Preventive Medicine - CiteSeerX

Preventive Medicine 31, 107–114 (2000) doi:10.1006/pmed.2000.0674, available online at http://www.idealibrary.com on

A Model of Smoking among Inner-City Adolescents: The Role of Personal Competence and Perceived Social Benefits of Smoking1 Jennifer A. Epstein, Ph.D.,2 Kenneth W. Griffin, Ph.D., and Gilbert J. Botvin, Ph.D. Institute for Prevention Research, Cornell University, Weill Medical College, New York, New York 10021

INTRODUCTION

Background: Based on current trends, smoking will remain a major public health problem in the 21st century. Effective smoking prevention approaches offer the best hope for decreasing the rise in adolescent smoking rates. Competence enhancement approaches to smoking prevention are among the most successful. Yet, there is not a full understanding of how effective prevention approaches work. This study tests whether a deficiency in competence (poor decision-making skills and low personal efficacy) is linked to acquiring beliefs in the perceived benefits of smoking and whether these perceived benefits are then related to subsequent smoking. Methods: A sample of 1459 students attending 22 middle and junior high schools in New York City participated. Students completed surveys at baseline, 1-year follow-up and 2-year follow-up during a regular class period. They self-reported smoking, decision-making skills, personal efficacy and beliefs in the perceived benefits of smoking. Results: The tested structural equation model had a good fit and was parsimonious and consistent with the theory underlying the competence approach to smoking prevention. Conclusions: This research highlights the importance of addressing decision-making skills, personal efficacy, and beliefs in the social benefits of smoking within adolescent smoking prevention programs. q 2000 American Health Foundation and Academic Press

Key Words: adolescence; cigarette smoking; ethnic minority populations; structural equation modeling.

1 This study was supported by Grant 1 R03 CA 73020 from the National Cancer Institute to J.A.E. Data collection for this study was supported by Grant 1 R18 CA 39280 from the National Cancer Institute to G.J.B. 2 To whom correspondence and reprint requests should be addressed at Institute for Prevention Research, Cornell University Medical College, 411 East 69th Street, KB 201, New York, NY 10021. Fax: (212) 746-8390, E-mail: [email protected].

Estimates indicate that in 1999 about 173,000 Americans are expected to die from cancer caused by tobacco use [1]. Still others will die of cardiovascular disease, chronic obstructive pulmonary disease, and other diseases associated with smoking. Notwithstanding decreases in cardiovascular deaths, smoking-related cancer deaths have continued to increase and since 1987 more women die from lung cancer than breast cancer, reversing a 40-year trend [1]. Cigarette smoking remains the leading preventable cause of premature disease and death in this country. In total, nearly one in every five deaths could have been averted if smokers had quit smoking or had never started smoking cigarettes at all. Unfortunately, over one million adolescents begin smoking each year in the United States. In fact, in the half-dozen-year period from 1991 to 1997, two national school-based surveys of adolescents, the Monitoring the Future Survey [2] and the Youth Risk Behavior Survey [3], reported that smoking rates increased among all youth, including the major ethnic groups (blacks, Hispanics, and whites). Considering that the vast majority of adult smokers start as adolescents, these trends suggest that smoking will most likely continue to be a major public health problem in the 21st century. For example, projections indicate that about five million people aged 17 years or under in 1995 are expected to die prematurely from smoking [4]. Furthermore, prospective research has demonstrated that smoking in adolescence increases the risk of later smoking as an adult [5,6]. Finally, ethnic minority populations appear at high risk for smoking due to increased cigarette advertising and promotions directed at them [7–9], with economicallydisadvantaged youth living in inner-city regions thought to be at greatest risk [10]. Given that most adult smokers began smoking as adolescents and the acknowledged difficulty in quitting smoking, adolescent smoking prevention should be a key strategy in reducing smoking. Great progress has

107

0091-7435/00 $35.00 Copyright q 2000 by American Health Foundation and Academic Press All rights of reproduction in any form reserved.

