Our study found that social and role model influences were clearly associated with the initiation and reinitiation of smoking in this population of young military adults. These factors were significant predictors of smoking status even after adjusting for other known risk factors in the model, such as family history of smoking, alcohol use, and depressive symptoms. Further, these results suggest that social role models were more predictive of smoking (Taylor, Conard, Koetting O’Byrne, Haddock, & Poston, 2004
) in this population of young adults than traditionally predictive demographic factors, such as age, gender, and education level (Eaton et al., 2006
The strongest social influence was that of the military training leaders and classroom instructors. Our findings indicate that perceptions of leaders’ tobacco use are associated with smoking initiation and reinitiation in these young adults. The role of professionals in perpetuating the belief that smoking is a culturally accepted and encouraged behavior in the military increases the risk for smoking among newly enlisted personnel. These young adults may believe that if their military role models smoke, then it must be a socially acceptable behavior despite prominent USAF antismoking messages and policies. These findings seem to mirror the associations seen between parental smoking habits and young adult smoking behavior (Buller et al., 2003
Our study also found that peer norms played a strong role in the initiation and reinitiation of smoking. Despite the fact that only 32% of young adults in this population smoke cigarettes (Bray et al., 2006
; Bray & Hourani, 2007
), respondents who were smokers perceived this percentage to be much higher. The close-knit nature of the military student population and the fact that smokers are more visible as they are required to smoke in designated outdoor locations may have contributed to the perception that the majority of students smoke. Even though this perception is incorrect, the belief appears to be most powerful for those who are most vulnerable to initiate or reinitiate smoking. This finding is consistent with the literature that salient peer models and social norms are some of the strongest influences on the initiation of smoking (Buller et al., 2003
; Kandel, Kiros, Schaffran, & Hu, 2004
; Killen et al., 1997
; Maxwell, 2002
; Urberg, Degirmencioglu, & Pilgrim, 1997
). Interestingly, although technical training students who were smokers were more likely to perceive that the majority of their classmates smoked, they did not hold this same perception toward permanent party personnel. This may be partially explained by the relative separation of students and permanent party personnel. Students are housed on a separate part of the military bases, and social interaction is highly discouraged between permanent party personnel and students. As such, the proximity and perceived similarity of the social role models may influence the impact of the model on one’s behavior and may help to explain these differences (Kandel et al., 2004
; Maxwell, 2002
; Urberg et al., 1997
Having a roommate that smoked was predictive of initiation of smoking among previous nonsmokers, but was not predictive of reinitiation of smoking among those who had smoked previously. It may be that the presence of a smoking roommate may be enough of an influence to tempt a nonsmoker to try smoking, whereas for those who have a history of smoking other factors, including addiction, outweigh the influence of a smoking roommate.
Several factors that were used as control variables were found to be significant predictors of smoking, including race/ethnicity, pre-military alcohol use, whether the personnel were under an enforced ban on smoking, and depression. Based on existing literature, it is not surprising that previously using alcohol or being white resulted in greater likelihood to initiate or reinitiate smoking in these young adults (Ames, Cunradi, & Moore, 2002
; Eaton et al., 2006
; Griesler et al., 2002
; Haddock, Klesges, Talcott, Lando, & Stein, 1998
; Klesges et al., 1999
; Voorhees, Schreiber, Schumann, Biro, & Crawford, 2002
). Although we did not collect data during basic training, other researchers have indicated that smoking in basic training is virtually nonexistent due to a strict ban (Klesges et al., 2006
). This is consistent with our finding that the existence of a smoking ban was the strongest predictor of smoking, which demonstrates that early training restrictions are beneficial in reducing smoking. However, as smoking restrictions for military students are relaxed, personnel are at increased likelihood to smoke. It may also be that once permitted to smoke, general messages from the public and the Air Force about the harms of smoking are disregarded. Depression was more predictive of reinitiation than initiation of smoking. This is consistent with findings about the link between depression and smoking where depression has predicted greater relapse to smoking and lower long-term abstinence rates (Burgess et al., 2002
; Niaura et al., 2001
These findings have important implications for smoking prevention and early smoking cessation intervention efforts. The findings that smoking behavior by role models and norms about peer smoking both increase the likelihood of smoking suggest that changing role models’ smoking behavior and modifying (possibly inaccurate) perceptions of who smokes may help reduce smoking initiation or reinitiation. Role models and peers might be used to highlight more accurate social norms, promote a nonsmoking norm, or help with early cessation efforts. In addition, the finding that those who have been in training longer are more likely to smoke suggests that restrictions on smoking during earlier weeks of training may be reasonably successful, but when restrictions are lifted, personnel are more likely to initiate or reinitiate smoking. It could be that another method to decrease smoking rates among young military personnel is to increase the time period that smoking is not permitted.
One of the strengths of this study is that it was one of the first to evaluate the relative contributions of multiple social influences to smoking initiation among large numbers of USAF technical training students. This study included the four largest technical training bases in the USAF and we feel that our results provide a reasonably accurate representation of this population. However, the fact that this study was limited to USAF military personnel may affect the ability to generalize the results to other military services and the civilian sector.
One of our study’s limitations is the amount of missing data; however, this was addressed through multiple imputation. The length of the survey may have accounted not only for the missing data but also for our response rate. It is also important to note that the cross-sectional nature of these data prevent any solid conclusions about the causal order of the association between perceptions of peer and role model smoking and the initiation/reinitiation of smoking. For example, it could be that these findings resulted from the “false consensus effect,” or the tendency for individuals to overestimate the degree to which other people share their beliefs and behaviors (Ross, Greene, & House, 1977
). This would suggest that those who smoke would overestimate the overall prevalence of smoking, which would explain the reported associations between smoking and perceptions of others’ smoking. However, if these results were in fact due to the false consensus effect, one would also expect to find a significant association between smoking and perceptions of the prevalence of smoking among permanent party personnel. The lack of such an association argues against a general false consensus effect explanation for these findings.
In summary, our study suggests that the likelihood of smoking initiation among newly enlisted military personnel who have recently undergone a period of forced abstinence is increased by (a) military role models who use tobacco, (b) by peer smoking behavior, and (c) by perceived smoking norms. Smoking cessation and prevention often ignores this young adult group (18 to 25 years old) and targets younger children or adult smokers (Backinger, Fagan, Matthews, & Grana, 2003
; Lantz, 2003
). Given the findings in this study, it is likely that the influence of peers and role models will need to be factored into successful prevention or early cessation efforts. Future research will need to explore the strength of peer and role model relationships in other settings and populations, such as high school, vocational, or college students. Further, these relationships should be explored prospectively to determine the strength of these factors as predictors of future smoking behavior.