Elsevier

Health & Place

Volume 8, Issue 3, September 2002, Pages 155-166
Health & Place

Youth smoking initiation: disentangling context from composition

https://doi.org/10.1016/S1353-8292(02)00003-5Get rights and content

Abstract

This paper examines the relative contributions of compositional and contextual effects on the one hand, and behavioural and material factors on the other, to help explain smoking initiation. We present results from a cross-sectional, multi-level analysis linking data from pre-adolescents, their households and their neighbourhoods across Quebec, Canada. Results show a significant geographical area variation in youth smoking initiation. The variation is explained not only by individual characteristics (parents’ smoking status, parents’’ education and pre-adolescents’ age), but also by aspects of the social structure at the neighbourhood level. When both the individual level and territory level predictors are entered in the model, only 2.66% of the between-territory variance in smoking initiation remain unexplained.

Introduction

A continually growing body of research has emerged over the past decade focussing on the respective contributions of contextual and compositional effects in public health research (Duncan et al., 1993; Duncan et al., 1998; Diez-Roux, 1998; Diez-Roux et al., 2000). A parallel debate on the theoretical constructs underlying community effects as well as on measurement of these effects is unfolding (Diez-Roux, 1998; Diez-Roux et al., 2000; Diehr et al., 1993; Birch et al., 1998). Much of both of these debates centres around the issue of whether the relationship to health of a particular variable, such as socioeconomic status (SES) aggregated at the community level, simply reflects the relationship of SES at the individual or family level, or whether there is an effect of community SES on individual health that goes over and above the effects of individual or family level SES (Robert and House, 2000). Otherwise stated, these studies attempt to tease out whether aggregate effects are artefacts of population composition measured at an individual level or genuine effects of some contextual characteristic. The key question in most of these studies is what is the origin of the variations in disease outcomes found between different communities; individual or aggregate attributes?

Analyses of context, in fact, tend to frame the origins of differential disease rates in terms of one of two issues, the first of which is lower versus higher levels of explanation. The origins of these effects may be due to what are frequently called compositional attributes (Duncan et al., 1998; Macintyre et al., 1993; Duncan et al., 1996; Duncan et al., 1999). These attributes are understood to be those of persons living in certain types of areas (Diez-Roux, 2000). The compositional factors most frequently examined are indicators of SES such as individuals’ social class, housing tenure, employment status, educational status, marital status, etc. Alternatively, the origins may lie in what is termed the contextual level, or in neighbourhood characteristics (Diez-Roux, 2000). Contextual effects are presumed to have an impact on the individual actor over and above the effects of her/his own characteristics. These macro-level variables may be either summary measures of compositional attributes, such as median income of an area, or they may be other than simple summaries of such variables such as values, norms or geographic characteristics of an area (Macintyre et al., 1993; Blalock, 1984; Sooman and Macintyre, 1995; Ellaway and Macintyre, 1996; Macintyre and Ellaway, 1998; Cheadle et al., 1992). These have been called supra-individual variables (Macintyre et al., 1993), environmental indicators (Cheadle et al., 1992) and integral variables (Diez-Roux, 1998).

For the most part this higher level of explanation has been underdeveloped in public health studies and is rarely examined as other than a deprivation index or the level of inequality at the census tract, region, or state levels (Diez-Roux, 2000; Duncan et al., 1999; Soobader and Leclere, 1999; Pickett and Pearl, 2001). Recently there has been discussion, however, of the relationship between the neighbourhood environments in which people live and disease outcomes (Sundquist et al., 1999). Attention has been focused on examining neighbourhoods in terms of access to healthy foods, leisure physical activities, cultural activities, safe recreation spaces, and smoking-free environments (Macintyre et al., 1993; Sooman and Macintyre, 1995; Ellaway and Macintyre, 1996; Macintyre and Ellaway, 1998).

An equally important issue has been disregarded, namely—what should be evaluated at the ecological level. In particular, we believe that the variance documented in recent years at the aggregate level originates from both the aggregate characteristics of people in places, as well as the supra-individual or ecological characteristics of places.

