 184–189.

Furman, W., & Buhrmester, R. (1985). Children’s perceptions of the personal relationships in their

social networks. Developmental Psychology, 21, 1015–1024.

Grusec, J., & Goodnow, J. (1994). Impact of parental discipline methods on the child’s internalization of values: A reconceptualization of current points of view. Developmental Psychology, 30, 4–19.

Hawkins, J., Catalano, R., & Miller, J. (1992). Risk and protective factors for alcohol and other drug prob-lems in adolescence and early adulthood: Implica-tions for substance abuse prevention. Psychological Bulletin, 112, 64–105.

Jaccard, J., Dittus, P., & Gordon, V. (1998). Parent–adolescent congruency in reports of adolescent sexual behavior and in communications about sex-ual behavior. Child Development, 69, 247–261.

Jackson, C., Bee-Gates, D., & Henricksen, L. (1994). Authoritative parenting, child competencies, and initiation of cigarette smoking. Health Education Quarterly, 21, 103–116.

Jackson, C., & Henricksen, L. (1997). Do as I say: Parent smoking, antismoking socialization, and smoking onset among children. Addictive Behaviors, 22, 107–114.

Koppstein, A. (2001). Tobacco use in America: Findings from the 1999 National Household Survey on Drug Abuse (Analytic Series: A-15, DHHS Publication No. SMA 02–3622). Rockville, MD: Substance Abuse and Mental Health Services Administration, Office of Applied Studies.

Lamborn, S., Mounts, S., Steinberg, L., & Dornbusch, S. (1991). Patterns of competence and adjustment among adolescents from authoritative, authoritar-ian, indulgent, and neglectful families. Child Development, 62, 1049–1065.

Lynch, B., & Bonnie, R. (1994). Growing up tobacco free: Preventing nicotine addiction in children and youths. Washington, DC: Institute of Medicine, National Academy Press.

Maccoby, E. E., & Martin, J. A. (1983). Socialization in the context of the family: Parent-child interac-tion. In E. M. Hetherington (Ed.), Socialization, personality, and social development (pp. 1–101). New York: Wiley.

MacKinnon, D., Lockwood, C., Hoffman, J., West, S., & Sheet, V. (2002). A comparison of methods to test mediation and other intervening variables. Psy-chological Methods, 7, 83–104.

Mayhew, K., Flay, B., & Mott, J. (2000). Stages in the development of adolescent smoking. Drug and Alcohol Dependence, 59, S61–S81.

by guest on June 3, 2013http://jpepsy.oxfordjournals.org/

Dow

nloaded from

Chassin et al.

Mize, J., & Pettit, G. (1997). Mothers’ social coaching, mother-child relationship style, and children’s peer competence: Is the medium the message? Child Development, 68, 312–332.

Muthen, L. K., & Muthen, B. O. (1998). Mplus: The comprehensive modeling program for applied researchers. Los Angeles: Muthen & Muthen.

Pelegrina, S., Garcia, A., Linares, M., & Casanova, P. (2003). Adolescents and their parents’ perceptions about parenting characteristics: Who can better pre-dict the adolescent’s academic competence? Journal of Adolescence, 26, 651–665.

Schaefer, E. (1965). Children’s reports of parental behavior: An inventory. Child Development, 36, 413–424.

Smith, G. M., & Fogg, C. P. (1979). Psychological ante-cedents of teen-age drug use. Research in Community and Mental Health, 1, 87–102.

Steinberg, L., Lamborn, S., Darling, N., Mounts, N., & Dornbusch, S. (1994). Over-time change in adjust-

ment and competence among adolescents from authoritative, authoritarian, indulgent, and neglect-ful families. Child Development, 65, 754–770.

Stice, E., Barrera, M., & Chassin, L. (1993). The relation of parental control and social support to adolescent externalizing outcomes: A longitudinal examination of curvilinear effects. Journal of Abnormal Child Psychology, 21, 609–629.

Tein, J., Roosa, M., & Michaels, M. (1994). Agreement between parent and child reports on parental behav-iors. Journal of Marriage and the Family, 56, 341–355.

Wald, N., Idle, M., Boreham, J., & Bailey, A. (1981). Carbon monoxide in breath in relation to smoking and carboxyhaemoglobin levels. Thorax, 36, 366–369.

Wolchik, S., Sandler, I., Millsap, R., Plummer, B., Greene, S., Anderson, E., et al. (2002). Six-year follow-up of preventive interventions for children of divorce: A randomized controlled trial. Journal of the American Medical Association, 288, 1874–1881.

by guest on June 3, 2013http://jpepsy.oxfordjournals.org/

Dow

nloaded from