Psychosomatic Medicine
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Rodriguez, D.
Right arrow Articles by Audrain-McGovern, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Rodriguez, D.
Right arrow Articles by Audrain-McGovern, J.
Psychosomatic Medicine 69:106-113 (2007)
© 2007 American Psychosomatic Society

ORIGINAL ARTICLES

Beliefs About the Risks of Smoking Mediate the Relationship Between Exposure to Smoking and Smoking

Daniel Rodriguez, PhD, Daniel Romer, PhD and Janet Audrain-McGovern, PhD

From the Department of Psychiatry (D.Rodriguez, J.A.-M.), University of Pennsylvania; Annenberg Public Policy Center (D.Romer), University of Pennsylvania, Philadelphia, Pennsylvania.

Address correspondence and reprint requests to Daniel Rodriguez, PhD, Department of Psychiatry, University of Pennsylvania, 3535 Market Street, Suite 4100, Philadelphia, PA 19104. E-mail: drodrig2{at}mail.med.upenn.edu.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 
Objective: Peer and family smoking are key predictors of adolescent smoking. Yet, it is unclear whether the effect of these variables is direct or indirect through the effects of mediating mechanisms. One possible mechanism is smoking risk beliefs. We hypothesized an indirect effect such that exposure to peer and family smoking may affect adolescents’ smoking through two sets of risk beliefs; beliefs about the personal harm of smoking, and beliefs about the general immediate harm of smoking, and these beliefs may in turn affect smoking.

Methods: Our sample was 963 participants taking part in a longitudinal study of the biobehavioral determinants of smoking. We measured exposure to peer and household smoking in grade 10, smoking risk beliefs in grade 11, and modeled the effects of these variables prospectively on smoking one year post high school graduation in a Structural Equation Model (SEM).

Results: Beliefs about the personal harm and general immediate harm of smoking had significant and negative direct effects on smoking one year post high school. However, controlling for 10th grade smoking, only personal harm beliefs mediated the relationship between household smoking exposure and smoking behavior. Specifically, personal harm beliefs mediated the effect of having a household member who smokes on smoking one year post high school graduation.

Conclusions: The findings are consistent with the hypothesized mediation model and suggest that exposure to household smoking may affect adolescent smoking through its effects on beliefs about the personal harm of smoking, beyond the effects of previous smoking.

Key Words: smoking • risk beliefs • exposure to smokers

Abbreviations: GPA = grade point average; SEM = structural equation modeling; CFI = Comparative Fit Index; RMSEA = Root Mean Square Error of Approximation; WRMR = weighted Root Mean Residual; OR = odds ratio; CI = confidence interval.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
METHODS
 RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 
Adolescence is a critical period for smoking. Of those adults who have ever smoked regularly, the majority began smoking during adolescence and progressed to a regular habit by age 18 (1,2). It is well-established that exposure to peer and family smoking significantly increases the likelihood of experimentation and progression to a regular habit (1,3–10). Although exposure to others smoking increases the likelihood adolescents will smoke, it is unclear why exposure affects smoking. Usually these effects are interpreted as a modeling, normative, or instigation influence (e.g., 10,11,12), all of which suggest mediating mechanisms. We are proposing a specific mediation model by which the effect of exposure to others’ smoking may influence how adolescents conceive of the risks of smoking, and these beliefs in turn can affect smoking behavior. It is noted, however, that we assessed the model with observational data, and the use of terms like influence is only speculative with respect to the guiding mediation model.

Research suggests that adolescent smokers understand that smoking entails risk, yet their perceptions are flawed. For instance, even when adolescents acknowledge the mortality risks of smoking and years of life lost to smoking, they tend to underestimate the personal risk in comparison to average smokers (13–15). They also tend to inaccurately believe smoking is less risky than other risk behaviors (e.g., drunk driving), and underestimate the short-term impact of smoking (e.g., smoking the very next cigarette). Adolescent smokers also tend to be overly optimistic about their ability to quit before smoking affects their health (15,16). Therefore, adolescents may initiate and progress to regular smoking because they underestimate the risks and overestimate the ease of quitting before smoking results in disease.