108

EPSTEIN, GRIFFIN, AND BOTVIN

been made in the smoking prevention field during the past two decades. Competence enhancement approaches to smoking prevention have been identified as among the most successful [11,12]. The competence enhancement approach conceptualizes smoking as socially- learned, purposeful, and functional behavior resulting from the interplay between social and personal factors. This approach, theoretically derived from social learning theory [13] and problem behavior theory [14], posits that the lack of generic skills for coping with life increases vulnerability to interpersonal and intrapersonal pressures to smoke. For example, some adolescents who are not successful academically or socially may begin to smoke as an alternative means of achieving popularity, social status, or self-esteem. These adolescents are most likely deficient in personal and social skills that would help them master the everyday challenges of this turbulent developmental stage. Individuals begin to acquire such basic skills during childhood; as they mature these skills typically increase. By the time individuals become adolescents, many have acquired a repertoire of skills (e.g., decision-making, refusing unreasonable requests) learned through a combination of modeling and reinforcement. The development of such skills depends upon having opportunities to observe and practice them. Personal efficacy is also a necessary part of developing social competence. Early evaluations testing competence enhancement approaches have shown 56–67% prevalence reductions in smoking among white middle-class students [15,16]. With the addition of booster sessions, these reductions have been as high as 87% relative to untreated controls [16]. Long-term follow-up data from one of the largest school-based prevention studies ever conducted found reductions in smoking 6 years after the initial baseline assessment [17]. Evidence of effectiveness of this approach in decreasing cigarette smoking has also been extended to inner-city Hispanic youth [18,19] and African–American youth [20] in the 7th grade, as well as among Hispanic students in a longer-term study through the end of the 10th grade [21]. While competence enhancement approaches to prevention have been effective, research that elucidates which factors are critical components of prevention programs is needed. Based on the theory underlying the competence enhancement approach, the goal of the present study was to examine whether a deficiency in competence is linked to beliefs promoting smoking as a means to popularity and whether these beliefs are in turn related to subsequent smoking. Specifically, greater competence as evidenced by sound decisionmaking skills and high personal efficacy will decrease the tendency to associate positive characteristics with cigarette smoking. The current study is meant to fill a gap in the smoking etiology literature and contribute to developing a greater understanding of how competence

enhancement approaches to smoking prevention work. This study concentrates on adolescents who did not receive any smoking prevention program to examine the natural processes associated with smoking behavior during middle school. Although some studies have found a direct association between smoking and poor decision-making skills [22] or efficacy [23,24], others conducted with predominantly ethnic minority innercity adolescents have not [25,26]. Thus a better and more complete test of the importance of decision-making and personal efficacy in adolescent smoking is to examine its indirect effect on smoking through beliefs in the perceived social benefits of smoking. To date, no research has tested such a model of adolescent smoking behavior. Consequently, general competence serves as a foundation for decreasing the possibility that an adolescent believes that cigarettes are a source of popularity and high status. Confidence in personal efficacy and skills like decision-making deflect such beliefs. It would not be possible for beliefs in the positive benefits of cigarettes to lower levels of general competence. Moreover, general competence does not always have a direct relationship with adolescent smoking. Finally, this research will focus on a predominantly minority innercity sample of adolescents especially since ethnic minority populations have been identified as being at high risk for smoking [27]. Yet little longitudinal information concerning smoking among minority adolescents is available. METHOD

Overview A total of 22 middle and junior high schools in New York City with 25% or more Hispanic students participated. These schools were the non-treatment control schools in a longitudinal smoking prevention trial [19]. New York City schools serve urban youth from lowincome families; the majority of the 22 schools served youth from families with average incomes at or below 150% of the Federal poverty level. All sixth and seventh graders in English-speaking, mainstream classes completed questionnaires. Cornell Medical College’s Institutional Review Board approved of the protocol for the study and the consent procedure. More than 90% of eligible students completed this initial survey. Students completed surveys at baseline, 1-year follow-up, and 2-year follow-up. Participants At baseline, 2400 students completed questionnaires. The retention rates over the course of the study (81% at 1 year and 63% at 2 years) compare favorably with retention rates for similar school-based studies [28–30].

