As primary socializing agents, parents can exert an important influence on their offspring’s smoking behavior. Recent studies on the role of parents in the development of adolescents’ smoking have concentrated on parents’ anti-smoking socialization practices (e.g., Chassin et al.
2005; Harakeh et al.
2005; Jackson and Henriksen
1997). Anti-smoking socialization practices reflect specific ways by which parents attempt to prevent smoking onset or smoking maintenance of their children, such as setting smoking-specific rules and giving rewards for not smoking. One important aspect of anti-smoking socialization is the communication about smoking-related issues. Through conversations, parents cannot only explain the house rules (Clark et al.
1999) but also can discuss reasons for not smoking, which is related to a lower risk of adolescent’s smoking (Chassin et al.
1998). Nevertheless, whether conversations are effective or not seems to depend on the quality of the parent–child communication. The quality of communication appears to be a protective factor, which indicates that parents who discuss smoking-related issues in a constructive and respectful manner with their children can prevent adolescents from smoking (Harakeh et al.
2005).
At the same time, parents also influence their children through their own smoking. It has been shown that adolescents have a higher risk to start smoking or to continue smoking when one or both parents are smokers (see review by Mayhew et al.
2000). This can be explained by genetic factors (e.g., Brody et al.
2006) and by children modeling parental behaviors (Bandura
1977). However, the question arises how strong these effects of parental smoking are compared to the attempts by which parents actively try to prevent their children from smoking. Perhaps if parents who smoke apply smoking-specific socialization strategies in a successful way, they might simultaneously also diminish the impact of their own smoking. Overall, effective smoking-specific socialization practices seem to be related to lower rates of smoking among adolescents with non-smoking and smoking parents (Clark et al.
1999; den Exter Blokland et al.
2006; Harakeh et al.
2005; Jackson and Henriksen
1997).
Although parents are important socialization agents, peer relationships become increasingly important in the adolescent years, and peer influences are often considered one of the main sources for adolescents’ involvement in smoking (see reviews by Avenevoli and Merikangas
2003; Kobus
2003). Close friends particularly seem to serve as significant (role) models for adolescents: Comparing the impact of close friends and peer groups, it appears that the smoking of the closest friend is more strongly related to onset of smoking in youngsters (Urberg et al.
1997), suggesting that adolescents mainly observe, model, and imitate the smoking behavior of their best friends. However, recent findings suggest that the magnitude of peer influences on risk behaviors might be overestimated, because in many studies the effects of friendship-selection processes were not taken into account (e.g., Jaccard et al.
2005). As previously stated by Kandel (
1978), homophily between friends at one point in time is not only the result of socialization processes, but also due to selection processes, which imply that adolescents select their friends on the basis of shared characteristics. It is known that both influence and selection processes contribute to homogeneity in peers with respect to smoking (Engels et al.
1997; Ennett and Bauman
1994; Fisher and Bauman
1988). These findings suggest that adolescents select their friends on the basis of comparable smoking attitudes and behaviors. In combination with reciprocal modeling influences, this process contributes to similarities in smoking behavior between adolescents and their friends.
As argued by Kandel (
1996), research on the relative influence of peers and parents on adolescents’ drug use have inflated the importance of peers and underestimated the influence of parents. One of the aspects that should be taken into account when disentangling the relative impact of both social factors is the contribution of parents to children’s peer selection (see also Melby et al.
1993; Rowe et al.
1994). Previous research indicated that parents can function as managers of their offspring’s peer relationships, for example by acting as an advisor and consultant (Ladd and Pettit
2002). During day-to-day conversations parents are able to communicate with their offspring about peer relationships, such as how to initiate friendships, manage conflicts, maintain relationships, deflect teasing, and so on. In this way, parents are able to provide advice or solutions to problems in peer relationships, or listen to their child’s self-generated assessments and solutions. One of the topics that could also be discussed is this respect is substance use. Parents might advise their child about how to deal with peers who let themselves in with substance use, as specific cigarette use. Thus, besides giving anti-smoking messages, parents might express their feelings about smoking peers and advise their adolescent in manners to resist pressure to smoke. Additionally, one might expect that parents affect their child’s affiliation with (non-)smoking peers through these conversations. Parents who discuss smoking-related issues in a constructive and respectful manner might prevent their child not only from smoking but also from affiliating with smoking friends. Presumably, this possible effect of parents on adolescents’ friendships might reflect
direct influences of the parental smoking-specific communication, but also
indirect influences, by which parents affect peer affiliation through the adolescents’ behavior: Adolescents who do not smoke themselves, are more likely to select friends who also do not smoke. Investigating the impact of parents on adolescents’ friendships is important as previous research showed that imitation plays a major role in smoking (Harakeh et al.
2007) and that the risk for smoking increased dramatically as the number of smoking models in the adolescent’s environment increased (Taylor et al.
2004). When parents are able to prevent their child from affiliating with smoking friends, they take away an important source of smoking modeling and, accordingly, decrease the likelihood that their child will smoke in the future.