To our knowledge, no study has prospectively assessed whether smoking risk beliefs mediate the relationship between exposure to smoking in mid-adolescence and smoking in late adolescence. However, one cross-sectional study found a negative relation between peer group approval of smoking and risk perception (15). Another prospective study found significant indirect effects of peer substance use and parental communication about substance use, on substance use, in a sample of predominantly African-American adolescents (17). Specifically, peer substance use, including tobacco, decreased perceived vulnerability to the negative effects of substance use (eg, lung cancer), which in turn increased substance use one year later. Parental communication about substance use had the opposite effect. Although this study (17) did not assess smoking risk beliefs per se, the results suggest the importance of exposure as an influence on perceived vulnerability and substance use.

The purpose of this study was to assess the impact of adolescent beliefs about the personal harm and general immediate harm of smoking on smoking behavior from mid to late adolescence, and whether these beliefs mediated the relationship between exposure to peer and household smoking and smoking behavior. It is proposed that greater family and peer exposure in 10th grade (age 15–16) will be associated with a lower level of beliefs about the personal harm and the general immediate harm of smoking, measured in 11th grade (age 16–17), which will in turn be associated with an increase in the likelihood of regular smoking (smoking within the past 30 days) one year post high school (age 19–20).


   METHODS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 
Participants were 963 adolescents (53% female, 63% white) completing the first five years of an eight-year longitudinal study of the biobehavioral predictors of smoking adoption, spanning ages 14 to 22. The cohort was formed in the 9th grade and is being followed through four years post high school. Participants attended one of five public high schools in northern Virginia when data collection initiated. Students were ineligible to participate if they had special classroom placement (e.g., English as a second language). Of the 2120 eligible adolescents, 72% of parents/guardians approached responded regarding permission for their adolescent to participate. Of those responding, 75% provided written consent, yielding an overall consent rate of 54% (1151). Assessment of differences between parents providing and declining consent revealed that among white parents, those with greater than a high school education were more likely to consent than parents with a high school education or less (18). There were no other significant differences between parents who did and did not consent.

Study participation also required student assent. Of the 1151 adolescents receiving parental consent, 1136 assented. We surveyed 1123 adolescents at baseline, with 13 absent. There were four follow-up assessments in high school, culminating in the 12th grade, during spring 2003, and one completed post high school. Of the 1068 participants with complete data on 30-day smoking in 10th grade, 817 had complete data for smoking 10th grade and one year post high school, and 935 participants had complete data on the five risk variables assessed in 11th grade. However, 739 participants had complete data on smoking and the perceived risks of smoking. Therefore, 69% of participants with complete data on smoking in grade 10 had complete data on the dependent variables in the multivariate model (i.e., smoking and perceived risks of smoking). Including the covariates, 706 participants (66%) had complete data on all model variables.

To account for missing data, multivariate modeling used all available data. Mplus software provides options for missing data modeling assuming data are either missing completely at random or missing at random (19). Thus, the final sample size for the full multivariate model with covariates, using all available data, was n = 963, excluding those participants missing data on the exogenous variables. University Institutional Review Board approval of the study protocol was obtained.

Procedure
High school data collection took place on-site, during a class common to all students (e.g., health, science). A research team member distributed the survey. The completed survey only contained an identification number. A research team member read aloud the instructions, emphasizing confidentiality, and encouraged questions if survey items were unclear. Data were collected via telephone post high school. Surveys took approximately 30 minutes to complete.


Race and Gender
Race and gender were self-reported. We dichotomized race because of the small number of participants within each race category other than white.

Smoking
We assessed smoking with 13 standard epidemiological questions, such as "Have you tried or experimented with cigarette smoking, even a few puffs?" and "When was the last time you smoked a cigarette (20)?" We used responses to these items to generate a five level ordered categorical variable representing increasing levels of past 30-day smoking from not smoking to smoking 11 or more cigarettes daily. The ordered categories are 0, did not smoke in the past month; 1, smoked "1 month ago or less;" 2, smokes " ... at least once a week;" 3, smokes daily but no more than 10 cigarettes per day; 4, smokes 11 or more cigarettes per day.

Risk of Smoking
We used five four-option Likert-type items requesting information on adolescents’ perceptions of immediate harm of smoking to a young person in general ("... someone who starts smoking a pack of cigarettes a day at age 16"), and the personal risk of smoking (e.g., "... is smoking very risky to your health ...") (15). Each item included a fifth response option, "Don’t know." Items answered "Don’t know" were coded missing. We treated the five risk items as ordered categorical variables because analysis indicated univariate skewness and/or kurtosis for each item. The five risk items were measured in the 11th grade only, and were modeled as indicators for two smoking risk beliefs latent variables (factors) consistent with the results of an earlier exploratory factor analysis (15); and three items for personal harm and two items for general immediate harm.