A MODEL OF SMOKING AMONG INNER-CITY ADOLESCENTS

This is particularly impressive considering the recognized difficulty of conducting longitudinal research with inner-city minority youth due to high rates of mobility and absenteeism. The panel sample across the three assessment consisted of 1459 students (61% of baseline participants). The mean age at baseline for the panel sample was 12.4 (SD 5 0.75). The sample was 54% girls and 46% boys. In terms of ethnicity, this sample was 54% Hispanic, 20% African–American, 16% white, 7% Asian and 3% other. Seventy percent of the respondents lived in two-parent households. Procedure All participating students completed questionnaires at each assessment that measured self-reported smoking and psychological factors hypothesized to be related to adolescent smoking. A team of three to five data collectors, who were members of the same ethnic groups as the participating students, administered the surveys following the same standardized protocol at each assessment. Students completed these surveys during a regular 40-min class period. Teachers were not involved in data collection activities; students were assured that their answers would remain confidential. Quality of self-report data was assured by: (a) the use of student identification codes rather than names to emphasize the confidential nature of the surveys, and (b) the collection of carbon monoxide breath samples before students completed the questionnaire. The collection of physiological measures of smoking status (carbon monoxide breath samples) enhances the veracity of self-reported smoking data [31–33]. Training for data collectors included a systematic means of collecting carbon monoxide samples. At the same time that the students completed the questionnaires, carbon monoxide samples were collected in a procedure that lasts less than 1 min per student. Overview of Measures Students completed one of two randomly distributed questionnaire forms containing the same items with the order reversed for the measures on the last half of the questionnaire. Half the sample completed each form, maximizing the amount of data collected within the available time and minimizing data loss due to fatigue, boredom, or inadequate time. All data were selfreported. Included on the questionnaires were items concerning race/ethnicity, gender, age, smoking behavior of respondents, personal efficacy, decision-making skills, and beliefs in the social benefits of smoking. All of the items/scales were derived from psychometricallyvalid and widely used instruments. However, since the items used to measure several of these variables had originally been developed for use with white, middleclass students, they were pilot-tested and revised where

109

necessary to ensure their suitability for the target population [18]. Cigarette Smoking One dichotomous (yes/no) item assessed recent smoking during the past month. An 11-point smoking index assessed current smoking. Specifically, students responded to the question, “How often do you currently smoke?” Response options ranged from “I have never smoked” (1) to “A pack or more each day” (11). Future smoking was assessed by asking students to rate their behavioral intentions to be a smoker in 2 years on a 5point scale ranging from “I definitely will not” (1) to “I definitely will” (5). These three measures represent the full spectrum of adolescent smoking as past, present, and future indicators. Similar measures have been used in past research [2,34]. Although intentions to smoke measures are often overlooked, numerous studies of adolescent smoking include this type of measure [35– 38]. Future intentions are particularly important to measure because prospective research indicates that such intentions are strongly associated with actual subsequent smoking [39]. Finally, in the current study in the 2-year follow-up, the range of correlations among the three smoking items was 0.44 to 0.74 (all P’s , 0.01). Decision-Making Four items assessed decision-making skills (a 5 0.76). These items were derived from a subscale of the Coping Inventory [40] concerning problem-solving and direct action. The decision-making items assessed the use of sound decision-making skills (e.g., “when I have a problem I get information that is needed to deal with the problem”). Responses were rated on a 5-point scale which ranged from “never” (1) to “almost always” (5). Personal Efficacy Personal efficacy was assessed using four items (a 5 0.73) from the personal efficacy subscale of the Spheres of Control Scale [41]. This scale measured the extent to which respondents believed they could achieve personal goals through their own efforts (e.g., “I can learn almost anything if I set my mind to it”). Responses were scored on a 5-point Likert scales which ranged from “strongly disagree” (1) to “strongly agree” (5). Social Benefits of Smoking The perceived social benefits of smoking (e.g., “smoking cigarettes makes you look cool”) were measured with four items (a 5 0.85) derived from the Teenager’s Self Test: Cigarette Smoking [42]. Responses for each item were scored using a 5-point Likert scale which ranged from “strongly disagree” (1) to “strongly agree” (5).

