ORIGINAL ARTICLE

Ethnic Variation in Parenting Characteristics andAdolescent Smoking

SOHAILA SHAKIB, Ph.D., MICHELE MOUTTAPA, M.A., C. ANDERSON JOHNSON, Ph.D.,

ANAMARA RITT-OLSON, DENNIS R. TRINIDAD, Ph.D., M.P.H.,

PEGGY E. GALLAHER, Ph.D., AND JENNIFER B. UNGER, Ph.D.

Purpose: To examine ethnic variation in parentingcharacteristics and in associations among parenting char-acteristics and adolescent smoking, and to assess thestrength of those associations separately by ethnicsubgroup.

Methods: Data were collected from a diverse sample(48.4% Hispanic/Latino, 22.9% Asian, 12.1% Non-His-panic white, and 16.6% Multiethnic) of 1846 SouthernCalifornia 6th-graders (mean age 11.3 years). Lifetimesmoking and parenting characteristics (parental smokingstatus, adolescents’ perceptions of parent–child commu-nication, and parental monitoring) were assessed. Acrossethnic groups, descriptive statistics for parenting andsmoking variables were calculated and compared. Logis-tic regression analyses controlled for demographics (so-cioeconomic status, generation status, family structure,age, and gender) and interaction terms (parenting char-acteristics � ethnicity), and were also run separately byethnicity.

Results: Asian adolescents reported less parental mon-itoring than Latino/Hispanics and less parental commu-nication than other ethnic subgroups. Odds ratios indi-cated parental monitoring (OR � .63; CI � .53-.74; p <.001) and parental communication (OR � .73; CI �.62–.86; p < .001) were protective whereas parental smok-ing (OR � 1.48; CI � 1.18–1.87; p < .01) was a risk factorfor smoking. Relative to Latinos, parental monitoring(OR � .50; CI � .26–.95; p < .05) was more protective, andparental communication (OR � 2.44; CI � 1.15–5.17; p <.05) less protective for white adolescents. All parentalcharacteristics were significantly associated with Latino/Hispanics smoking. Parental monitoring was signifi-cantly associated with Multiethnic and white smoking.No parenting characteristics were significant for Asians.

Conclusion: Ethnic differences in parenting character-istics and adolescent smoking should be consideredwhen designing prevention curriculum. Future researchshould investigate the nature of parent–child communi-cation among Asian immigrant populations. © Society forAdolescent Medicine, 2003

KEY WORDS:EthnicityAdolescentSmokingParental monitoringParental communication

The ethnic diversity of adolescents in the UnitedStates is transforming. Approximately 10.9 millionAsian Pacific Islanders [1] and 31.7 million Hispan-ic/Latinos inhabit the United States [2]. This ethnicdiversity is reflected in the adolescent population, aswell; currently, both Asian Pacific Islander and His-

From the Institute for Health Promotion & Disease PreventionResearch & Transdisciplinary Tobacco Use Research Center, Universityof Southern California, Keck School of Medicine, Alhambra, California.

Address correspondence and reprint requests to: Sohaila Shakib,Ph.D., USC Institute for Prevention Research, 1000 S. Fremont, Box 8,Alhambra, CA 91803.

This research was supported by the University of Southern CaliforniaTransdisciplinary Tobacco Use Research Center (TTURC), funded by theNational Institutes of Health (Grant 1 P50 CA84735-01) and theCalifornia Tobacco-Related Disease Research Program (TRDRP; Grant7PT-7004). Support for Dr. Shakib (T32 CA 09492) was providedthrough a National Cancer Institute postdoctoral fellowship awarded toC. Anderson Johnson, Ph.D., Director of the Institute for HealthPromotion & Disease Prevention Research, University of SouthernCalifornia. The authors thank Gaylene Gunning, Steven Cen, and theTTURC/IRP project staff for assistance with data collection and datamanagement.

Manuscript accepted March 10, 2003.

JOURNAL OF ADOLESCENT HEALTH 2003;33:88–97

© Society for Adolescent Medicine, 20031054-139X/03/$–see front matterPublished by Elsevier Inc., 360 Park Avenue South, New York, NY 10010doi:10.1016/S1054-139X(03)00140-X

panic/Latino are the fastest growing segments of theadolescent population [3]. Because adolescence is adevelopmental stage when health risk behaviors aretypically initiated [4], preventing health risk behav-iors among ethnically diverse adolescents is an im-portant public health goal. However, little researchhas been conducted on social influences that aresalient to the health behaviors of these adolescents.

Ethnic Variation in Adolescent Smoking

Adolescent smoking is an important health riskbehavior to examine. According to national data,increases in smoking have been identified amongAfrican-American, Hispanic/Latino, and white ado-lescents [5,6]. Although in the late 1990s, smokingbegan to decrease especially among African-Ameri-cans, between 1990 and 1996, in California an in-crease in smoking has been identified in Asians,Latinos, Native Americans, and white adolescents[7]. Because the majority of those who are nicotine-dependent report initiation during adolescence [8,9],smoking prevention efforts are one sensible strategyfor reducing the numbers of smokers. Before thedevelopment of adolescent smoking prevention cur-riculum for ethnically diverse adolescents, morestudies of social influences and adolescent smokingin the growing Asian Pacific Islander and Hispanic/Latinos adolescent population are necessary.