Exposure to Smoking
Items to assess exposure to smoking were the binary variable household smoking (1 = someone living in the household smokes, 0 = no one living in the household smokes), and the continuous variable number of friends who smoke (does your best friend smoke, and how many of your other four best male and other four best female friends smoke, ranging from 0–9 friends). Research has found that peers and household smoking are associated with smoking in adolescence (10, 21).

Other Covariates
We controlled for the available demographic variables: gender (2 = female, 1 = male) and race (1 = nonwhite, 0 = white). We also controlled for several environmental variables associated with adolescents not smoking, including past 12-month team sport participation (0 = no teams, 1 = 1 team, 2 = 2 teams, 3 = 3 or more teams), Grade Point Average (GPA), and school clubs and activities (4,22–25). We assessed all covariates in 10th grade (baseline).

Analysis Plan
Data analysis was conducted with structural equation modeling (SEM) using Mplus software, version 4.1 (19). SEM is theory-based representations of relations among a set of variables (26). In this study, all dependent variables (i.e., 30-day smoking, one year post high school and the five risk beliefs items) were ordered categorical, therefore we estimated model parameters with a Weighted Least Squares estimation technique (WLSMV) in which the diagonal weight matrix uses robust standard errors, and the chi-square test statistic is mean and variance adjusted (19). Model fit was evaluated with model chi-square, Comparative Fit Index (CFI), Root Mean Square Error of Approximation (RMSEA), and weighted Root Mean Residual (WRMR). Suggested criteria for model fit are nonsignificant model chi-square, CFI above 0.95, RMSEA below 0.05 to 0.08, and a WRMR value below 0.9 (27).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
DISCUSSION
 NOTES
 REFERENCES
 


Descriptive Statistics
Females comprised 53% of the sample. Regarding race, white participants comprised 63% of the sample, whereas 8% were African American, 12% were Hispanic, 11% were Asian/Pacific Islander, and 6% reported race as "other." Table 1 presents the bivariate correlations for all variables in the model, along with their means and standard deviations. Regarding smoking, there is a noticeable increase in the proportion of adolescents smoking from 10th grade to one year post high school. For instance, the proportion of adolescents smoking daily (encompassing the two highest ordered categories) increased from 5% at 10th grade to 11% in the year following high school. By contrast, the proportion of adolescents not smoking in the past 30-days decreased from 87% at 10th grade to 75% in the year following high school. Regarding the relationship between smoking and risk beliefs, examination of Table 1 shows significant and negative polychoric correlations between smoking in 10th grade and one year post high school, and the five observed risk variables, respectively.1


View this table:
[in this window]
[in a new window]

  		TABLE 1. Bivariate Polychoric Correlations (n = 963)

 

Table 2 presents distributions for the smoking risk variables. Although adolescents generally acknowledged that smoking is very risky to one’s health, and overwhelmingly endorsed the risk of smoking every day, there was less agreement on the effects of casual smoking (smoking "... say at parties or with friends"). Further, although they generally acknowledged the immediate harm of smoking, a substantial proportion of adolescents expressed some degree of disagreement that smoking is immediately harmful to a person’s health.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Frequency Distribution for the Smoking Risk Variables (N = 963)

 

Regarding exposure to smoking in the 10th grade, 25% of participants lived with a household member who smokes, and the average number of other friends smoking was 1.83 (SD = 2.27), with a range of 9. On the other hand, the average number of 10th grade extracurricular activities (e.g., drama club, yearbook) was 3.40 (SD = 1.51), range 8. Fifty one percent of 10th grade participants played at least one team sport in the past 12 months. Finally, the average 10th grade GPA was 3.17 (SD = 0.57), range 3.

Multivariate Analysis: Measurement Model

The Perceived Risks of Smoking
We began the multivariate analysis assessing the factor structure for the five smoking risk items. Previous research supported a two factor structure equally dividing four indicators of beliefs about the risks of smoking, one factor representing two personal harm indicator variables and a second factor representing two indicator variables of the general immediate harm of smoking (15). The current study included a fifth item "How risky do you think it would be for your health to smoke only once in a while, say at parties or with friends?" We added this new item to the factor representing personal harm.

