Tobacco outlet density and demographics

Analysing the relationships with a spatial regression approach

Danlin Yu, N. A. Peterson, M. A. Sheffer, Robert Reid, J. E. Schnieder

Research output: Contribution to journalArticleResearchpeer-review

61 Citations (Scopus)

Abstract

Objective: Studies of relationships between tobacco sales and socio-economic/sociodemographic characteristics are well documented. However, when analysing the data that are collected on geographic areas, the spatial effects are seldom considered, which could lead to potential misleading analytical results. This study addresses this concern by applying the spatial analysis method in studying how socio-economic factors and tobacco outlet density are related in New Jersey, USA. Study design: A spatial regression method applied to tobacco outlet and socio-economic data obtained in 2004 in New Jersey, USA. Method: This study assessed the association between tobacco outlet density and three demographic correlates - income, race and ethnicity - at the tract level of analysis for one state in the north-eastern USA. Data for 1938 residential census tracts in the state of New Jersey were derived from 2004 licences for 13,984 tobacco-selling retail outlets. Demographic variables were based on 2000 census data. When applying a regression model, the residuals of an ordinary least squared (OLS) estimation were found to exhibit strong spatial autocorrelation, which indicates that the estimates from the OLS model are biased and inferences based on the estimates might be misleading. A spatial lag model was employed to incorporate the potential spatial effects explicitly. Results: Agreeing with the OLS residual autocorrelation test, the spatial lag model yields a significant coefficient of the added spatial effect, and fits the data better than the OLS model. In addition, the residuals of the spatial regression model are no longer autocorrelated, which indicates that the analysis produces more reliable results. More importantly, the spatial regression results indicate that tobacco companies attempt to promote physical availability of tobacco products to geographic areas with disadvantageous socio-economic status. In New Jersey, the percentage of Hispanics seems to be the dominant demographic factor associated with tobacco outlet distribution, followed by median household income and percentage of African Americans. Conclusion: This research applied a spatial analytical approach to assess the association between tobacco outlet density and sociodemographic characteristics in New Jersey at the census tract level. The findings support the common wisdom in the public health research domain that tobacco outlets are more densely distributed in socio-economically disadvantaged areas. However, incorporating the spatial effects explicitly in the analysis provides less biased and more reliable results than traditional methods.

Original languageEnglish
Pages (from-to)412-416
Number of pages5
JournalPublic Health
Volume124
Issue number7
DOIs
StatePublished - 1 Jul 2010

Fingerprint

Tobacco
Demography
Spatial Analysis
Censuses
Economics
Spatial Regression
Vulnerable Populations
Licensure
Hispanic Americans
Research
Tobacco Products
African Americans
Public Health

Keywords

  • Demographic factors
  • New Jersey
  • Spatial regression
  • Tobacco outlet density

Cite this

Yu, Danlin ; Peterson, N. A. ; Sheffer, M. A. ; Reid, Robert ; Schnieder, J. E. / Tobacco outlet density and demographics : Analysing the relationships with a spatial regression approach. In: Public Health. 2010 ; Vol. 124, No. 7. pp. 412-416.
@article{18769096fac5449d95bdd687defc2c75,
title = "Tobacco outlet density and demographics: Analysing the relationships with a spatial regression approach",
abstract = "Objective: Studies of relationships between tobacco sales and socio-economic/sociodemographic characteristics are well documented. However, when analysing the data that are collected on geographic areas, the spatial effects are seldom considered, which could lead to potential misleading analytical results. This study addresses this concern by applying the spatial analysis method in studying how socio-economic factors and tobacco outlet density are related in New Jersey, USA. Study design: A spatial regression method applied to tobacco outlet and socio-economic data obtained in 2004 in New Jersey, USA. Method: This study assessed the association between tobacco outlet density and three demographic correlates - income, race and ethnicity - at the tract level of analysis for one state in the north-eastern USA. Data for 1938 residential census tracts in the state of New Jersey were derived from 2004 licences for 13,984 tobacco-selling retail outlets. Demographic variables were based on 2000 census data. When applying a regression model, the residuals of an ordinary least squared (OLS) estimation were found to exhibit strong spatial autocorrelation, which indicates that the estimates from the OLS model are biased and inferences based on the estimates might be misleading. A spatial lag model was employed to incorporate the potential spatial effects explicitly. Results: Agreeing with the OLS residual autocorrelation test, the spatial lag model yields a significant coefficient of the added spatial effect, and fits the data better than the OLS model. In addition, the residuals of the spatial regression model are no longer autocorrelated, which indicates that the analysis produces more reliable results. More importantly, the spatial regression results indicate that tobacco companies attempt to promote physical availability of tobacco products to geographic areas with disadvantageous socio-economic status. In New Jersey, the percentage of Hispanics seems to be the dominant demographic factor associated with tobacco outlet distribution, followed by median household income and percentage of African Americans. Conclusion: This research applied a spatial analytical approach to assess the association between tobacco outlet density and sociodemographic characteristics in New Jersey at the census tract level. The findings support the common wisdom in the public health research domain that tobacco outlets are more densely distributed in socio-economically disadvantaged areas. However, incorporating the spatial effects explicitly in the analysis provides less biased and more reliable results than traditional methods.",
keywords = "Demographic factors, New Jersey, Spatial regression, Tobacco outlet density",
author = "Danlin Yu and Peterson, {N. A.} and Sheffer, {M. A.} and Robert Reid and Schnieder, {J. E.}",
year = "2010",
month = "7",
day = "1",
doi = "10.1016/j.puhe.2010.03.024",
language = "English",
volume = "124",
pages = "412--416",
journal = "Public Health",
issn = "0033-3506",
publisher = "Elsevier",
number = "7",

}

Tobacco outlet density and demographics : Analysing the relationships with a spatial regression approach. / Yu, Danlin; Peterson, N. A.; Sheffer, M. A.; Reid, Robert; Schnieder, J. E.