110

EPSTEIN, GRIFFIN, AND BOTVIN RESULTS

Rates of smoking in this sample were typical for this age group: at baseline, 15.4% of participants reported that they had ever smoked, 10.6% smoked in the past year, and 3.3% smoked in past month. Smoking rates increased substantially from the baseline (sixth and seventh graders) to 2-year follow-up (eighth and ninth graders), with 30.4% of eighth and ninth graders reporting that they had ever smoked, 22.0% reporting smoking in the past year, and 10.6% reporting smoking in the past month. Attrition Analyses Significant mean differences on the smoking behavior measures were observed between the panel sample and dropout students over the course of the study (P’s , 0.001). Dropouts at the 1-year follow-up assessment were more frequent smokers (M 5 1.77) compared to the panel students (M 5 1.43). Likewise, dropouts at the 2-year assessment smoked more frequently (M 5 1.70) than panel students (M 5 1.37). However, the loss of high-end smokers due to attrition (absenteeism and relocation) would be expected to attenuate the estimation of variable relationships only minimally. Moreover, the loss of heavier smokers would make it more difficult

to detect a relationship between smoking behavior and potentially related factors. Treatment of Missing Data Since listwise deletion is inefficient due to the discarding of potentially useful data, procedures were used to help maximize the available data [43]. Analysis of non-response patterns showed that complete data on all 20 relevant items were available for 66% of the 1459 cases; an additional 9% of cases were missing one item, and 25% were missing two or more items. To maximize the number of cases available for analysis with structural equation modeling (which requires complete data) we used mean substitution for those cases that were missing only one item (5% of the items) and dropped those cases with two or more missing items. After adjusting for missing data, the final sample size was N 5 1,094 (75% of the panel sample of 1,459). Confirmatory Factor Analysis Prior to testing the hypothesized structural model, a confirmatory factor analysis (CFA) was performed to assess how well the observed measures reflect the hypothesized latent constructs. The EQS computer program [43] was used for the CFA and structural equations modeling (SEM). Five latent factors were specified

FIG. 1. Confirmatory factor analysis model. Large circles represent latent constructs, rectangles are measured variables, and singleheaded arrows designate residual variances. All loadings are statistically significant at P , 0.001.

111


in the measurement model (see Fig. 1). Three of the latent factors were composed of variables measured at the baseline assessment in sixth or seventh grade: the Baseline Smoking factor had loadings ranging from 0.45 to 0.94; the Decision-Making latent factor had loadings ranging from 0.65 to 0.68, and the Personal Efficacy latent factor had loadings ranging from 0.48 to 0.74. The Social Benefits of Smoking latent factor measured 1 year later in the seventh or eighth grade had loadings ranging from 0.59 to 0.90. The 2-Year FollowUp Smoking latent factor assessed in eighth or ninth grade had loadings ranging from 0.67 to 0.98. Factor loadings for all latent constructs were highly significant (P’s , 0.001) and in the expected direction, indicating that the measurement model was properly specified and that each factor was statistically reliable based on the hypothesized model. Several criteria were used to evaluate the overall fit of the CFA model and subsequent SEMs, including: (a) the x 2 to degree of freedom ratio, which should be less than 5.0 [44]; (b) the standardized root mean squared residual (SRMR), which should be less than .05; and (c) several fit indices including the Normed Fit Index (NFI), the Nonnormed Fit Index (NNFI), and the Comparative Fit Index (CFI). Each of these indices is derived by comparing the predicted covariation in the hypothesized model to that of the null model, with values greater than 0.90 indicating an excellent fit of the model to the data. The model fit criteria are those recommended in guidelines for reporting SEM results [45]. According to these criteria, the CFA model was a good to excellent fit; x 2/df 5 2.7; SRMR 5 0.043; NFI 5 0.95; NNFI 5 0.96; and CFI 5 0.97. The latent factor intercorrelations from the CFA model are shown in Table 1. Not surprisingly, the strongest relationship was between Baseline Smoking and 2-Year Follow-Up Smoking (r 5 0.39, P , 0.001). Decision-making and Personal Efficacy were moderately associated (r 5 0.29, p , 0.001). The Smoking outcome variable was significantly correlated in the expected direction with each of the four predictor variables. In summary, the CFA analysis demonstrated that the measurement model was excellent, with high factor