Instead of two factors, it was also possible that the five risk beliefs items indicated a single global factor representing conceptually general risk beliefs about smoking. To assess this possibility, we compared for fit to the data a single factor with the two factor model. Because the single-factor model is nested within the two-factor model (i.e., it can be derived by fixing the inter-factor correlation in the two factor model to 1.0), the comparison was made with a chi-square difference test (26). Because factors are latent variables, and factors do not have scales, we constrained the factor loading to 1 for the first item in each set (i.e., "... is smoking very risky to your health?" for personal harm, and "... no risk at all ... for the very first few years?" for general immediate harm) (26). Anchoring the two factors provided each an interpretable scale. The single general risk beliefs factor model did not fit the data well, {chi}2 (4, 1041) = 155.92, p < .0001, CFI = 0.92, RMSEA = 0.19, WRMR = 2.33. The second model with one factor for personal harm and a second factor for general immediate harm, however, fit substantially better, {chi}2 (3, 1041) = 1.53, p = .67, CFI = 1.00, RMSEA = 0.00, WRMR = 0.21. The chi-square difference test, {chi}2 (1, 1041) = 154.38, p < .0001, favored the two factor model. Therefore, the data supported the two-factor risk beliefs model for the five observed risk beliefs items in this analysis, over a single-factor model representing general risk beliefs. Conceptually this may suggest that there is not a unidimensional smoking risk beliefs construct as measured by these five risk beliefs items, but rather a construct that identifies smoking risks directly relevant to the individual and a construct that identifies general smoking risks that apply to others.

Multivariate Analysis: Full Model
Figure 1 presents the results of the final multivariate model with standardized path coefficients for significant model effects only. It is noted that the full model also included paths from all covariates to both risk factors and smoking one year post high school. However, to reduce visual clutter only significant paths are shown. This model fit the data well, {chi}2 (25, 963) = 28.17, p = .30, CFI = 1.00, RMSEA = 0.011, WRMR = 0.51. For comparison, we also estimated a model with the single general risk beliefs factor representing the relationship among the five observed risk variables. As expected, this model did not fit as well, {chi}2 (31, 963) = 143.42, p < .0001, CFI = 0.91, RMSEA = 0.06, WRMR = 1.28, supporting the two-factor structure for smoking risk beliefs. Table 3 presents the nonstandardized path coefficients, standard errors, and test statistics for the regressions within the final multivariate model with two risk beliefs factors. The following section reviews results for the primary outcome 30-day smoking and mediator latent variables personal harm and general immediate harm.


Figure 117
View larger version (11K):
[in this window]
[in a new window]

  		Figure 1. Structural equation model with standardized regression coefficients for significant model effects. *p < .05. **p < .0001.

 

View this table:
[in this window]
[in a new window]

  		TABLE 3. Non-standardized Path Coefficients, Standard Errors, and Test Statistics for the Regression Analyses in the Model (N = 963)

 

Direct Effects

Smoking Behavior One Year Post High School
There were three significant direct effects on 30-day smoking, one year post high school. We present these results in terms of odds ratios (OR), along with the 95% confidence intervals (CI), to facilitate interpretation. We calculated OR and 95% CI exponentiating the nonstandardized log odds Beta path coefficients (Table 3), resulting in an estimate of the odds of change in the dependent variable for a unit increase in the independent variable (28). As expected, 10th grade smoking increased the odds of greater smoking (from once in the past 30 days to smoking >10 cigarettes daily within the past 30 days) one year post graduation 59% (OR = 1.59, CI = 1.36–1.85). Regarding risk beliefs, personal harm beliefs had a significant direct effect on smoking, such that for each unit decrease in personal harm from very risky to not at all risky, there was an associated increase of 34% (OR = 1.34, CI = 1.12–1.59) in the odds of greater smoking one year post graduation. Beliefs about the general immediate harm of smoking also had a significant effect on smoking one year post high school. Specifically, for each unit decrease in belief in the immediate harm of smoking from strongly agree to strongly disagree, there was an associated increase of 27% (OR = 1.27, CI = 1.02–1.60) in the odds of greater smoking one year post high school. These results suggest that beliefs that smoking is not personally harmful, and that smoking is not immediately harmful generally, both measured in the 11th grade, were associated with an increase in the odds of a higher level of smoking two years later, beyond the effects of previous smoking. Collectively, the predictors in the model accounted for 35% of the variation in smoking one year post high school. There were no other significant direct effects on 30-day smoking one year post high school.