Parental Characteristics and Adolescent Smoking

Theories of social influences have been used exten-sively to study adolescent smoking. Studies thathave applied social learning theory [10–12] as atheoretical framework assume that adolescents learnbehaviors from significant others, through imitation,modeling, or vicarious learning. For instance, anadolescent observing his/her parent smoke for relax-ation may be consciously or unconsciously moti-vated to smoke for similar rewards. The majority ofstudies find strong support for the direct effects ofparents’ smoking on adolescent smoking [13–24].

Other theories have emphasized the influence ofsocial relationships. Studies of families have exam-ined the effects of qualities of parent–child relation-ships on adolescent smoking. In general, perceptionsof lack of parental concern and respect [20], parent–child conflict [15,24], and a lack of parental socialsupport [25,26] have been associated with adolescentsmoking. More mixed support has been found forthe effect of parental communication on adolescentsmoking [14,27].

Studies have examined parents’ role in adolescentopportunities to smoke. These associations betweenadolescent smoking and parental monitoring haveidentified parental monitoring as protective againstadolescent smoking [15,17,23,24,28,29]. Relative tothis research, few studies have examined ethnicvariation in parenting characteristics and adolescentsmoking [6].

Ethnic Variation in Parenting Characteristics

Evidence suggests that traditional frameworks ofadolescent health may be inappropriate across ethnicgroups. Some studies have identified ethnic variationin parenting characteristics; including parentingstrategies [30], expectations of parent–child relation-ships [31,32], and parental monitoring [33,34]. Rela-tive to Whites, less evidence for authoritative parent-ing styles (high parental demands with high parentalresponsiveness) has been found in African-Ameri-can, Asian-American, and Hispanic/Latino families[35]. Furthermore, some evidence exists that parent-ing characteristics might operate differently acrossethnic groups. In a study of school achievement,white adolescents benefited more from authoritativeparenting practices, than Asian, African-American,and Hispanic/Latino adolescents [36].

Using Appropriate Controls

Although numerous differences also exist [37,38],relative to other U.S. families important social andstructural characteristics are shared among the ma-jority of Asian Pacific Islander and Hispanic/Latinofamilies. Living in the United States, both AsianPacific Islander and Hispanic/Latino families tend tobe concentrated in urban metropolitan areas such asLos Angeles [1,2], to live in larger extended familieswith more children [39], and to have family memberspresent in the household who have recently immi-grated to the United States [40]. Because of theseunique differences, few studies of Asian Pacific Is-lander and Hispanic/Latino adolescents have effec-tively controlled for the influence of these factors. Asa result, the effects of race/ethnicity, independentfrom socioeconomic status, family level characteris-tics, and acculturation are difficult to ascertain.

Many of these factors influence adolescent healthrisk behavior. Living in urban metropolitan areas[41], particularly in disadvantaged communities [42],is associated with increased psychosocial risk factorsfor substance use as well as more exposure to oppor-

August 2003 PARENTING AND ETHNIC VARIATION IN SMOKING 89

tunities for use, particularly among ethnic minorityyouth [43]. Acculturation to the culture of the UnitedStates is related to family disruption; including fam-ily conflict, breakdowns in family communication,and alienation across generations [37,38]. For exam-ple, in immigrant families where values found in thenative culture vary extensively from those in the hostculture (i.e. more polarized versus less polarizedgender roles) there is evidence for higher divorceand/or separation rates [44]. Across numerous eth-nic groups, acculturation has been implicated as arisk factor for alcohol, drug use [45,46], and smoking[47].

Obstacles to Understanding Ethnic Variation inFamily Influences and Adolescent Smoking

More research on ethnic variation in adolescentsmoking is necessary. Most previous research hasbeen conducted with white samples [48]. Often,when non-white adolescents have been included inthe sample, their numbers have not been largeenough to make meaningful statistical comparisonswith white adolescents [19]. To compensate, manystudies have included ethnicity as a control variableand have not examined ethnicity by parenting vari-able interactions [49]. Therefore, more research onethnic variation in associations among parentingcharacteristics and adolescent health behaviors,clearly, is needed.

The Purpose of This Study

In a sample containing a sizeable number of AsianPacific Islander and Hispanic/Latino adolescents,this study examines parental influences on smoking.More specifically, the following are examined: (a)ethnic variation in parenting characteristics (parentalsmoking status, perceived parent–child communica-tion, and parental monitoring); (b) whether relation-ships among parenting characteristics and adoles-cent smoking are stronger or weaker in one ethnicgroup than in another; (c) the relative importance ofparenting characteristics on adolescent smoking foreach ethnic group separately.

MethodsSample

The data described in this article are from a baselinesurvey of a longitudinal school-based experimentaltrial of culturally tailored smoking prevention strat-

egies with ethnically diverse urban 6th-graders at-tending participating California middle schools.

School selection. Because the study focuses on thetwo largest ethnic groups in California (Hispanicsand Asians), the sample selection procedure wasdesigned to select schools with large proportions ofHispanic and Asian students. Data from the Califor-nia Board of Education and the Roman CatholicArchdiocese of Los Angeles and Orange Countiesidentified and classified schools as Asian, Latino, orethnically diverse. Schools were predominantlyAsian or Latino if: (a) at least 50% of the studentswere the ethnicity of interest, or (b) at least 35% of thestudents were the ethnicity of interest, and less than25% of the students were the other ethnicity. Schoolswere classified as ethnically diverse for not meetingthe above criteria (e.g., one school was 38% Asianand 44% Latino, and another was 25% Asian and29% Latino).