Discussion
The aim of the present study was to investigate whether the quality and frequency of parental smoking-specific communication and parental smoking are associated with adolescents’ friendship-selection processes. Furthermore, we investigated whether adolescents and their best friends seem to influence each other over time, and what role parental smoking-specific communication and parental smoking play in these processes. The results show that a high quality of parental smoking-specific communication was related to a lower likelihood of adolescent smoking. The frequency of communication was found to be positively associated with adolescent smoking. To investigate the impact of parental smoking-specific communication on adolescents’ friendship-selection processes, we differentiated two groups: one group of adolescents who reported to have the same best friend over a period of two years, thus the group in which no selection processes occurred; and one group of adolescents with changing friendships over this same period. The latter group enabled us to investigate friendship-selection processes. In line with previous findings, we found evidence for friendship selection based on smoking, because adolescent’s smoking affected best friend’s smoking over time (Engels et al.
1997; Ennett and Bauman
1994). Moreover, our results demonstrated direct and indirect associations between the quality and frequency of parental smoking-specific communication and best friend’s smoking, which indicate that parents can affect their child’s friendship-selection processes (see also Brown et al.
1993). Thus, parents seem to be able to affect their child’s friendship-selection processes
directly, but also
indirectly: When parents succeed in their attempts to prevent their child from smoking, this subsequently will decrease the likelihood that their child will affiliate with smoking friends.
Additionally, the group of adolescents with stable friendships enabled us to examine whether adolescents and their best friends influence each other over time, and whether parents play a role in this. We tested whether parental smoking-specific communication is still related to their child’s smoking, even when their child affiliates with the same friend for a longer period of time. The findings suggest that parents do not appear to directly affect their child’s smoking over a period of one year. It is generally known that during adolescence a transformation of close relationships takes place; as friendships grow closer, the intensity and exclusivity of the parent–child relation decrease (e.g., Laursen and Bukowski
1997). This might explain why adolescents with stable friendships are no longer affected by their parents in terms of smoking. However, that no direct associations were found between parental smoking-specific communication and future adolescent smoking does not necessary imply that parents are not influential. The cross-sectional associations between parenting and adolescent smoking were strong, which might indicate that parents influence their children on the short term. Moreover, adolescents’ smoking behavior is quite stable over time. This could imply that parental smoking-specific communication indirectly affects adolescent smoking one year later, suggesting that when parents prevent their children from smoking during early adolescence, they increase the chance that their children remain non-smokers in the later years.
Smoking parents appear to be less constructive and supportive in their smoking-specific communication than non-smoking parents. This is probably because smoking parents may be more uncertain about their possibilities to prevent their child from smoking, because their advice does not match their actions (e.g., Jackson and Dickinson
2003). Furthermore, longitudinally we found no significant association between parental smoking and adolescent’s smoking when controlling for parental smoking-specific communication. Our interpretation is that what parents express through their smoking-specific communication has more impact on their child’s behavior than the direct effects of parental smoking. However, one should keep in mind that this does not imply that parental smoking does not play a role, because it does affect the quality (and credibility) of parental smoking-specific communication. Prevention programs, therefore, should focus not only on strengthening parents in their ways of communication, but also on smoking cessation among parents.
Findings from our study indicate some directions for future research. When parents do affect their children’s friendship-selection processes through smoking-specific communication, the question arises whether other parental practices affect these selection processes as well. An important aspect of parenting is monitoring and the extent to which parents have knowledge about their children’s behavior, friends, and whereabouts (Stattin and Kerr
2000). It would be interesting to examine whether monitoring and parental knowledge is indirectly related to a lower likelihood of adolescent smoking through the selection of non-smoking friends (see Engels et al.
2005). Mounts (
2000) identified several ways parents can deal with adolescents’ peer relationships, including
guiding, which is talking about the consequences of being friends with particular people. Another way is
prohibiting, which is when parents forbid their child to associate with specific peers. Finally,
supporting is a way in which parents encourage their child to undertake activities with peers they like and provide an environment at home where adolescents can have their friends over. It is important to keep in mind that during adolescence parents are challenged to find a balance between autonomy granting and exerting control. Focusing on several peer management skills might lead to an improved understanding of the parental role in adolescents’ friendship selection and could answer the question whether parents should function especially as friendship-formation gatekeepers or whether they should function mainly as advisors or consultants. Additionally, it would be interesting to take the quality of the parent-adolescent relationship into account.
Future research should focus not only on the impact parents have on their children, but also on the effects of the children on their parents. Investigating bidirectional associations between parental practices and adolescent’s behavior allows ascertaining the actual effects of parents on their children, without ignoring the impact children have on their parents. Only a few studies have investigated the bidirectional relationships between parenting and adolescent’s problem behavior, and they clearly support the idea that parents do respond to the problem behavior of their offspring (see Lytton
1990; Stice and Barrera
1995). Recent research has shown that parents also respond to their children’s smoking behavior by increasing smoking-specific conversations (Huver et al.