Personal Harm
There were four significant direct effects on the factor representing beliefs about the personal harm of smoking. Being female was associated with a greater belief (ß = 0.23, z = 3.07, p = .0021) about the personal harm of smoking. As expected, exposure to smoking had a significant negative effect on personal harm beliefs. Having friends who smoke was associated with a lower level of personal harm beliefs (ß = –0.04, z = –2.44, p = .0147). Moreover, having at least one household member who smokes was associated with a lower level of beliefs about the personal harm of smoking (ß = –0.22, z = –2.77, p = .0056). Finally, 10th grade smoking was associated with lower beliefs in the personal harm of smoking one year later (ß = –0.20, z = 4.21, p < .0001).

General Immediate Harm
Only gender had a significant effect on general beliefs about the immediate harm of smoking. Females had higher beliefs about the general immediate harm of smoking (ß = 0.24, z = 3.30, p = .001). There were no other significant direct effects on the factor representing general immediate harm.

Indirect Effects
We tested for indirect effects of variables representing exposure to smoking on smoking one year post high school, through beliefs about the personal harm and general immediate harm of smoking. We also tested the indirect effects of 10th grade smoking through these same pathways. The indirect effects were calculated with Delta method computed standard errors (19).


Tenth Grade Smoking
Tenth grade smoking had a significant total effect (direct and indirect paths) on smoking one year post high school (ßtotal = 0.54, z = 7.101, p < .0001). Only one indirect effect was significant, though, the indirect effect of 10th grade smoking through personal harm (ßindirect = 0.07, z = 3.103, p = 0.0019). Moreover, the 95% confidence interval did not include zero (0.03, 0.11), suggesting that beliefs about the personal harm of smoking resulted in part from existing smoking behavior, and that they in turn affected smoking one year post high school. However, the direct effect remained significant (ßdirect = 0.45, z = 5.721 p < .0001), suggesting only partial mediation; although smoking may affect one’s beliefs about its personal harm, these beliefs themselves may have a limited role in predicting 30-day smoking in late adolescence beyond the direct effects of tenth grade smoking.

Tenth Grade Household Smoking
Household smoking had a significant indirect effect on smoking one year post high school through smoking risk beliefs (ßtotal-indirect = 0.06, z = 1.988, p = .0468). However, the indirect effect was only significant through beliefs about the personal harm of smoking (ßindirect = 0.05, z = 2.298, p = .0216). Moreover, the 95% confidence interval did not include zero (0.01, 0.10), suggesting that beliefs about the personal harm of smoking is a potential mediator in the relationship between exposure to household smoking and smoking behavior; exposure to household members whom smoke was associated with a greater belief that smoking is not personally harmful. Higher beliefs regarding personal harm, in turn were associated with an increase in the likelihood of smoking within the past 30 days, one year post high school. The direct effect of household smoking in 10th grade to smoking one year post high school was not significant (p = .15), suggesting significant mediation.

Tenth Grade Peer Smoking
Regarding the effects of 10th grade peer smoking on smoking one year post high school, the total effect (ßtotal = 0.05, z = 2.106, p = .0352) was significant, as was the total indirect effect (ßtotal-indirect = 0.02, z = 2.383, p = .0172) through both personal harm and general immediate harm. However, neither specific indirect effect was significant. Only the indirect effect through personal harm approached significance (ßindirect = 0.01, z = 1.958, p = 0.05), with the 95% confidence interval including zero (0.00, 0.02).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
NOTES
 REFERENCES
 
The aim of this study was to assess the association of adolescents’ beliefs about the personal harm and general immediate harm of smoking, measured in the 11th grade, with smoking behavior one year post high school (late adolescence), and whether these beliefs could explain the relationship between exposure to peer and or household smoking in 10th grade and smoking behavior in late adolescence. The findings suggest personal harm but not the general immediate harm of smoking is one possible mechanism by which exposure to household smoking affects smoking behavior in late adolescence. General immediate harm did not account for this relationship, and neither smoking risk variable explained how having peers who smoke affects smoking behavior from mid to late adolescence.

