Smoking, not smoking: how important is where you live?

Abstract

To explore and describe the social-environmental influence upon the likelihood of smoking tobacco for the Australian experience, in particular, Victoria. A multilevel (hierarchically structured regression) method was used. The micro-level units (characteristics of individuals) came from the annual surveys conducted by the Cancer Council of Victoria from 1990-97. The dependent variable was smoking status. The socio-demographic variables of age, marital status, education, employment status and ethnicity of individuals were used. The macro-level units (partitioned by postcode) came from the Australian Bureau of Statistics' Socioeconomic Index for Areas, split into quartiles. Residential neighbourhood had a unique but modest influence in the likelihood of smoking for both men and women. Also as the level of disadvantage increased within the residential area so did the odds of smoking for both men and women; however, the effect was not consistent. The odds of smoking were highest in the most disadvantaged areas for men, contrasting with women for whom the highest odds were in areas of more but not most disadvantage. The level of disadvantage of the residential neighbourhood has a unique, statistically significant influence, but not to the same degree as previously published Australian research. The effect is consistent across individual characteristics such as age and level of education for men. The effect is small and less consistent for women. Indeed, area of residence seems less important for women as a whole, suggesting differential influences according to gender.