Student recruitment. All 6th-grade students at-tending the participating schools were invited toparticipate in the study. The study includes twocomponents with different informed consent re-quirements according to California law and the uni-versity’s Institutional Review Board, which ap-proved this study. The curriculum evaluationportion of the study required only implied parentalconsent and active student assent (i.e., studentscould choose to participate if their parents did notprovide a written refusal). The more extensive sur-vey of cultural and psychosocial variables, whichcontained more personal questions not directly re-lated to the content of the curriculum, required activewritten parental consent (i.e., students could chooseto participate only if their parents provided activewritten informed consent).

Of the 4427 students invited to participate, 3326(75.13%) provided active written informed parentalconsent whereas 681 (15.38%) provided implied con-sent. Of those students who provided active writtenparental consent, 217 did not complete the surveybecause they were absent from school on the day ofdata collection or they chose not to participate. Thedata for the current analyses are from the moreextensive survey requiring active written parentalconsent (parents signed and returned consent forms).In comparison to those students who provided activeparental consent (i.e., their parents neither providedactive consent nor refused to allow their child’sparticipation), those who provided implied consentwere significantly more likely to be male or African-

90 SHAKIB ET AL JOURNAL OF ADOLESCENT HEALTH Vol. 33, No. 2

American, to have poorer grades in school, and weremore likely to be lifetime ever-smokers.

Procedure

Students completed a 160-item paper-and-pencil sur-vey in their classrooms. The survey was distributedby a trained data collector, not previously acquaintedwith students during a single class period (45–50minutes). Surveys were identified only by a code, notwith students’ names or any other identifying infor-mation. Because the students all were attendingEnglish-language schools with classes conductedonly in English (as required by California law), abasic level of English-language proficiency was as-sumed, and surveys were provided only in English.However, students were encouraged to ask datacollectors to clarify meanings of unfamiliar words.

MeasuresLifetime Smoking

Owing to the low prevalence of smokers in this6th-grade sample, the question, “Have you ever triedsmoking, even a few puffs?” was used as the out-come variable. Those students who responded “no”and “yes” to the question were respectively coded as“Never smokers” and “Ever smokers.”

Ethnicity

Self-reported ethnicity was assessed with a series ofdichotomous questions (e.g., “Are you white, Chi-nese/Chinese-American, Pacific Islander, Filipino,Korean/Korean-American, Vietnamese/Vietnam-ese-American, Latino/Hispanic, or Black/African-American?”). Students were classified as Asian(those who answered “yes” to being Chinese, PacificIslanders, Filipinos, Koreans, or Vietnamese), Latino,or Multiethnic (those who selected at least two of thefollowing ethnic categories: Asian, Latino, andWhite). African-Americans were excluded from theanalyses because of a small sample size (n � 40). Inaddition, those students who choose none of theethnic identification questions (n � 212) were alsoexcluded from the analyses.

Parent Influence Variables

Parental communication. To assess perceived com-munication with parents, the following questionsadapted from the Cohen et al [50] communication

scale were asked: “How often do you talk with yourparents about what is on your mind?”; “How oftendo you ask your parents for advice?”; “How often doyou tell your parents your secrets?”; “If you had aproblem would you be able to talk to your parentsabout it?” Four response options were available,ranging from: 1 � “Very often,” to 4 � “Never”.Responses were averaged together and standardizedto a mean of 0 and a standard deviation of 1 to createa scale score. The Cronbach alpha was .81.

Parental monitoring. To ascertain perceptions ofparental monitoring, a five question series wasadapted from the Cohen et al [50] parental monitor-ing scale asking about parents’ knowledge and con-trol over student whereabouts: “Do you ever goplaces that your parents don’t want you to go?”;“When you go out with friends, do your parents askyou where you are going?”; “Are you allowed to goout with friends that your parents don’t know?”;“How important is it to your parents to know whereyou are at all times?”; “How often do your parentsreally know where you are?” Response optionsranged from: 1 � “Very often,” to 4 � “Never”.Responses were averaged together and standardizedto a mean of 0 and a standard deviation of 1 to createa scale score. The Cronbach alpha was .61.

Parental smoking status. To assess parental smok-ing status, the following question was asked: “Thinkabout the 2 adults you spend the most time with:How many smoke?” Response options were: 0 �“None,” 1 � “1 of them,” and 2 � “2 of them.”

Covariates

To control for confounding effects, the followingdemographic variables [age, gender, generation sta-tus, family household, and socioeconomic status(SES)] were included in the analyses as covariates.

Generation status. These three questions ascer-tained students’ generation status in the U.S: “Inwhat country were you born?”; “In what countrywas your father born?”; and “In what country wasyour mother born?” For all questions, response op-tions were: “United States” and “Other.” The “Oth-er” option included a line for writing in the countryname. Responses were coded in line with definitionsby Portes and Rumbault [38]: (a) “Immigrant-secondgeneration” if at least one parent was born outside ofthe United States and; (b) “third-generation” if thechild and both parents were born in the United

August 2003 PARENTING AND ETHNIC VARIATION IN SMOKING 91

States. In general, at this age, most foreign-bornadolescents are considered the 1.5 generation. In 26cases, the student was foreign-born and at least oneparent was U.S.-born. These cases were not includedin subsequent analyses.