The finding that personal harm mediated the relationship between household smoking exposure and smoking in late adolescence, suggests the possibility that household smoking may affect smoking behavior through its effect on specific smoking beliefs. Household members have a unique opportunity to influence the development of youth health behaviors by modeling healthy behaviors and discussing health risks, such as smoking. Perhaps household members who smoke are less likely to discuss the risks of smoking, which may lead youth to develop inaccurate beliefs about the personal risks of smoking. In support of this possibility, one prospective study found a significant indirect effect from parental communication about substance use on future substance use, through its effects on the perceived vulnerability to the risks of substance use, and willingness to use substances (17). Thus, youth whose parents discussed substance use were less likely to use substances as they perceived it as having negative health consequences. Alternatively, household members, including parents who smoke may communicate with their teens that smoking is a health risk, but their communications are ineffective as they are discordant with their behavior (29). Although only speculative at this point, such findings suggest the need for more research on the effects of household smoking on adolescent risk beliefs.

Contrary to expectation, the indirect effect from peer smoking to personal harm and from personal harm to smoking one-year post high school only approached significance. This finding suggests that the beliefs regarding the personal harms associated with smoking may not be a strong mechanism by which peer smoking affects smoking behavior from mid to late adolescence. Other variables may better account for the relationship between peer smoking and smoking from mid to late adolescence (30,31).

Although personal harm was a significant mediator, general immediate harm did not mediate the relationship between smoking exposure in mid-adolescence and smoking behavior in late adolescence. This may be due to the general nature of the general immediate harm items. Whereas, the personal harm items reflect the beliefs that smoking is risky to one’s self (e.g., "In your opinion, is smoking very risky for your health?"), items representing general immediate harm were impersonal, assessing beliefs about the effects of smoking on others (e.g., "There is usually no risk to the person at all for the very first few years?"). Although, this factor still had a significant direct and negative effect on smoking, beliefs about the effects of smoking on others did not account for how exposure to other smokers affects smoking.

Although it appears that household smoking in mid-adolescence indirectly affects smoking behavior in late adolescence through personal harm beliefs, this is only one mechanism and other potential mediating mechanisms and moderating variables should be explored within a more comprehensive model. For instance, one study found that receptivity to tobacco advertising mediated the relationship between exposure to smoking, and adolescent smoking progression (30). Potential moderating variables include parental education about smoking and parenting style (6–8,24,32,33). For instance, one study found that authoritative parenting was associated with believing parental communication about smoking, and a reduction in the odds of an adolescent currently smoking (24). Household smoking restrictions also warrant investigation as household smoking bans appear to affect adolescents’ perceptions of the prevalence of adults smoking, and adult approval of smoking (32). Thus, a more comprehensive analysis of how parenting and household factors affect smoking initiation and progression in childhood and mid to late adolescence is warranted (34).

There are several limitations of this study. First, beliefs about the risks of smoking were measured once only. It would be interesting to know whether risk beliefs change with time, what factors affect initial beliefs and rate of change, and how changes in risk beliefs influence changes in smoking behavior. Second, although we controlled for several important covariates, we did not control for other variables related to adolescent smoking, such as receptivity to tobacco advertising, genetic factors, physical activity, peer substance use (other than smoking), parenting style, parental monitoring, and discussion of smoking. Further, although we controlled for gender and race, we were not able to control for other demographic variables related to adolescent smoking, such as socioeconomic status (35). Third, Although 75% of those parents who responded provided consent and the differences between those who provided consent and those who declined were relatively small and few (18), caution is warranted in generalizing the results of this study, especially in light of the study’s consent rate (54%). However, the sample was nationally and locally representative on basic demographic characteristics (36,37). Finally, although the results of this study suggest mediation, it should be noted that the model accounted for 35% of the variation in smoking one year post high school, and the mediated path accounted for a little less than 5% of the variance in late adolescent smoking. Moreover, with only household smoking and personal harm in the model, 19% of the variance in late adolescent smoking was explained. This suggests that other mechanisms may play a more important role in accounting for the relationship between exposure to smoking and adolescent smoking behavior.

The aim of this study was to assess a mediation model whereby beliefs about the risks of smoking mediated the relationship between exposure to smoking and future smoking behavior. Consistent with the hypothesized mediation model, the results suggest that beliefs about the personal harm of smoking, assessed in mid-adolescence, mediated the relationship of household smoking, also assessed in mid-adolescence, with smoking behavior in late adolescence. The findings of this observational study suggest the importance of assessing how adolescents construe smoking health risks, in addition to evaluating smoking exposure in the household, as greater exposure may contribute to a lower perception of the harms associated with smoking. Future studies should explore more fully how environmental factors influence the development of adolescent smoking risk beliefs, and whether there are critical periods for the appropriation of risk beliefs.