The second way in which context is examined is in terms of two categories of disease correlates: “material/structuralist factors” or “behavioural” factors (Townsend and Davidson, 1988; Blaxter, 1990; Glendinning et al., 1995; Stronks et al., 1996; Macintyre, 1997). With the former it is believed that material conditions contribute to class gradients in health—conditions that have been operationalised as education or income. The latter on the other hand focuses on the contribution of health damaging behaviours, such as smoking or alcohol consumption, to social class gradients. What both classes of correlates have in common is that they attempt to explain how locally based social phenomenon influence people's biology—their health status.

We adopt an approach for understanding the effects of social correlates on disease outcomes by returning to the term “lifestyle” in the original sense given to the concept by Max Weber. Weber (1922) viewed lifestyle as being both a reflection of one's social status as well as what one consumes. Weber operationalised lifestyle as the actualisation of choices as influenced by life chances. Life chances include rights, norms, and social relationships. Chance is therefore socially determined and the social structure is an arrangement of chances. Lifestyles are not, therefore, random behaviours unrelated to structure, but are choices influenced by life chances. Thus, one of Weber's contributions to the definition of lifestyle is to introduce a dialectic between choice and structure in lifestyle formation.

A similar conceptualisation of lifestyle may assist public health researchers disentangling material from behavioural factors and therefore identifying appropriate targets for intervention. While not completely equivalent, much of the time the “material factors” used in public health studies, such as SES, are taken to be operationalisations of the social structure, with health behaviours understood as examples of choice.

The purpose of this study was to empirically contrast these two assumptions using data on smoking initiation among pre-adolescents in Québec, Canada. In this exploratory study, therefore, we first “contextualise” smoking initiation by examining the relationship between both material/structural factors (chances) and behavioural factors (choices) at the individual level. We do not treat behaviour and material conditions as separate generators of disease, but rather conceive of behaviour as being embedded in material conditions (Macintyre, 1997). In this way, material and behavioural factors will not be opposed, or one controlled for the other, but viewed as jointly forming the practice of smoking in neighbourhoods. Second, in an effort to operationalise aspects of the social structure (or what others term context), and to further contextualise smoking initiation, we analyse what we call “agents”, the resources that they make available, and their relationship to smoking initiation. Using these data we test whether there are aspects of the social structure at the neighbourhood level, in addition to other classic indicators of SES, that influence smoking initiation prevalence. Our hypothesis is that what are frequently called contextual, or higher-level effects can be partitioned into both individual aggregate effects (that is, aggregate variables constructed by summarising the characteristics of groups of people) as well as supra-individual influences. As such, what we later term pro-smoking areas may have an impact on youth smoking as a result of both the composition of the community (individual attributes), as well as the structurally encouraging attributes that abound in the community with reference to smoking. Lastly, we discuss whether there might be community-level effects that are constant across different types of people, that is, combinations of variables at the higher level that might explain variation in health outcomes above and beyond that explained by individual level effects.

Section snippets

Research design and sample

The results presented here are part of a research project concerned with the intermediate role that families play between community health promotion and individual behaviour and the way in which community characteristics are associated with families’ and youths’ health behaviours (cigarette smoking, physical activity, and dietary fat consumption). A cohort of families was assembled in 1995 based on the selection of a fourth grade index child in 47 participating elementary schools in

Sample characteristics

Table 1 shows descriptive statistics for the 694 pre-adolescents and Table 2 depicts information for the 32 territories. In Table 1 we note that there is variation in each of the variable categories with a high proportion of pre-adolescents having been initiated to smoking by grade six (34.3%).

