Advertising can influence food consumption by directly promoting food purchasing and indirectly by influencing social norms. Food advertising is associated with obesity in children
] and marketing more generally has been identified as one of the major factors contributing to the obesity epidemic
Our research confirms the observations of previous studies, which documented that unhealthy food advertisements are more prevalent in low income and minority communities
]. We also found different amounts of advertising in the two cities we surveyed. This higher number of ads, but lower percentage of food ads in Los Angeles, likely reflects the city’s focus on entertainment, as documented in another study
Most research on the effect of outdoor advertising has focused on tobacco and alcohol. For instance, many studies have shown that African-American neighborhoods (regardless of community income) have more alcohol and tobacco outdoor ads than White areas
]. In a study similar to ours, researchers found that exposure to outdoor alcohol advertising was associated with problem drinking in women
Most of the other evidence around the effect of advertising has looked at children, who are likely more susceptible to advertising
]. Alcohol advertisements have been found to cluster around schools and predicted alcohol consumption intentions by children at those schools
]. Children exposed to tobacco advertising are more likely to smoke
]. Other research has shown that unhealthy ads (including food, tobacco, and alcohol) have clustered around child-serving institutions
]. In the area of obesity, many studies have documented an association between various types of advertising and consumption of unhealthy foods and obesity
]. One other study looked at restaurant and food store signage and its relationship to obesity in adults living in the surrounding area
]. The authors found that while advertisements for unhealthy foods around restaurants were associated with obesity, signage around convenience stores was not associated with obesity.
Thus, the summary of research in other areas points to an effect of outdoor advertising on the intentions of the viewers of those ads. This analysis finds parallel results to the previous research on alcohol, tobacco, and food: those who live in areas with higher percentages of food advertising have greater odds of obesity than those living in areas with a lower percentage of food ads. While we controlled for several factors that were likely to confound the association, there may be other unmeasured variables (e.g. individual preferences, urbanization of census tract, roadway structure, exposure to television advertisements) that could explain the association.
Our study has limitations. We used self-reported information on height, weight, and soda drinking. Self-reports of dietary practices are always subject to recall bias, but 24-hour recalls seem to be better than longer-term food frequency questionnaires
]. BMI data are likely under-reported by about 1 unit (kg/m2
], which could have shifted the BMI distribution in our study downward. If respondents in our study reported a lower BMI than was true, we would be less likely to find an association between obesity and advertisements. Thus, the actual association may be stronger than the one we report.
Our data is also limited to two metropolitan areas. Further studies should try to replicate these results in different geographic settings. The survey data were collected by phone surveys. While the response rates were similar to those in another large national survey, our results could be biased by lack of response from those without landlines. However, one study found that the odds of obesity were similar in those with landlines and wireless phones
]. While those without any phone service may have a higher odds of obesity than those with a landline, the proportion of the population in this category is less than 2%
]. If those without phone service have a higher prevalence of obesity and are likely to live in areas where minorities predominate, and thus advertisements are prevalent, our effect estimates are likely to be biased towards the null hypothesis.
Furthermore, research on outdoor advertising is limited by the difficulty in determining who is actually exposed to the advertisements over a particular time period. We assumed that people were exposed to advertisements in their own neighborhoods. However, people may also be exposed to advertisements on their way to work and other destinations; advertisements and exposures may also change over time. Our findings indicate that the exposure to food ads around a person’s home does measure some aspect of the food environment that is associated with obesity. Future studies taking into account work location, time spent at work/home, and means of transportation to work may find a stronger association. Additionally, future studies could use actual locations of respondents to evaluate whether distance to advertisements matters.
Our independent variable, percent of food advertisements devoted to food, has some limitations. In census tracts where the total number of ads is small, but all (100%) of the ads are for food, we may have overestimated the impact of food ads. However, we controlled for the total number of ads in a tract, which may also be a measure of census tract geographic size. Even if there are only two ads in a geographically small census tract, they could be highly impactful in promoting consumption.
When looking at associations between individuals and their neighborhood’s characteristics, there is always a possibility of self-selection bias. People may deliberately move into neighborhoods in which particular foods are advertised. However, this is unlikely to be the dominant reason for selection of a neighborhood. Additionally, evidence from research on physical activity and the built environment suggests that self-selection has a weak influence on associations with the built environment