OBJECTIVES AND METHODS

This study assessed the types and percentage of retail businesses in demographically and economically diverse business districts in the Boston area that sell tobacco products; the amount and characteristics of tobacco point-of-sale advertising in those districts; and any evidence of target marketing to specific ethnic groups.

Businesses open to the public were observed to enumerate their tobacco sales and point-of-sale advertising. The study was carried out within two Latino enclaves and a predominantly African American area, and in predominantly white, non-Hispanic urban and suburban areas of greater affluence. In addition, owners and managers of businesses selling tobacco in Latino business enclaves were interviewed.

Because our original objective was to learn about tobacco in the Latino enclave economy, we first selected the largest Latino enclave in Boston, in the Jamaica Plain neighbourhood; and the north side of Lawrence, Massachusetts, the largest predominantly Latino community in Massachusetts. We identified census tracts of similar economic status but differing ethnic composition, then selected at random a predominantly African American district in North Dorchester. For contrast, we purposefully selected Newbury Street in Boston's Back Bay, an affluent neighbourhood with shops and nightlife that attract visitors.

Because the Latino enclave in North Lawrence is very large, four census tracts with Latino population greater than 50% were selected at random, plus two comparison tracts selected at random from predominantly white non-Hispanic South Lawrence.

We conducted a second round of observations in the summer of 1999, in most areas covered in year 1 (carried out in the summer of 1998), and in additional commercial districts in five comparatively affluent, predominantly white, non-Hispanic exurban and suburban communities, chosen purposefully. We did not repeat observations in part of South Lawrence, because of hostility from proprietors in that area, nor in Newbury Street as less than 4% of establishments there sold tobacco and most had no exterior advertisements. In neighbourhoods observed twice, three shops selling tobacco closed and two new ones were established, and a restaurant in North Lawrence removed a vending machine; otherwise the total number of establishments and the number selling tobacco products remained the same over the two years. In addition, 94% of establishments were consistent in the presence or absence of exterior advertising from year to year (χ² = 97, p < 0.0005) and the number of changes in each direction was equal (four each). Consequently, in the results section we report year 2 observations combined with data from those neighbourhoods observed only in year 1, in order to achieve the largest possible sample of neighbourhoods.

We used field tested forms to record information about the establishment, exterior and interior advertisements and promotional displays, including location, language, ethnicity of model, if any, and brand of tobacco product, plus other information not reported here.

We developed a merchant interview in English, which was translated into Spanish, and translated back by a different individual. Discrepancies were resolved, and the questionnaire was field tested. We endeavoured to interview proprietors or managers of all establishments selling tobacco in the Jamaica Plain district and a random sample of those in the North Lawrence district in 1999.

Data collectors underwent classroom training, practice observations and interviews, and validation of the various kinds of observations. Only observers who demonstrated high reliability continued on the study. Observers carried maps showing the territory they were to cover.

All private businesses open to the public were observed and included in the database, including retail stores and services such as physicians' offices, dentists, and travel agencies. All restaurants, bars, department stores, pharmacies, liquor stores, grocery stores, and convenience stores or bodegas (small grocery and variety stores) were entered to determine whether tobacco was sold and to record any interior advertising. Businesses not expected to sell tobacco products, such as clothing stores, were noted. When in doubt, to confirm the absence of tobacco sales, observers entered the premises.

Analyses were conducted using SPSS. Files were structured with the individual advertisement, the establishment or the neighbourhood as the unit of analysis. Information about per capita income for the corresponding tracts from the 1990 census was added to the neighbourhood level file.

RESULTS

Per capita income of neighbourhoods ranged from $7620/year (North Lawrence) to $46 490 (Newbury Street). The other predominantly minority districts (Jamaica Plain and North Dorchester) had per capita income under $10 000. The suburban districts had per capita income ranging from $21 039 to $28 025. South Lawrence was intermediate at $13 937.

The number of storefronts in each business district, and the density of store spaces, are elements of the built environment that depend considerably on historical accident. However, the nature of businesses occupying the available spaces responds to economic forces. We found that the proportion of businesses in a district selling tobacco products was negatively associated with per capita income (table 1).

The correlation between per capita income in a district, and the percentage of establishments selling tobacco in year 2, plus the year 1 observations in Newbury Street and South Lawrence, by Spearman's ρ, was −0.794 (p = 0.006). The correlations for year 1 and year 2 observations taken separately were also significant (ρ = −0.9, p = 0.037, and ρ = −0.7, p = 0.036, respectively).