Family structure. Family structure was assessedwith: “Mark all the people who live with you athome.” Response options included “Mother,” “StepMother,” “Father,” “Step Father,” “Sister(s) or Broth-er(s),” “Grandparent(s),” “Uncle(s) or Aunts,”“Cousins,” “Friends of the family,” and “Other.”Living with both biological parents was classified as“two-parent nuclear families.” Living with either oneparent or one parent and a stepparent was catego-rized as “alternative families.” Single parent andstepparent families were grouped together becausesmoking rates were similar among these groups.Because students from alternative families hadhigher rates of smoking (11.7%) than those classifiedotherwise (7.6%), this variable was dichotomized (1� alternative; 0 � two-parent nuclear families).

Socioeconomic status. Because questions askingabout parental occupation, education, and incomelack day-to-day relevance to experiences, responsesto traditional measures are difficult for adolescentsthis age to recall. The following more tangible ques-tions were used from the American Household Sur-vey as proxy measures: “How many people live inthe home where you spend most of your time(including you)?” (Students chose from 6 responseoptions ranging from “2 people” to “7 or morepeople”) and “How many rooms does your house orapartment have (Don’t count the kitchen or bath-room)?” (Students choose 7 response options rangingfrom “1 room” to “7 or more rooms”). For eachstudent, a proxy SES score was calculated by divid-ing the number of rooms by the number of peopleliving in the household.

Data Analysis

Descriptive statistics and differences by ethnicity. Forthe demographic and lifetime smoking variables,frequencies were calculated and compared acrossethnic groups using Chi-square analyses. Mean stan-dardized scores were calculated for parental moni-toring and parental communication. Parenting vari-ables were compared across ethnic groups usingANOVAs. Duncan’s post hoc tests assessed whetherethnic groups varied significantly on the parenting

variables. Bivariate correlations for all independentvariables revealed no multicollinearity problems.

Logistic regression analyses with interactionterms. Multiple logistic regression analyses assessedthe associations among parental influences and thesmoking variable, including the main effects for eachparent influence variable, ethnicity, parent influence� ethnicity interaction terms, and covariates. Theinteraction terms were constructed to assess whetherassociations among parental influences and adoles-cent smoking varied across ethnic groups. Being thelargest ethnic group in the sample, Latinos were thereference group used.

Logistic regression analyses for each ethnic group. Toexamine which parental influences had the strongestassociations with adolescent smoking for a particularethnic group, separate logistic regression analyseswere run for each ethnic group to assess the uniqueeffects of each independent variable while control-ling for the demographic covariates.

ResultsDemographic Characteristics of the Respondents

Of the 3109 students surveyed, slightly over half(59.3%) provided completed data on all variables ofinterest. The analytic sample size was 1846. Nosignificant differences between the analytic sampleand those with incomplete data were identified forage, gender, ethnicity, smoking, socioeconomic sta-tus, parental monitoring, and parental smoking.However, students in the analytic sample were morelikely to live with both biological parents (p � .0001),to come from the “immigrant second generation” (p� .05), and to report higher parental communication(p �. 05).

The analytic sample was ethnically diverse (48.4%Hispanic/Latino, 22.9% Asian American, 16.6% Mul-tiethnic, 12.1% white) with slightly more females(52.5% female) and early adolescents (mean age of11.3 years; SD � .51 years). Multiethnic studentswere a variety of ethnic combinations (78.8% Hispan-ic/white, 17.9% Asian/white, 9.8% Asian/Hispanic,and 3.3% Hispanic/Asian/white).

Demographic information identified several eth-nic differences (See Tables 1 and 2). Latino/Hispanicand Multiethnic students reported the highest ratesof smoking. A larger proportion of Hispanic/Latinowere female, whereas a larger proportion of Whiteswere male. Virtually all Asian and white students

92 SHAKIB ET AL JOURNAL OF ADOLESCENT HEALTH Vol. 33, No. 2

were classified respectively as “immigrant secondgeneration” and “third generation.” Asian respon-dents were more likely to live in two-parent house-holds. Hispanic/Latino and Multiethnic studentswere slightly older in age. Whites reported thehighest scores on measures of socioeconomic status,followed by Asians, Multiethnics, and Hispanics/Latinos.

Mean standardized scores for parental influencevariables by ethnicity were calculated (see Table 3).Latinos reported the highest rates of parental moni-toring, and Asians reported the lowest rates of pa-rental monitoring and parental communication. Nosignificant ethnic differences for parental smokingexisted.

Ethnic Variation in the Association BetweenParent Influence Variables and Smoking

Table 3 shows the main and interactive effects ofparent influences, ethnicity on adolescent smoking.

Again, all analyses were controlled for age, gender,ethnicity, family structure, generation status, andSES.

Main effects. Asian and white students (p � .06)appeared to have lower smoking rates than thosewho were Hispanic/Latino. Among the parentalinfluence variables, parental monitoring and paren-tal communication were protective factors, and par-ents’ smoking behavior was a risk factor for adoles-cent smoking.

Interactions With Parental Variablesand Ethnicity

In these analyses, an interaction term with a signifi-cant odds ratio suggests that the association betweenthe parental influence variable and smoking is eitherhigher or lower for the ethnic group than for His-panic/Latinos. The direction of the interaction effectdepends on the main effect of the parent variable.