In: Public Health, Vol. 124, No. 7, 01.07.2010, p. 412-416.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Tobacco outlet density and demographics

T2 - Analysing the relationships with a spatial regression approach

AU - Yu, Danlin

AU - Peterson, N. A.

AU - Sheffer, M. A.

AU - Reid, Robert

AU - Schnieder, J. E.

PY - 2010/7/1

Y1 - 2010/7/1

N2 - Objective: Studies of relationships between tobacco sales and socio-economic/sociodemographic characteristics are well documented. However, when analysing the data that are collected on geographic areas, the spatial effects are seldom considered, which could lead to potential misleading analytical results. This study addresses this concern by applying the spatial analysis method in studying how socio-economic factors and tobacco outlet density are related in New Jersey, USA. Study design: A spatial regression method applied to tobacco outlet and socio-economic data obtained in 2004 in New Jersey, USA. Method: This study assessed the association between tobacco outlet density and three demographic correlates - income, race and ethnicity - at the tract level of analysis for one state in the north-eastern USA. Data for 1938 residential census tracts in the state of New Jersey were derived from 2004 licences for 13,984 tobacco-selling retail outlets. Demographic variables were based on 2000 census data. When applying a regression model, the residuals of an ordinary least squared (OLS) estimation were found to exhibit strong spatial autocorrelation, which indicates that the estimates from the OLS model are biased and inferences based on the estimates might be misleading. A spatial lag model was employed to incorporate the potential spatial effects explicitly. Results: Agreeing with the OLS residual autocorrelation test, the spatial lag model yields a significant coefficient of the added spatial effect, and fits the data better than the OLS model. In addition, the residuals of the spatial regression model are no longer autocorrelated, which indicates that the analysis produces more reliable results. More importantly, the spatial regression results indicate that tobacco companies attempt to promote physical availability of tobacco products to geographic areas with disadvantageous socio-economic status. In New Jersey, the percentage of Hispanics seems to be the dominant demographic factor associated with tobacco outlet distribution, followed by median household income and percentage of African Americans. Conclusion: This research applied a spatial analytical approach to assess the association between tobacco outlet density and sociodemographic characteristics in New Jersey at the census tract level. The findings support the common wisdom in the public health research domain that tobacco outlets are more densely distributed in socio-economically disadvantaged areas. However, incorporating the spatial effects explicitly in the analysis provides less biased and more reliable results than traditional methods.

AB - Objective: Studies of relationships between tobacco sales and socio-economic/sociodemographic characteristics are well documented. However, when analysing the data that are collected on geographic areas, the spatial effects are seldom considered, which could lead to potential misleading analytical results. This study addresses this concern by applying the spatial analysis method in studying how socio-economic factors and tobacco outlet density are related in New Jersey, USA. Study design: A spatial regression method applied to tobacco outlet and socio-economic data obtained in 2004 in New Jersey, USA. Method: This study assessed the association between tobacco outlet density and three demographic correlates - income, race and ethnicity - at the tract level of analysis for one state in the north-eastern USA. Data for 1938 residential census tracts in the state of New Jersey were derived from 2004 licences for 13,984 tobacco-selling retail outlets. Demographic variables were based on 2000 census data. When applying a regression model, the residuals of an ordinary least squared (OLS) estimation were found to exhibit strong spatial autocorrelation, which indicates that the estimates from the OLS model are biased and inferences based on the estimates might be misleading. A spatial lag model was employed to incorporate the potential spatial effects explicitly. Results: Agreeing with the OLS residual autocorrelation test, the spatial lag model yields a significant coefficient of the added spatial effect, and fits the data better than the OLS model. In addition, the residuals of the spatial regression model are no longer autocorrelated, which indicates that the analysis produces more reliable results. More importantly, the spatial regression results indicate that tobacco companies attempt to promote physical availability of tobacco products to geographic areas with disadvantageous socio-economic status. In New Jersey, the percentage of Hispanics seems to be the dominant demographic factor associated with tobacco outlet distribution, followed by median household income and percentage of African Americans. Conclusion: This research applied a spatial analytical approach to assess the association between tobacco outlet density and sociodemographic characteristics in New Jersey at the census tract level. The findings support the common wisdom in the public health research domain that tobacco outlets are more densely distributed in socio-economically disadvantaged areas. However, incorporating the spatial effects explicitly in the analysis provides less biased and more reliable results than traditional methods.

KW - Demographic factors

KW - New Jersey

KW - Spatial regression

KW - Tobacco outlet density

UR - http://www.scopus.com/inward/record.url?scp=77954177845&partnerID=8YFLogxK

U2 - 10.1016/j.puhe.2010.03.024

DO - 10.1016/j.puhe.2010.03.024

M3 - Article

VL - 124

SP - 412

EP - 416

JO - Public Health

JF - Public Health

SN - 0033-3506

IS - 7

ER -