   NOTES
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
REFERENCES
 
1However, this does not suggest that risk beliefs were associated with a mean level change in smoking from 10th grade to one year post high school. We tested whether risk beliefs (personal harm and general immediate harm) affected rate of change in smoking from baseline (10th grade) to last follow-up (one year post high school), and the effect was not significant. Back

Received for publication December 15, 2005; revision received August 1, 2006.

This study was supported by National Cancer Institute/National Institute on Drug Abuse grants P50 CA/DA 84718, NCI R01 CA109850, and NCI RO1 CA096836.

DOI:10.1097/PSY.0b013e31802e0f0e


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 

  1. Chassin, L, Presson CC, Rose JS, Sherman SJ. The natural history of cigarette smoking from adolescence to adulthood: demographic predictors of continuity and change. Health Psychol 1996;15:478–84.[CrossRef][Medline]
  2. US Department of Health and Human Services. Tobacco and the clinician: interventions for medical and dental practice. National Institutes of Health; 1994.
  3. Audrain-McGovern J, Rodriguez D, Tercyak KP, Neuner G, Moss HB. The impact of self-control indices on peer smoking and adolescent smoking progression. J Pediatr Psychol 2006;31:139–51.[Abstract/Free Full Text]
  4. Audrain-McGovern J, Rodriguez D, Tercyak KP, Cuevas J, Rodgers K, Patterson F. Identifying and characterizing adolescent smoking trajectories. Cancer Epidemiol Biomarkers Prev 2004;13:2023–34.[Abstract/Free Full Text]
  5. Conrad K, Flay B, Hill D. Why children start smoking cigarettes: predictors of onset. Br J Addict 1992;87:1711–24.[CrossRef][Medline]
  6. Farkas AJ, Gilpin EA, White MM, Pierce JP. Association between household and workplace smoking restrictions and adolescent smoking. JAMA 2000;284:717–22.[Abstract/Free Full Text]
  7. Proescholdbell RJ, Chassin L, MacKinnon DP. Home smoking restrictions and adolescent smoking. Nicotine Tob Res 2000;2:159–67.[Abstract]
  8. Wakefield MA, Chaloupka FJ, Kaufman NJ, Orleans CT, Barker DC, Ruel EE. Effect of restrictions on smoking at home, at school, and in public places on teenage smoking: cross sectional study. BMJ 2005;321:333–7.
  9. Wang MQ, Fitzhugh EC, Westerfield RC, Eddy JM. Family and peer influences on smoking behavior among American adolescents: an age trend. J Adolesc Health 1995;16:200–3.[CrossRef][Medline]
  10. Wills TA, Cleary SD. Peer and adolescent substance use among 6th-9th graders: latent growth analyses of influence versus selection mechanisms. Health Psychol 1999;18:453–63.[CrossRef][Medline]
  11. Bandura A. Social foundations of thought and action: a social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall; 1986.
  12. Fishbein M, Ajzen I. Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley; 1975.
  13. Chassin L, Presson CC, Rose JS, Sherman SJ. From adolescence to adulthood: age-related changes in beliefs about cigarette smoking in a midwestern community sample. Health Psychol 2001;20:377–86.[CrossRef][Medline]
  14. Gerrard M, Gibbons FX, Reis-Bergan M. The effect of risk communication on risk perceptions: the significance of individual differences. J Natl Cancer Inst Monogr 1999;25:94–100.[Abstract/Free Full Text]
  15. Romer D, Jamieson P. The role of perceived risk in starting and stopping smoking, in Smoking: Risk, perception, & Policy, P. Slovic, ed. Thousand Oaks, CA: Sage Publications; 2001.
  16. Slovic P. Affect, analysis, adolescence, and risk, in Reducing adolescent risk: toward an integrated approach, D. Romer, ed. Thousand Oaks, CA: Sage Publications, Inc: 2003:44–48.
  17. Gerrard M, Gibbons FX, Vande Lune LS, Pexa NA, Gano ML. Adolescents’ substance-related risk perceptions: antecedents, mediators and consequences. Risk Decision and Policy 2002;7:175–91.[CrossRef]