loadings for all indicator variables, and that the outcome latent factor of 2-Year Follow-Up Smoking was significantly correlated with all predictor latent factors. Structural Equations Model To test a formal model of the relationships between the predictor latent factors and outcome Smoking latent factor, a SEM was tested. The formal structural equations model differs from the CFA model in that arrows representing path coefficients have been added to show the hypothesized direction of relationships among the latent factors. As recommended by MacCallum et al. [46], the first step involved testing a saturated model, which estimated the paths from all exogenous latent factors to the construct of Social Benefits of Smoking and to the outcome Smoking latent factor, as well as the path from the Social Benefits of Smoking to the Smoking outcome. In addition, the covariances among all exogenous latent factors were estimated in testing the saturated model. The error terms for each matching smoking indicator at the two time points were freely estimated because it was expected that measurement error would be similar. This model had three exogenous latent factors (Baseline Smoking, Decision-Making, and Personal Efficacy), one mediating latent factor (Social Benefits of Smoking), and one outcome latent factor (2-Year Follow-Up Smoking). Subsequently, paths that were not significant were trimmed from the model (the direct effects each from baseline Decision-Making and Personal Efficacy to 2-Year Follow-Up Smoking). The results of testing this final model are illustrated in Fig. 2. Each of the three baseline exogenous latent factors directly predicted Social Benefits of Smoking: DecisionMaking (b 5 20.11, P , 0.01) and Personal Efficacy (b 5 20.14, P , 0.001) predicted lower perceptions of the Social Benefits of Smoking while Baseline Smoking predicted higher perceptions of the Social Benefits of Smoking (b 5 0.09, P , 0.01). Furthermore, these three predictors accounted for 5% of the variance in the perceptions of social benefits in the final model. The perceived Social Benefits of Smoking predicted 2-Year Follow-Up Smoking (b 5 0.14, P , 0.001). Finally, as

TABLE 1 Correlations among Latent Factors from Confirmatory Factor Analysis

1. 2. 3. 4. 5.

Latent factor

1

2

3

4

5

Baseline Smoking Decision-Making Personal Self-Efficacy Social Benefits 2-Year Smoking

— 20.10* 20.05 0.11* 0.39**

— — 0.29** 20.16** 20.13**

— — — 20.18** 20.10*

— — — — .18**

— — — — —

* P , 0.01. ** P , 0.001.

112


FIG. 2. Structural equation model. Large circles represent latent factors and small circles with numbers reflect residual variances; *P , 0.01; **P , 0.001.

expected, Baseline Smoking predicted 2-Year FollowUp Smoking (b 5 0.37, P , 0.001). Overall, the predictors explained 17% of the variance in 2-year smoking. In terms of goodness-of-fit indices, there was an excellent fit of the model to the data, x 2/df 5 2.2; SRMR 5 0.043; NFI 5 0.96; NNFI 5 0.97; and CFI 5 0.98. In summary, the findings shown in Fig. 2 indicate that each of the three exogenous latent factors significantly predicted the perceived Social Benefits of Smoking, which in turn predicted 2-Year Follow-Up Smoking. The model was not modified further in an attempt to improve its fit since the fit indices are in the excellent range and because conducting specification searches to improve the fit can capitalize on chance characteristics of the data and lead to an unstable model with limited generalizability [46]. DISCUSSION

In the present study, findings indicated that deficiencies in competence (poor decision-making skills and low personal efficacy) measured at the baseline assessment had an indirect relationship with smoking 2 years later through the perceived social benefits of smoking measured at the 1-year follow-up assessment. Neither decision-making skills nor personal efficacy had a direct relationship with subsequent smoking; both decisionmaking and personal efficacy affected smoking indirectly through the perceived social benefits of smoking. According to these findings, the theory underlying the

competence enhancement approach to smoking prevention received support using longitudinal etiologic data. Based on social learning theory [13] and problem behavior theory [14], the competence enhancement approach views smoking as socially learned, purposeful, and functional behavior resulting from the interplay between social and personal factors. This approach postulates that adolescents who lack basic skills to cope with life are more receptive to internal and external pressures to smoke. The findings suggest that adolescents who are less competent due to poorer decision-making skills and low personal efficacy are more susceptible to prosmoking messages, particularly those messages promoting the social benefits of smoking. This in turn appears to lead to positive beliefs about the perceived social benefits of smoking (i.e., being cool or popular), which motivates adolescents to smoke. Adolescents who have not been able to observe and practice sound decision-making skills may be at a distinct disadvantage when confronted with social influences to smoke from family, friends, and the media. Competence enhancement approaches to smoking prevention include training in general problem-solving and decision-making skills to reduce adolescents’ susceptibility to such external forces to smoke [11,12]. Prevention studies conducted with inner-city Hispanic youth [18,19] and African–American youth [20] have demonstrated that a competence enhancement approach to smoking prevention can decrease smoking prevalence relative to a control group in the seventh grade. Thus


