We appreciated the thoughtful comments provided by the four commentaries on our paper (Slomkowski et al. 2005). The commentaries included a number of converging opinions as well as specific suggestions to improve our understanding of sibling influences on smoking. All of the commentators agreed with our position that genetically informative designs may be utilized to generate evidence of environmental influence on behavioral phenotypes, especially when theory-driven measures of putative environmental influence are utilized. Thus Merikangas’ (2005) point that behavioral genetic designs may inform prevention strategies by yielding information on environmental mechanisms is well-taken, especially with reference to initiation of substance use in adolescence, which often shows evidence of shared environmental influences of equal or greater magnitude than genetic effects (Rende & Waldman, in press).

Vink (2005) provides a number of important conceptual and methodological considerations for such behavioral genetic work. One critical point is the measurement of the phenotype, an issue also addressed by Brandon & Brandon (2005). These authors emphasize both the complexity of smoking phenotypes and progress that is being made in the field in measuring a range of smoking indices, especially with adolescent populations. The data presented by Vink provide expansion of our findings by focusing on ‘maximum number of cigarettes smoked in a day’, a more specific indicator of adolescent smoking than the frequency measure used in our study. Vink also brings attention to the seminal work of Rose and colleagues (Rose et al. 1990), which is a good reminder that behavior genetics research has, for many years, generated important information on the social environment. As suggested by Vink, there are indeed a number of behavioral genetic models well-suited to address social influences on smoking, especially as they may be adapted to various representations of the smoking phenotype (e.g. categorical vs. continuous). We add to these good points the importance of broadening our conception and measurement of environmental influences in behavioral genetic models. For example, our paper expands the concept of social contact between siblings to include ratings of affection for each other and time spent with mutual friends, along with time spent together. Much of our current work emphasizes intensive methods for studying social interaction (e.g. micro social coding of videotaped interactions; recording sibling social contact and its contextual and affective features using experience sampling methodology) and these too offer a variety of methodological options for behavioral genetic research.

The inclusion of mutual friendships as part of the ‘sibling effect’ implies that sibling influence functions as part of a broader social context for smoking, a point emphasized by both Merikangas and Conger. Conger (2005) argues that there is a need to understand the larger social environment that includes parents as well as peers. Conger presents some of his seminal work demonstrating a variety of mechanisms by which parent, sibling and peer influences intersect. For example, a critical point for prevention is the finding that parent and sibling smoking may influence the acquisition of peers who smoke. Such directionality of effects suggests that family smoking may provide inroads to non-familial influences on smoking, a notion that is consistent with our model of sibling effects on smoking. The various levels of influence across social relationships discussed by Conger reinforces Merikangas’ point that social contextual effects reflect complex mechanisms which may be multiplicative rather than additive as well as her suggestion that the social contextual basis of adolescent smoking become a focus of prevention models.

In addition to this perspective on the social context, Brandon & Brandon (2005) offer an important emphasis on potential mechanisms by which social relationships transmit risk for smoking. Their focus on ‘units of communication’ reinforces the idea that social influences have impact via social cognitive processes, again a critical point for prevention studies. They offer an intriguing idea that sibling effects may operate in part by influencing the creation of expectancies about the consequences of smoking. This suggestion makes the broader point that work on social contextual influences on smoking may and should be linked with specific mechanisms that influence the individual.

In summary, the four commentaries offer a broad array of stimulating ideas on social influence, prevention, measurement, and methodology. What is impressive is the convergence from a number of perspectives on the importance of sibling influence on smoking. Sibling designs offer unique opportunities to examine familial and non-familial influences and provide opportunities to test both etiological and prevention models. Genetically informative sibling designs offer a further layer of potential by allowing for more insight into environmental factors as well as carry promise for eventually integrating both social and genetic influences on smoking.