  18. Audrain J, Tercyak KP, Goldman P, Bush A. Recruiting adolescents into genetic studies of smoking behavior. Cancer Epidemiol Biomarkers Prev 2002;11:249–52.[Abstract/Free Full Text]
  19. Muthén BO. Mplus technical appendices. Los Angeles, CA: Muthén & Muthén; 1998–2004.
  20. Kann L, Kinchen SA, Williams BI, Ross JG, Lowry R, Grunbaum JA, Kolbe LJ; State and Local YRBSS Coodinators. Youth risk behavior surveillance–United States, 1997. MMWR CDC Surveill Summ 1998;47:1–89.[Medline]
  21. Audrain-McGovern J, Rodriguez D, Tercyak KP, Cuevas J, Rodgers K, Patterson F. Identifying and characterizing adolescent smoking trajectories. Cancer Epidemiol Biomarkers Prev 2004;13:2023–34.[Abstract/Free Full Text]
  22. Simons-Morton B, Chen R, Abroms L, Haynie DL. Latent growth curve analyses of peer and parent influences on smoking progression among early adolescents. Health Psychology 2004;23:612–21.[CrossRef][Medline]
  23. Audrain-McGovern J, Rodriguez D, Tercyak KP, Epstein LH, Goldman P, Wileyto EP. Applying a behavioral economic framework to understanding adolescent smoking. Psychol Addict Behav 2004;18:64–73.[CrossRef][Medline]
  24. Castrucci BC, Gerlach KK. Understanding the association between authoritative parenting and adolescent smoking. Matern Child Health J 2006;10:217–24.[CrossRef][Medline]
  25. Rodriguez D, Audrain-McGovern J. Team sport participation and smoking: analysis with general growth mixture modeling. J Pediatr Psychol 2004;29:299–308.[Abstract/Free Full Text]
  26. Loehlin JC. Latent variable models: An introduction to factor, path, and structural equation analysis. 4th ed. Mahwah, NJ: Lawrence Erlbaum Associates; 2004.
  27. Muthén LK, Muthén BO. Mplus User’s Guide, ed. 2nd. Los Angeles, CA: Muthén & Muthén; 2001.
  28. Hosmer DW, Lemeshow S. Applied Logistic Regression. 2nd ed. New York: John Wiley & Sons, Inc; 2000.
  29. Chassin L, Presson CC, Rose J, Sherman SJ, Davis MJ, Gonzalez JL. Parenting style and smoking-specific parenting practices as predictors of adolescent smoking onset. J Pediatr Psychol 2005;30:333–44.[Abstract/Free Full Text]
  30. Audrain-McGovern J, Rodriguez D, Patel V, Faith MS, Rodgers K, Cuevas J. How do psychological factors influence adolescent smoking progression? The evidence for indirect effects through tobacco advertising receptivity. Pediatrics 2006;117:1216–25.[Abstract/Free Full Text]
  31. Kobus K. Peers and adolescent smoking. Addiction 2003;98(Suppl 1):37–55.
  32. Conley Thomson C, Siegel M, Winickoff J, Biener L, Rigotti NA. Household smoking bans and adolescents’ perceived prevalence of smoking and social acceptability of smoking. Prev Med 2005;41:349–56.[CrossRef][Medline]
  33. Jackson C, Dickinson D. Enabling parents who smoke to prevent their children from initiating smoking: results from a 3-year intervention evaluation. Arch Pediatr Adolesc Med 2006;160:56–62.[Abstract/Free Full Text]
  34. Curry SJ, Mermelstein RJ. Do as I say, not as I do: does it work for tobacco use prevention? Arch Pediatr Adolesc Med 2006;160:102–3.[Free Full Text]
  35. Soteriades ES, DiFranza JR. Parent’s socioeconomic status, adolescents’ disposable income, and adolescents’ smoking status in Massachusetts. Am J Public Health 2003;93:1155–60.[Abstract/Free Full Text]
  36. Fairfax County Youth Survey Report, Communities that Care. Fairfax County, VA: Developmental Research & Programs; 2001.
  37. State and county quick facts. U.S. Census Bureau; 2001.



This article has been cited by other articles:


Home page
 AJPHHome page
A. V. Song, H. E. R. Morrell, J. L. Cornell, M. E. Ramos, M. Biehl, R. Y. Kropp, and B. L. Halpern-Felsher
Perceptions of Smoking-Related Risks and Benefits as Predictors of Adolescent Smoking Initiation
Am J Public Health, March 1, 2009; 99(3): 487 - 492.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Rodriguez, D.
Right arrow Articles by Audrain-McGovern, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Rodriguez, D.
Right arrow Articles by Audrain-McGovern, J.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS