We report in the current study an overall prevalence of current cigarette smoking among the studied cohort in 2004 of 15.7%, with lower smoking rates at lower ages from 4.0% in those 12 years old or younger to 25.7% in those 16 years or older. Males predominated in being smokers as has been shown in other studies [23–25]. However, in absolute terms, the difference in prevalence was rather minimal, 15.9% and 15.3% in males and females respectively. Our study also found that having a close friend who was a smoker, or living in a household with a smoker were both independently associated with being a current smoker among the adolescents. Previous studies have reported similar correlation between these characteristics and smoking [26–28].
In a prospective study on adolescent alcohol drinking reported by Fisher et al , having adults who drink in the home, underage sibling who drinks, peer who drinks, possession of or willingness to use alcohol promotional items, and positive attitudes toward alcohol were associated with an increased likelihood of alcohol initiation. The mechanisms operating in the case of alcohol i.e. role modeling, acceptability of a behavior within the home and among siblings and peers, easy access to a substance and positive attitude towards a behavior, may be operational in the case of smoking.
Due to the cross sectional nature of the data collection in the NYTS, it is not possible to ascribe causation or determine the exact sequence of events in an adolescents' smoking trajectory. It is plausible to consider that adolescents who befriend smokers are more likely to be influenced into smoking. It is equally plausible that adolescents who smoke are more likely to choose other smokers as their friends [30, 31]. Adult smokers within the home are also less likely to discourage adolescents from smoking. Furthermore the easy availability of cigarettes when other people are smokers in the home may facilitate initiation and maintenance of smoking among the adolescents. Conley Thomson et al  have reported that household smoking bans limits smoking among adolescents. Adults who do not smoke themselves are more likely to discourage smoking in the home and elsewhere than smokers. This study also found that as age increases, the likelihood of being a current smoker also increases.
We also found that white-non Hispanic adolescents were more likely to be smokers than African-American, Asian-Americans and Hispanic. American Indians however were as likely as Whites to be smokers. We did not determine why this may be the case from the available data. However, a literature review report by Tauras  found that Hispanics and African Americans were more responsive to changes in cigarette prices than whites. Furthermore, one study that was reviewed reported that adolescent white males who smoked were responsive to changes in smoke-free air laws, while adolescent blacks who smoked were responsive to changes in youth access laws. This may suggest that different racial/ethnic groups may respond to different public health interventions.
In agreement to previous studies [23, 24], perception that smoking was not harmful to health and having pocket money were associated with being a current smoker among adolescents. In an environment where adolescents may be legally employed to earn income from employment, it is possible that much of an adolescent's income may not be coming from parents. As such expecting parental supervision on adolescent spending may be particularly difficult. Also, it is possible that the amount of pocket money available to the adolescent may have just been a surrogate variable for the unmeasured time that the adolescent spends at work. Ramchand et al  have reported that the longer the hours an adolescent worked, the more likely she or he was to be a smoker.
Overall, 1 in 10 adolescents in the current study thought smoking was "cool", 1 in 3 owned an item with a tobacco brand, and exposure to pro-tobacco advertisements at gas stations, on the internet and smoking by characters in movies/television exceeded 40%. This amount of exposure should be seen in the light of evidence that media exposure to tobacco advertisements is associated with adolescent smoking [35–37].
Limitations of the study
This study had several limitations. First, the results reported in this study were based on self-reports. Study participants may have intentionally mis-reported history of smoking or done so inadvertently. In addition, the levels of reliability and validity in the specific settings where data collection occurred may result in differential bias. Furthermore, self-reported history of smoking was not validated with laboratory markers such as exhaled carbon monoxide, and blood or urine cotinine [38–40]. Brener et al  however have reported high reliability of this methodology to assess adolescent health risky behaviors. However, in the design and administration of the surveys, various steps were taken to mitigate such bias, for example students were participated anonymously. This was aimed to prevent intentional mis-reporting for fear of reprisals from school officials. However, it is not possible to gauge how far the study participants completed the questionnaires as accurately as possible.
Smoking rates among students who were habitually absent may have been different from the rates among their counterparts who attended school regularly. The external validity of the results from the current study may therefore be limited to students who were not absent on the day of administration. It is however likely that absent students are more likely to be smokers than those present in schools [42, 43], so our findings are likely to be underestimates.
Among persons aged 16 or 17 years in the United States, about 5% were not enrolled in a high school education and had not completed high school in 2000 . The questionnaire was offered only in English, and so with the increase in non-US born adolescents who may have difficulties in comprehension, there may have been misreporting. Finally, the factors assessed in this study were limited to socio-demographic variables. Key psychosocial variables known to affect substance use such as perceived norms, outcome expectancies, ability to resist peer appeals (self-efficacy), depression, truancy and stress were not included.