Table 1. Frequencies on Categorical Covariates by Ethnicity

Latino(N � 893)

Asian(N � 422)

White(N � 224)

Multiethnic(N � 307)

Chi-square

n % n % n % n %

Ever smoked (lifetime smoking) 19.6**Ever smoked 99 11.1 18 4.3 13 5.8 29 9.5Never smoked 794 88.9 404 95.7 211 94.2 278 90.5Gender 10.0*

Male 394 44.1 204 48.3 121 54.0 158 51.5Female 499 55.9 218 51.7 103 46.0 149 48.5

Generation statusImmigrant 783 87.7 415 98.3 81 36.2 245 79.8 425.8***Non-immigrant 110 12.3 7 1.7 143 63.8 62 20.2

Family structureTwo parent 676 75.7 367 87.0 179 79.9 237 77.2 22.7***Alternative 217 24.3 55 13.0 45 20.1 70 22.8

* p � .05; ** p � .01; *** p � .0001.

Table 2. Mean Age, Socioeconomic Status (SES) and Parent Influence Variables by Ethnicity

Latino (N � 893) Asian (N � 422) White (N � 224) Multiethnic (N � 307) F-TestMean Mean Mean Mean

Age (years) 11.4a 11.2b 11.3b 11.3a 5.4**SES (rooms/people) 0.8d 1.3b 1.5a 1.1c 98.9***Parental monitoring

(standardized).08a �.09b .00ab .03ab 2.9*

Parental communication(standardized)

.10a �.15b .14a .00a 7.58***

Parental smoking (0, 1, or 2parents smoke)

0.41 0.33 0.35 0.38 1.48

* p � .05; ** p � .001; *** p � .0001.Note: a, b, c, and d are groupings derived from Duncan’s Multiple Range Test.

August 2003 PARENTING AND ETHNIC VARIATION IN SMOKING 93

Among the interaction terms computed, white �parental communication and Multiethnic � parentalcommunication were significant. Relative to Hispan-ic/Latinos, parental communication was a weakerprotective factor and parental monitoring was astronger protective factor for Whites.

Associations Between Parental InfluenceVariables and Smoking by Ethnicity

Table 4 shows the odds ratios and 95% confidence

intervals for parenting variables and lifetime smok-ing for each ethnic group separately to illustrate thepatterns of effects. For Latinos, parental monitoringand parental communication were protective factors,and parental smoking was a risk factor. Relative tofemales, Hispanic/Latino males reported higherrates of lifetime smoking. For Asians, age was theonly significant risk factor. For Whites and Multieth-nic students, parental monitoring was protectiveagainst lifetime smoking. For Multiethnic adoles-cents, living in alternative families was a significantrisk factor.

DiscussionIn a sample comprised of Asian, Hispanic/Latino,Multiethnic, and white Southern California 6th-grade adolescents, important relationships amongethnicity, parenting characteristics, and adolescentsmoking were examined. These results demonstratethat parental influences and the effects of theseinfluences on adolescent smoking vary among theethnic groups in this study.

Ethnic Differences in the Frequency of ReportingParenting Characteristics

Important ethnic differences in reporting parentingcharacteristics were identified. Latinos reportedhigher parental monitoring than Asians, whereaswhite and Multiethnic adolescents reported interme-diate parental monitoring. Asian adolescents re-ported less parental communication than all othergroups in the current study. Furthermore, parent–

Table 3. Ethnic Differences in Parental Influences onLifetime Smoking

Lifetime Smoking

OR 95% CI

Ethnic main effectsLatinos 1.00Asians (vs. Latinos) 0.40* 0.23, 0.69Whites (vs. Latinos) 0.53 0.27, 1.01Multiethnics (vs. Latinos) 0.80 0.51, 1.27

Parent Influence main effectsParental monitoring 0.63*** 0.53, 0.74Parental communication 0.73*** 0.62, 0.86Parents’ smoking 1.48** 1.18, 1.87

Ethnicity X Parent influence interaction effectsAsian X Parental monitoring 1.34 0.78, 2.28Asian X Parental communication 0.99 0.56, 1.78Asian X Parents’ smoking 1.19 0.53, 2.64White X Parental monitoring 0.50* 0.26, 0.95White � Parental communication 2.44* 1.15, 5.17White � Parents’ smoking 0.83 0.34, 2.02Multiethnic � Parental monitoring 0.68 0.43, 1.09Multiethnic � Parental communication 1.16 0.73, 1.85Multiethnic � Parents’ smoking 0.82 0.41, 1.63

* p � .05; ** p � .01; *** p � .001.Note. Main effects and interaction effects controlled for age,

gender, ethnicity, family structure, and SES.

Table 4. Odds Ratios of Lifetime Smoking by Ethnicity

Latino (n � 893) Asian (n � 422) White (n � 224)Multiethnic(n � 307)

OR (CI 95%) OR (CI 95%) OR (CI 95%) OR (CI 95%)

Demographic covariatesAge (years) 1.42 0.94, 2.14 3.37** 1.44, 7.88 1.32 0.39, 4.40 1.45 0.65, 3.23Gender

Female 1.00 1.00 1.00 1.00Male 1.96** 1.24, 3.11 1.06 0.38, 2.97 4.08 0.78, 21.33 1.24 0.51, 3.01

Family structureTwo parents 1.00 1.00 1.00 1.00Alternative 1.37 0.84, 2.23 0.34 0.04, 2.71 0.31 0.03, 2.84 3.80** 1.56, 9.28

SES 0.72 0.50, 1.05 0.94 0.47, 1.87 0.31 0.09, 1.14 0.77 0.43, 1.40Independent variablesParental monitoring 0.68*** 0.54, 0.85 0.92 0.55, 1.52 0.30*** 0.15, 0.60 0.42*** 0.27, 0.66Parental communication 0.63*** 0.50, 0.78 0.72 0.41, 1.25 1.48 0.70, 3.13 0.83 0.54, 1.27Parental smoking 1.38* 1.01, 1.88 1.71 0.80, 3.66 1.19 0.50, 2.85 1.22 0.64, 2.32

* p � .05; ** p � .01; *** p � .0001.

94 SHAKIB ET AL JOURNAL OF ADOLESCENT HEALTH Vol. 33, No. 2

child communication was not related to Asian ado-lescent smoking, but was related to smoking amongLatino, white, and Multiethnic adolescents. Com-pared with these other ethnic groups, Asian parentsmight have different ideas about what constitutesappropriate parenting during adolescence. Contem-porary and historical cross-cultural studies of ado-lescence have identified variation in the meaning ofadolescence across societies, including differences inchild-rearing beliefs and practices, parent–child re-lationships [31,32,34] and normative social affilia-tions [33]. Research conducted with adolescents liv-ing in China does demonstrate the importance ofparent–child communication and the family as asource of primary social support. For example, sur-vey findings report that the majority of urban Chi-nese youth seriously consider information from theirparents about how to choose appropriate marriagepartners [51]. There is also evidence that smoking isrelated to parent–child relationships among Chineseyouth [52]. The questions in our survey were aboutasking parents for advice, understanding rule mak-ing principles, and sharing secrets with parents. Thisstyle of communication may be more culturallyappropriate for Western families than for Asians.Communication between parent and child may occurin a different manner among Asian families andfuture studies should attempt to explore ethnic dif-ferences in parental communication.

Ethnic Variation in Parenting Characteristics andAssociation With Smoking

In comparison to Latinos/Hispanics, parental moni-toring had a marginally stronger effect on adolescentsmoking for whites. Parental communication, on theother hand, had a significantly stronger protectiveeffect for Hispanic/Latinos than for white adoles-cents. Among Hispanic/Latinos, successful commu-nication with parents might reduce acculturativestress [37,38], thereby reducing adolescent smoking.The finding that parental monitoring is more protec-tive against white adolescent smoking is consistentwith the evidence that white adolescents benefitmore from an authoritative parenting style [36],which include high parental demands.

The Relative Importance of Parenting Variablesand Adolescent Smoking

In addition to examining variations in parentingcharacteristics and adolescent smoking across ethnic

groups, the current study examines the relative im-portance of each parenting characteristic separatelyfor each ethnic group. In doing so, the current studycan identify the relative importance of parentalsmoking status, parental monitoring, and parentalcommunication for that ethnic group. According tothese analyses, all of the parenting characteristics,especially parental monitoring and communication,were strongly related to Hispanic/Latino adolescentsmoking. In addition, the magnitude of these associ-ations revealed both parenting variables were ofrelatively similar importance for these adolescents.

Numerous similarities between white and Multi-ethnic adolescents in parenting relative importanceof parenting characteristics existed. For both groups,parental monitoring was significant whereas paren-tal communication was not. Furthermore, for bothparental monitoring and parental communication,the magnitude of the association of each with smok-ing was similar. Interestingly though, for Multiethnicadolescents, coming from an alternative family (notliving with both biological parents but with a singlemother or in a stepparent family) was also significantfor adolescent smoking.

Limitations

The current study has several limitations that shouldbe reported. First, this study only assessed adoles-cents’ perceptions of their parents’ monitoring andcommunication behaviors. These perceptions there-fore may or may not accurately reflect parentalbehaviors. Future research should examine both ad-olescents’ and parents’ perceptions of these behav-iors. Adolescents’ subjective assessments of the qual-ity of communication with their parents might bewhat in reality have the greatest influence on theirhealth risk behaviors, but this is not yet known.Second, the associations reported here are cross-sectional, and therefore, the direction of causalitycannot be inferred. Parenting practices reflect com-plex processes. For instance, it is possible that theparenting practices reported in this study reflect aresponse to adolescents’ prior behavior, includingtheir smoking behaviors. Therefore, longitudinalstudies are necessary to understand the direction ofcausality. Third, because the Asian adolescents hadsuch low rates of smoking, it is possible that notenough variation in smoking patterns existed todetect statistically significant effects. Future studiesshould develop more appropriate parenting ques-tions and should examine older samples of Asianadolescents who smoke more. Fourth, because those

August 2003 PARENTING AND ETHNIC VARIATION IN SMOKING 95

adolescents with missing data were more likely tolive in non-intact families and to report low parentalcommunication, it is unknown whether the currentanalyses will generalize to those adolescents. Futurestudies should take measures to find out how toobtain more complete data from those adolescents.Fifth, family related characteristics were only col-lected from adolescents who provided active paren-tal consent. Those adolescents who provided activeparental consent were more likely to be African-American, to have poorer grades in school, and weremore likely to be lifetime smokers. Therefore, theresults of this study might not generalize to thoseadolescents. Finally, the current sample was notlarge enough to examine whether ethnic variation inassociations among parenting characteristics and ad-olescent smoking varied by gender. An examinationof the model by gender revealed the parental smok-ing by gender interaction was stronger for girls.Therefore, future studies should obtain sample sizeslarge enough to examine if associations vary withinethnic groups by gender.

Parental influences have an impact on adolescentsmoking behavior, and that impact varies in differentethnic groups. Parental monitoring appears to beparticularly relevant for white and Multiethnic ado-lescents, and parental communication appears to berelevant for Latino adolescents. Both parenting char-acteristics seem amendable to intervention, and fu-ture studies should attempt to increase these charac-teristics among specific ethnic groups.

References1. Humes K, McKinnon J. The Asian Pacific Population in the

United States. Washington, DC: U.S. Census Bureau, 2000:20–529.

2. Therrien M, Ramirez RR. The Hispanic Population in theUnited States. Washington, DC: U.S. Census Bureau, 2000:20–535.

3. Bureau UC. Population Profile of the United States, 1999.Washington, DC: U.S. Government Printing Office, 2001:23–205.

4. Schulenberg J, Maggs JL, Hurrelmann K. Negotiating devel-opmental transitions during adolescence and young adult-hood: Health risk and opportunities. In: Schulenberg J, MaggsJL, Hurrelmann K (eds). Health Risks and DevelopmentalTransitions During Adolescence. New York: Cambridge Uni-versity Press, 1997:1–19.

5. Kopstein AN. Trends in adolescent smoking in the UnitedStates: Data from the National Household Survey on DrugAbuse, 1994 through 1998. In: U.S. Department of Health andHuman Services (ed). Changing Adolescent Smoking Preva-lence: Where It Is and Why? Washington, DC: NationalInstitutes of Health, National Cancer Institute, 2002:51–67.

6. U.S. Department of Health and Human Services. Tobacco Useamong U.S. Racial/Ethnic Minority Groups—African Ameri-

can, American Indians and Alaskan Natives, Asian Americansand Pacific Islanders, and Hispanics: A Report of the SurgeonGeneral. Atlanta, GA: U.S. Department of Health and HumanServices, Centers for Disease Control and Prevention, NationalCenter for Chronic Disease Prevention and Health Promotion,Office on Smoking and Health, 1998.

7. Burns DM. Cigarette smoking among adolescents in Califor-nia, 1990–1996. In: U.S. Department of Health and HumanServices (ed). Smoking and Tobacco Control: Changing Ado-lescent Smoking Prevalence. Washington, DC: U.S. Depart-ment of Health and Human Services, National Institutes ofHealth, National Cancer Institute, 2002:69–99.

8. Choi WS. Which adolescent experimenters progress to estab-lished smoking in the United States? Am J Prev Med 1997;13:385–91.

9. Choi WS. Determining the probability of future smokingamong adolescents. Add Res Theory 2001;96:313–23.

10. Bandura A, Walters RH. Social Learning and PersonalityDevelopment. New York: Holt, Rinehart, and Winston, 1963.

11. Bandura A. Social Learning Theory. Englewood Cliff, NJ:Prentice-Hall, 1977.

12. Bandura A. Self-efficacy: The Exercise of Control. New York:W. H. Freeman and Company, 1997.

13. Farkas AJ, Distefan JM, Choi WS, et al. Does parental smokingcessation discourage adolescent smoking? Prev Med 1999;28:213–8.

14. Hundleby JD, Mercer GW. Family and friends as socialenvironments and their relationship to young adolescents’ useof alcohol, tobacco, and marijuana. J Marriage Fam 1987;49:151–64.

15. Biglan A, Duncan TE, Ary TV, Smolkowski K. Peer andparental influences on adolescent tobacco use. J Behav Med1995;18:315–30.

16. Glendinning A, Shucksmith J, Hendry L. Family life andsmoking in adolescence. Soc Sci Med 1997;44:93–101.

17. Jackson C, Henricksen L, Dickinson D, et al. A longitudinalstudy predicting patterns of cigarette smoking in late child-hood. Health Educ Behav 1998;25:436–447.

18. Sokol-Katz J, Dunham R, Zimmerman R. Family structureversus parental attachment in controlling adolescent deviantbehavior: A social control model. Adolescence 1997;32:199–215.

19. Grielsler PC, Kandel DB. Ethnic differences in correlates ofadolescent cigarette smoking. J Adolesc Health 1998;23:167–80.

20. Pederson LL, Koval JJ, McGrady GA, Tas SL. The degree andtype of relationship between psychosocial variables and smok-ing status for students in grade 8: Is there a dose responserelationship? Prev Med 1998;27:337–47.

21. O’Loughlin J, Paradis G, Reuaud L, Gomez LS. One-yearpredictors of smoking initiation and of continued smokingamong elementary school children in multiethnic, low-in-come, inner-city neighborhoods. Tob Control 1998;7:1–15.

22. Beck KH. Associations between parent awareness, monitor-ing, enforcement, and adolescent involvement with alcohol.Health Educ Res 1999;14:765–75.

23. Mott J, Crowe P, Richardson J, Flay B. After-school supervi-sion and adolescent cigarette smoking: Contributions of thesetting and intensity of after-school self-care. J Behav Med1999;22:35–58.

24. Simons-Morton B, Crump AD, Haynie DL, et al. Psychosocial,school, and parent factors associated with recent smokingamong early-adolescent boys and girls. Prev Med 1999;28:138–48.

96 SHAKIB ET AL JOURNAL OF ADOLESCENT HEALTH Vol. 33, No. 2

25. Aquilino WS, Supple AJ. Long-term effects of parentingpractices during adolescence on well-being outcomes inyoung adulthood. J Fam Issues 2001;22:289–308.

26. O’Byrne KK, Haddock CK, Poston WSC. Parenting style andadolescent smoking. J Adolesc Health 2002;30:418–25.

27. Doherty WJ, Allen W. Family functioning and parental smok-ing as predictors of adolescent cigarette use: A six-yearprospective study. J Fam Psychol 1994;8:347–53.

28. Chilcoat H, Anthony J. Impact of parent monitoring oninitiation of drug use through late childhood. J Am AcadChild Adolesc Psychiatry 1996;35:91–100.

29. Richardson J, Radziszewska B, Dent CW, Flay BR. Relation-ship between after-school care of adolescence and substanceuse, risk taking, depressed mood, and academic achievement.Pediatrics 1993;92:32–8.

30. Laosa L. Maternal teaching strategies in Chicano and Anglo-American families: The influence of culture and education onmaternal behavior. Child Dev 1980;51:759–65.

31. Facio A, Batistuta M. Latins, Catholics and from the far south:Argentinian adolescents and their parents. J Adolesc 1998;21:49–68.

32. Gjerde P, Shimizu H. Family relationships and adolescentdevelopment in Japan. J Res Adolesc 1995;5:281–318.

33. Pomfret DM. Representations of adolescence in the moderncity: Voluntary provision to work in Nottingham and Saint-Etienne, 1890–1914. J Fam Hist 2000;459–79.

34. Mazor A. Kibbutz adolescents. J Youth Adolesc 1993;22:569–72.

35. Steinberg L, Darling N, Fletcher AC. Authoritative parentingand adolescent adjustment: An ecological journey. In: Moen P,Elder JGH, Luscher K (eds). Examining Lives in Context:Perspectives on the Ecology of Human Development. Wash-ington, DC: American Psychological Association, 1995:423–66.

36. Steinberg L, Dornbusch SM, Brown BB. Ethnic differences inadolescent achievement: An ecological perspective. Am Psy-chol 1992;47:723–9.

37. Portes A, Rumbaut RG. Immigrant America: A Portrait. LosAngeles, CA: University of California Press, 1998.

38. Portes A, Rumbaut RG. The Story of the Immigrant SecondGeneration: Legacies. Los Angeles, CA: University of Califor-nia Press, 2001.

39. Day JC. Projections of the number of households and familiesin the United States 1995–2010. U.S. Bureau of the Census,

Current Population Reports. Washington, DC: U.S. Govern-ment Printing Office, 1996:25–1129.

40. Hansen KA, Faber CS. The foreign-born population: 1996.Washington, DC: U.S. Bureau of the Census Current Popula-tion Reports, 1997:20–494.

41. Gutierrez ECZ, Kendall C. The globalization of health anddisease: The health transition and global change. In: Scrim-shaw SC (ed). Handbook of Social Studies in Health andMedicine. Thousand Oaks, CA: Sage Publications, 1998:84–99.

42. Robert SA, House JS. Socioeconomic inequalities in health:Integrating individual-, community-, and societal level theoryand research. In: Scrimshaw SC (ed). Handbook of SocialStudies in Health and Medicine. Thousand Oaks, CA: SagePublications: Thousand Oaks, 1998:115–35.

43. Crockett LJ. Cultural, historical, and subcultural contexts ofadolescence: Implications for health and development. In:Hurrelmann K (ed). Health Risks and Developmental Transi-tions During Adolescence. Cambridge: Cambridge UniversityPress, 1997:23–53.

44. Landale NS, Ogena N. Migration and union dissolutionamong Puerto Rican women. Int Migr Rev 1995;29:671–93.

45. Epstein JA, Botvin GJ, Diaz T. Ethnic and gender differences insmoking prevalence among a longitudinal sample of inner-city adolescents. Health Educ Res 1998;23:160–6.

46. Epstein JA, Botvin GJ, Diaz T. Linguistic acculturation associ-ated with higher marijuana and polydrug use among Hispanicadolescents. Subst Use Misuse 2001;36:477–99.

47. Unger JB, Truz TB, Chen X, et al. English language use as arisk factor for smoking initiation among Latino and Asianadolescents: Evidence for mediation by tobacco related beliefsand social norms. Health Psychol 2000;19:403–10.

48. Tyas SL, Pederson LL. Psychosocial factors related to adoles-cent smoking: A critical review of the literature. Tob Control1998;7:409–20.

49. Wills TA, Vaughan R. Social support and substance use inearly adolescence. J Behav Med 1989;12:321–39.

50. Cohen DA, Richardson J, LaBree L. Parenting behaviors andthe onset of smoking and alcohol use: A longitudinal study.Pediatrics 1994;94:368–75.

51. Riley NE. Interwoven lives: Parents, marriage, and guanxi inChina. J Marriage Fam 1994;56:791–806.

52. Shakib S, Zheng H, Mouttapa M, et al. Family functioning,familial smoking status, and adolescent smoking behavior inWuhan, China. Transcultural Perspectives on Tobacco Useand Health Promotion Conference, 2002.

August 2003 PARENTING AND ETHNIC VARIATION IN SMOKING 97