the current study suggests that enhancing decisionmaking skills is a critical component in preventing smoking among inner-city minority adolescents and this effect works in part by decreasing the belief in perceived social benefits of smoking. Adolescents with good decision-making skills were better able to gather information, decide on the best alternative, and consider the consequences. They generated counter-arguments to pro-smoking messages and rejected beliefs in the perceived social benefits of smoking. Furthermore, competence enhancement approaches to smoking prevention are designed to increase participants’ feelings of personal efficacy through training in a broad array of personal and social skills. As shown in the model tested in the current study, increasing personal efficacy is also a critical factor preventing subsequent smoking and this effect works in part by decreasing the perceived benefits of smoking. Looking at this model from the perspective of adolescents who were deficient in their level of personal competence (poor decision-making skills, low personal efficacy), it was found that these adolescents were more likely to believe that smoking offers social benefits such as looking cool, having more friends, and being better liked. For these adolescents, smoking cigarettes has served a functional value. Consequently, adolescents holding these beliefs at the 1-year follow-up were more likely to engage in smoking behavior by the 2-year follow-up assessment. Finally, the model points to the importance of the beliefs in perceived social benefits of smoking as a predictor of subsequent smoking. Prevention efforts should clearly focus on any means of affecting these perceptions. Although items reflecting acceptability and prevalence of smoking were not included, it could be expected that peer norms would affect perceived social benefits. Therefore, these other domains need to be addressed as well particularly since past prevention research highlights the importance of these factors [19,47,48]. Several limitations of this study should be considered. First, due to the limited numbers of variables that could be tested in the hypothesized model, it is acknowledged that other key predictors exist including smoking-specific versions of the decision-making and personal efficacy measures. Based on the limited number of predictors tested, the model did not explain a high proportion of the variance in the perceived social benefits or smoking or the outcome measure. If other general competence measures had been included, they may have contributed to a greater proportion of the variance. Decision-making and personal efficacy are by no means the only constructs of general competence, merely the only ones available here. In addition, smoking-specific versions of decision-making and personal efficacy may have increased the amount of variance explained as well because of their presumed stronger

113

relationship with the perceived benefits of smoking. Future research should explore other models with a complete array of general competence measures and smoking-specific competence measures. Yet, the model is parsimonious and consistent with previous research concerning adolescent smoking behavior [11,12] and reasonable within the context of the relevant theory. Second, since this sample was school-based, caution is warranted in generalizing these findings to non-school samples. Also because this study focused on students residing in New York City, it is conceivable that the results might be different from other urban areas. Yet additional research conducted in other inner-city regions most likely would confirm the generalizability of the theoretical foundation specified here. Attrition analyses indicated the loss of high-end smokers, which may have attenuated the estimation of variable relationships. However, this study was conducted in the middle school period when dropout rates remain low and absentee data were minimized by pursuing absentees on two return data collections. Despite these limitations, the longitudinal model of smoking among inner-city adolescents has several implications for smoking prevention. To be effective, smoking prevention approaches should incorporate the teaching of decision-making skills. These skills proved to be an important foundation for adolescents in middle school. Similarly, smoking prevention approaches should contain components designed to increase personal efficacy. To the extent that competence approaches create alternative means of achieving popularity and self-esteem through training in social, communication, and assertive skills, this should undermine the adoption of beliefs in the social benefits of smoking that then lead to smoking. REFERENCES 1. American Cancer Society. Cancer facts and figures. Atlanta: American Cancer Society, 1999. 2. Johnston LD, O’Malley PM, Bachman JG. National survey results on drug use from monitoring the future study, 1975–1997, Vol. I, Secondary school students. Rockville, MD: National Institute of Drug Abuse, 1998. 3. Centers for Disease Control and Prevention. Tobacco use among high school students. MMWR 1998;47:229–33. 4. Centers for Disease Control and Prevention. Projected smokingrelated deaths among youth—United States. MMWR 1996;45: 971–4. 5. Chassin L, Presson CC, Sherman SJ, Edwards D. The natural history of cigarette smoking: Predicting young-adult outcomes from adolescent smoking patterns. Health Psychol 1990;9: 701–16. 6. Chassin L, Presson CC, Rose J, Sherman SJ. The natural history of cigarette smoking from adolescence to adulthood: Demographic predictors of continuity and change. Health Psychol 1996;15: 478–84. 7. Chen VW. Smoking and the health gap in minorities. Annu Epidemiol 1993;3:159–64.

114


8. Moore DJ, Williams J, Qualls WJ. Target marketing of tobacco and alcohol-related product to ethnic minority groups in the United States. Ethnicity Dis 1996;6:83–98. 9. U.S. Department of Health and Human Services. Tobacco use among U. S. racial/ethnic minority groups—African Americans, American Indians and Alaska Natives, Asian Americans and Pacific Islanders, and Hispanics: A report of the Surgeon General. Atlanta, GA: U.S. Department of Health and Human Services, CDC, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, 1998. 10. Oetting ER, Beauvais F. Adolescent drug use: Findings of national and local surveys. J Consult Clin Psychol 1990;58:385–94. 11. Botvin GJ, Epstein JA. Preventing cigarette smoking among children and adolescents. In: Seidman DF, Covey L, editors. Helping the hard-core smokers: A clinician’s guide. Mahwah, NJ: Erlbaum, 1999. 12. Botvin GJ, Epstein JA, Botvin EM. Adolescent cigarette smoking: Prevalence, causes and intervention approaches. Adolescent Med: State of the Art Rev 1998;9:299–313. 13. Bandura A. Social learning theory. Englewood Cliffs, NJ: Prentice-Hall, 1977. 14. Jessor R, Jessor SL. Problem behavior and psychosocial development: A longitudinal study of youth. New York: Academic Press, 1977. 15. Botvin GJ, Eng A. The efficacy of multicomponent approach to the prevention of cigarette smoking. Prev Med 1982;11:199–211. 16. Botvin J, Renick NL, Baker E. The effects of scheduling format and booster sessions on a broad-spectrum psychosocial smoking prevention program. J Behav Med 1983;6:359–79. 17. Botvin GJ, Baker E, Dusenbury L, Botvin EM, Diaz T. Longterm follow-up results of a randomized drug abuse prevention trial in a white middle-class population. J Am Med Assoc 1995;1106–12. 18. Botvin GJ, Dusenbury L, Baker E, James-Ortiz S, Kerner J. Skills training approach to smoking prevention among Hispanic youth. J Behav Med 1989;12:279–96. 19. Botvin GJ, Dusenbury L, Baker E, James-Ortiz S, Botvin EM, Kerner J. Smoking prevention among urban minority youth: Assessing effects on outcome and mediating variables. Health Psychol 1992;11:290–99. 20. Botvin GJ, Batson H, Witts-Vitale S, Bess V, Baker E, Dusenbury L. A psychosocial approach to smoking prevention for urban black youth. Pub Health Rep 1989;104:573–82. 21. Botvin GJ. Smoking prevention among New York Hispanic youth: Results of a four-year evaluation study [Final report to the National Cancer Institute (NIH), Grant R18 CA 39280]. New York: Cornell Univ Med College, 1994. 22. Wills TA. Stress, coping, and tobacco and alcohol use in early adolescence. In: Shiffman S, Wills TA, editors. Coping and substance use. San Diego: Academic Press, 1985:67–92. 23. Ellickson PL, Hays RD. Beliefs about resistance self-efficacy and drug prevalence: Do they really affect drug use? Intl J Addict 1990;25:1353–78. 24. Lawrance L, Robinson L. Self-efficacy as a predictor of smoking behavior in young adolescents. Addict Behav 1986;11:367–82. 25. Botvin GJ, Goldberg CJ, Baker E, Dusenbury L, Botvin EM. Correlates and predictors of smoking among black adolescents. Addict Behav 1992;17:97–103. 26. Botvin GJ, Epstein JA, Schinke SP, Diaz T. Predictors of smoking among inner-city minority youth. J Dev Behav Pediatr 1994; 15:67–73. 27. Glynn TJ, Anderson DM, Schwarz L. Tobacco-use reduction among high-risk youth: Recommendations of a National Cancer Institute advisory panel. Prev Med 1991;20:279–91.