Businesses that commonly sold tobacco were convenience stores/bodegas, gas stations, pharmacies, and liquor stores. The convenience store/bodega was the most frequent source of tobacco products, particularly in the minority neighbourhoods. In Jamaica Plain, 88% of establishments that sold tobacco were this type of business; in North Lawrence, 71%, and in North Dorchester, 75%. Convenience stores were 50% or less of stores selling tobacco in the remaining neighbourhoods. Convenience stores/bodegas also constituted a higher percentage of all business in the minority communities than in the more affluent areas (table 1).

Combining the year 2 observations with year 1 observations in Newbury Street and South Lawrence, three (1.8%) of the 161 interior ads with a human image featured an evidently Latino model. There were no evidently Latino models in exterior advertisements. There were seven African American models in exterior ads (5.8% of ads with models) and 16 in interior ads (10.0%), scattered among neighbourhoods.

Advertisements for mentholated brands were heavily concentrated in Jamaica Plain, North and South Lawrence, and particularly North Dorchester (where 32.3% of all ads, both interior and exterior, were for mentholated brands). Overall, in the predominantly minority neighbourhoods, 29% of advertisements were for these mentholated brands versus 10% in the non-minority neighbourhoods (χ² = 67.2, p < 0.0005).

Brookline Village had the highest number of exterior advertisements per store selling tobacco, 7.7. This was explained by just two establishments, with 20 and 22 exterior ads respectively. The minority neighbourhoods otherwise tended to have more exterior ads per store selling tobacco than the non-minority neighbourhoods (5.51 v 4.41). For year 2 alone, this difference was significant by t test, but not by non-parametric tests. However, including the year 1 data from Newbury Street and South Lawrence, despite the Brookline Village anomaly, the minority neighbourhoods tended to have more exterior ads per store selling tobacco (U = 1773.5, p = 0.017). In every comparison, in contrast, the non-minority neighbourhoods tended to have more interior ads per store that sold tobacco, although this difference for the combined years only approached significance. Because a higher percentage of stores in the minority neighbourhoods sell tobacco, the number of advertisements divided by the total number of business establishments was much higher in the minority and low income neighbourhoods. Exterior advertisements per store, including establishments that did not sell tobacco, were 0.95 in minority neighbourhoods in the combined data set versus 0.15 in the non-minority neighbourhoods (U = 155973.5, p < 0.0005).

Interviews were successfully carried out in 72.2% (n = 13) of all stores selling tobacco in Jamaica Plain and with all of the random sample of 12 establishments in North Lawrence (100% response). Twenty interviews were conducted in Spanish, and 23 of 25 respondents were Latino, principally Dominican.

Excluding one who answered “one million dollars”, Latino respondents said they would require an average of $31 678/year to give up tobacco sales. Many respondents said they did not know, or refused to answer, questions about required compensation to give up tobacco promotional allowances. Mean required compensation to give up tobacco advertising was $1280 for Latino managed businesses (n = 10); mean compensation to give up slotting allowances and product discounts was $1354 (n = 7). One merchant reported receiving money in exchange for tobacco advertising, but otherwise promotional allowances were in the form of product discounts and free merchandise.

DISCUSSION

Storefront advertising of tobacco products is more prevalent in low income communities than in more affluent neighbourhoods, as a higher proportion of businesses in the low income communities sell tobacco products. The tendency for stores in non-minority neighbourhoods to have more interior tobacco advertising may be explained in part simply by their typically larger size.

Mentholated brands were marketed most heavily in the predominantly African American neighbourhood, and disproportionately in the Latino neighbourhoods. Although we observed relatively few Spanish language ads or ads featuring evidently Latino models, Latino grocery and variety store owners are heavily dependent on tobacco sales.

It costs tobacco distributors little to obtain a high prevalence of tobacco advertising in the Latino neighbourhoods. In exchange for about $2600 annually in product discounts and branded premiums or store supplies, the companies fill store windows with their advertising.

A limitation of this study is its confinement to the Boston metropolitan area. That similar patterns exist elsewhere in the USA seems plausible, but is unconfirmed. The sample of neighbourhoods is small, so the findings should be viewed with caution. Future research may establish the generalisability of these findings to other regions.