2007). Finally, one might expect that parents react not only to their child’s behavior, but also to behavior of their child’s best friend: When parents notice that their child is associating with smoking friends they may adjust their parenting practices in order to prevent their child from smoking.
In the current analysis, we examined the quality of smoking-specific communication as a general construct. It would be interesting—and important for prevention—to explore more specific aspects of a highly qualitative parent–child discussion about smoking-related issues. For example, some studies have shown that rigidity in parent–child interactions is linked to externalizing behavioral problems (Hollenstein et al.
2004). Investigating parent–child discussions in an observational setting would allow for analyzing several parent–child communication characteristics, such as flexibility versus rigidity, which subsequently provides more insight into the relation between the parent–child communication and adolescent’s smoking. Another important implication for future research involves the individual characteristics of youngsters. It is plausible that individual differences play an important role in the associations between parental smoking-specific communication and adolescent’s smoking. For example, adolescents who are highly responsive to their parents’ viewpoint might comply with their parents’ advice concerning smoking, while adolescents who are not very responsive to their parents’ viewpoint might neglect their parents’ advice (Jackson
2002).
Another point of interest concerns socialization between adolescents and their best friends. Our findings demonstrated that best friends smoking was marginally related to individual smoking. Although this impact was small, it existed while taking into account the associations between parental smoking and parental smoking-specific communicating, implicating that friend’s smoking contribute to adolescent smoking beyond the impact of parents. Therefore, a next step for future research is to gain more insight into this socialization mechanism. For instance, it might be possible that friends who are older exert a stronger influence on adolescent smoking than a friend who is of the same age or younger. Additionally, sex constellation of the friendship pair might contribute to the strength of the socialization as well, as same-sex friends might exert more influence on each other than different-sex pairs. In short, it is interesting to examine which factors are responsible for the strength of friends’ smoking socialization.
The present study is the first to investigate the relationship between parental smoking-specific communication, smoking of adolescents, and smoking of adolescents’ best friend. Findings provide preliminary evidence that parents are able to affect not only their offspring’s smoking through their smoking-specific communication, but also the peers their child will associate with. However, our pattern of results might not be generalizable to the population as a whole because we used a selective sample of intact families. Therefore, it is necessary to replicate these findings in other samples, such as single-parent families, and specific ethnic groups. Additionally, although we controlled for adolescents’ age in our model, it is plausible that the influence of parents and friends on adolescent smoking might vary during the teenage years. A recent study from Steinberg and Monahan (
2007) showed that middle adolescence is an especially significant period for the development of the capacity to resist peer influence. Therefore, it would be interesting to replicate these findings in large samples of different age groups. The fact that we had to combine the data of the younger and older adolescents also limited our ability to investigate whether siblings influence each other with respect to smoking and their selection of (non-)smoking friends (see Melby et al.
1993).
Some other limitations of this study should be addressed. Although we used self-reports of the adolescents and their parents, best friends’ smoking status was based on adolescents’ perceptions. These perceptions might differ from best friend’s actual smoking, as adolescents tend to project their own behavior on their friends (e.g., Bauman and Ennett
1996). However, in a previous study with the same sample as used in the present study, Harakeh et al. (
2007) examined whether adolescents were accurate in their reports about their best friend’s smoking. Results of that study indicated a substantial agreement between adolescents’ reports and self-reports of the best friends. Another limitation is that adolescents were asked for information only about their best friend and not for information about other important friends. Therefore, we were not able to examine among the adolescents with changing friendships whether their best friends stay close friends and continued to influence them, or disappeared completely from the adolescent’s immediate environment during the measurement period.
Finally, it is important to stress that we were not able to take genetic influences into account. Recently, scholars have emphasized the importance of encompassing both genetic and social influences in research designs, as this will provide more insight into the specific causes of particular behaviors (e.g., Rutter et al.
1997). For example, it might be possible that adolescents’ friendship-selection processes are influenced by genetic factors as previous research indicated that individuals create their own social environment based on their genetic propensities (e.g., Rose
2007). Thus, adolescents may select friends with a smoking status that fits their own due to specific genetically influenced characteristics, like for example sensation seeking (e.g., Zuckerman
2007). However, there is also evidence that a favorable social environment can dramatically reduce or even eliminate the adverse influence of genetic factors (e.g., Reiss and Neiderhiser
2000; Rose
2007). If parents communicate in a constructive and respectful manner with their offspring, this will create such a favorable environment which then could operate as a protective factor by which possible disadvantageous genetic influences can be suppressed.
In conclusion, the present findings suggest that parental smoking-specific communication affects adolescent’s smoking directly but also indirectly by influencing the friends the adolescents will associate with. Therefore, smoking prevention programs should focus on strengthening parents in the way they communicate about smoking-related issues. Parents who discuss smoking-related issues in a constructive and respectful manner will be more likely to influence their child’s smoking. In addition, parents who communicate in this way might also decrease the likelihood that their child will affiliate with smoking friends.