Table 2 gives the distribution for the variables to be employed in the second level of the hierarchical analyses. Again there is important variance across the 32 territories for all of the second-level

Discussion

The results of this study demonstrate that both individual level predictors and territory-level predictors are associated with smoking initiation among pre-adolescents. Furthermore, territory-level effects seem to explain variation in smoking initiation above and beyond that explained by individual-level predictors. These results support other research findings showing some form of area effect on smoking (Duncan et al., 1999; Glendinning et al., 1997). There is definitively variation in smoking

Conclusion

We can state from the results of this hierarchical model that it appears that “Weberian” chance, operationalised by both individual SES as well as by the supra-individual variables at level-two, are related to the “choice” that pre-adolescents are making to start smoking. We would elaborate further on this by adding that the social structure is an arrangement of “chances”, both compositional and contextual, and that further research could focus on disentangling aspects of the social structure

Acknowledgements

This research was made possible by the National Health Research and Development Program (NHRDP) research grant #6605-4006-210. During the development of the project K. L. Frohlich was supported in part by Health Canada through an NHRDP Research Training Award (6605-5226-47R) as well as through a GM Fellowship from the Canadian Institute of Advanced Research (CIAR) Population Health Program. L. Potvin is supported through a Scientist Award (H3-17299-AP007270) from the Medical Research Council of

References (37)

  • S. Macintyre

    The black report and beyond. What are the issues?

    Social Science and Medicine

    (1997)
  • S. Macintyre et al.

    Social and local variations in the use of urban neighbourhoodsa case study in Glasgow

    Health and Place

    (1998)
  • S.A. McGraw et al.

    Sociocultural factors associated with smoking behaviour by Puerto Rican adolescents in Boston

    Social Science and Medicine

    (1991)
  • M.-J. Soobader et al.

    Aggregation and the measurement of income inequalityeffects on morbidity

    Social Science and Medicine

    (1999)
  • A. Sooman et al.

    Health and perceptions of the local environment in socially contrasting neighbourhoods in Glasgow

    Health and Place

    (1995)
  • S.L. Bailey et al.

    Potential mediators, moderators, or independent effects in the relationship between parents’ former and current cigarette use and their child's cigarette use

    Addictive Behavior

    (1993)
  • S. Birch et al.

    Modelling the community as a determinant of health

    Canadian Journal of Public Health

    (1998)
  • H.M. Blalock

    Contextual-effects modelstheoretical and methodological issues

    Annual Review of Sociology

    (1984)
  • Cited by (71)

    • The tobacco endgame: The neglected role of place and environment

      2018, Health and Place
      Citation Excerpt :

      Particularly important has been an increased focus on the density of tobacco retailers, especially near schools, with recent papers extending consideration to e-cigarette retail and use (Bostean et al., 2018, 2016; Corsi and Lippert, 2016; Larsen et al., 2017; Lipperman-Kreda et al., 2015; Yu and Lippert, 2017). Other studies have examined youth smoking more broadly, seeking to understand the relevance of neighbourhood effects (Frohlich et al., 2002) and disentangle the general impact of proximity to tobacco retail (Mennis et al., 2016). Collectively these studies offer pointers to local supply-side interventions to hasten the endgame; similar possibilities exist with work on bans on smoking in children's playgrounds (McIntosh et al., 2015).

    • The interactive effect of neighborhood peer cigarette use and 5HTTLPR genotype on individual cigarette use

      2014, Addictive Behaviors
      Citation Excerpt :

      For instance, the smoking behaviors of close and extended friends are associated with one's own cigarette smoking (Alexander, Piazza, Mekos, & Valente, 2001), as are adolescents' perceptions of peer smoking behaviors (Chassin, Presson, Sherman, Corty, & Olshavsky, 1984) and the school and neighborhood smoking rates (Eitle & Eitle, 2004). Similarly, neighborhood and area groupings show evidence of significant clustering of smoking behavior (Chahine, Subramanian, & Levy, 2011; Duncan, Jones, & Moon, 1999) and smoking initiation in adolescence (Frohlich, Potvin, Gauvin, & Chabot, 2002). Thus, the smoking behaviors of adolescents are linked through common affiliation in social institutions (schools and neighborhoods) as well as voluntary associations (friendships).

    View all citing articles on Scopus
    View full text