28. Botvin GJ, Schinke SP, Epstein JA, Diaz T, Botvin EM. The effectiveness of culturally-focused and generic skills training approaches to alcohol and drug abuse prevention among minority youth: Two-year follow-up results. Psychol Addict Behav 1995; 9:183–94. 29. Pentz MA, MacKinnon, Flay BR, Hansen WB, Johnson CA, Dwyer JG. Primary prevention of chronic diseases in adolescence: Effects of the midwestern prevention project on tobacco use. Am J Epidemiol 1989;130:713–24. 30. Snow DL, Tebes JK. Panel attrition and external validity in adolescent substance use research. J Consult Clin Psychol 1992;60:804–7. 31. Bauman KE, Koch GG, Bryan ES. Validity of self-reports of adolescent cigarette smoking. Int J Addict 1988;17:1131–6. 32. Evans RI, Hansen WB, Mittlemark MD. Increasing the validity of self-reports of smoking behavior in children. J Appl Psychol 1977;62:521–3. 33. Murray DM, O’Connell DM, Schmid LA, Perry CL. The validity of smoking self-reports by adolescents: A reexamination of the bogus pipeline procedure. Addict Behav 1987;12:7–15. 34. Botvin GJ, Epstein JA, Baker E, Diaz T, Ifill-Williams M. Schoolbased drug abuse prevention with inner-city minority youth. J Child Adolescent Substance Abuse 1997;6:5–19. 35. Coogan PF, Adams M, Geller AC, Brooks D, Miller DR, Lew RA, Koh HK. Factors associated with smoking among children and adolescents in Connecticut. Am J Prev Med 1998;15:17–24. 36. Presson CC et al. Predictors of adolescents’ intentions to smoke: Age, sex, race, and regional differences. Int J Addict 1984;19: 503–19. 37. Stern RA, Prochaska JO, Velicer WF, Elder JP. Stages of adolescent cigarette smoking acquisition: Measurement and sample profiles. Addict Behav 1987;12:319–29. 38. Sussman S, Brannon BR, Dent CW, Hansen WB, et al. Relations of coping effort, coping strategies, perceived stress, and cigarette smoking among adolescents. Int J Addict 1993;28:599–612. 39. Conrad KM, Flay BR, Hill D. Why children start smoking cigarettes: Predictors of onset. Br J Addict 1992;87:1711–24. 40. Wills TA. Stress and coping in early adolescence: Relationships to substance use in urban school samples. Health Psychol 1986;5:503–29. 41. Paulhus D. Sphere-specific measures of perceived control. J Personal Social Psychol 1983;44:1253–65. 42. U.S. Public Health Service. Teenager’s self-test: Cigarette smoking. Washington, DC: Centers for Disease Control, 1974. [DHEW Publication No. CDC 74-8723] 43. Bentler PM. EQS structural equations program manual. Encino, CA: Multivariate Software, Inc., 1995. 44. Bollen KA. Structural equations with latent variables. New York: Wiley, 1989. 45. Raykov T, Tomer A, Nesselroade JR. Reporting structural equation modeling results in psychology and aging: Some proposed guidelines. Psychol Aging 1991;6:499–503. 46. MacCallum RC, Roznowski M, Necowitz LB. Model modification in covariance structure analysis: The problem of capitalizing on chance. Psychol Bull 1992;111:409–504. 47. Donaldson SI, Graham JW, Hansen WB. Testing the generalizability of intervening mechanism theories: Understanding the effects of adolescent drug use prevention interventions. J Behav Med 1994;17:195–216. 48. Hansen W, Graham J. Preventing alcohol, marijuana, and cigarette use among adolescents: Peer pressure resistence training versus establishing conservative norms. Prev Med 1991;20: 414–430.