Context
This study used simulated customers inquiring about e-cigarettes at two types of shops (i.e., vape and tobacco) in three different cities in Orange County, CA to discover marketing claims made in different racial/ethnic communities. Rates of adult smoking in Orange County (10.8%) are below the Healthy People 2020 target and lower than the state adult smoking rate (11.6%) [
18]. Yet recent data show that current and ever electronic cigarette use in Orange County middle school and high school students during the 2013–2014 school year was higher than use of conventional cigarettes, suggesting that ENDS are increasing in popularity [
19].
Research assistants posed as shoppers, asked the salesperson how e-cigarettes compare to conventional cigarettes, and then recorded claims made by salespeople using a standardized form designed for this purpose. Informed consent was waived for simulated customer interactions. The use of deception in simulated customer studies allows for behaviors to be observed without changing the behavior because of the presence of an observer [
20]. The gains in study validity with the use of deception by simulated customers could not be maintained if informed consent from individual participants was sought. Modified informed consent procedures for store-level consent to be included in the study may also bias study results. Under opt-out consent procedures, the validity of consent is questionable because lack of refusal may not reflect actual consent on the part of ENDS retailers but lack of attention to informed consent notices. Thus, given the minimal risk posed to human subjects involved in simulated customer interactions, anonymous collection of data, and the challenge to scientific and consent validity of alternate procedures, a waiver for obtaining informed consent for simulated customer interactions was requested and granted by the IRB. The research protocol was approved by the California State University Fullerton Institutional Review Board (HSR 15_0072, February 8, 2015). Simulated customer events (
N = 68) took place from May 2015 to July 2015.
Sample
Shops. For the purposes of this study, e-cigarette or “vape” shops were defined as retail outlets that sell e-cigarettes, e-cigarette components, and/or e-cigarette liquids, exclusively. Tobacco shops were defined as retail outlets for which conventional tobacco products (e.g., cigarettes, chewing tobacco) make up at least 50% of all products sold, and that also sell some type of electronic nicotine delivery device.
Cities. In Orange County, CA, the two largest racial/ethnic minority groups are Latinos (34.2%) and Asians (18.9%), with Mexicans the largest Latino subgroup in the county (29.3%; 85.8% of Latinos) and Vietnamese (6.3%; 33.1% of Asians) and Koreans (3.0%; 15.8% of Asians) two of the largest Asian ethnic groups in the county [
21]. These three ethnic groups are important vulnerable populations for tobacco control; Koreans and Vietnamese smoke at higher rates than the general population and more than other Asian subgroups [
22], and Latinos are the largest minority group in California.
Three cities were selected for inclusion in the study using purposive sampling, a type of non-probability sampling in which decisions about the elements to be included in the sample are made by the researcher based on a variety of criteria pertinent to the given study [
23]. Specifically, the cities were chosen based on their historical and documented enclaves of Vietnamese, Korean, and Mexican populations [
24,
25] and the substantially larger proportion of business serving them than in the general population. Based on 2015 U.S. Census data [
21], 39.0% of the population in City A (174,721 total) was Asian, with 29.9% of the population identifying as Vietnamese and 2.8% as Korean. In City A, 36.7% of the population was Latino and 21% non-Hispanic White. Home to a two-mile stretch of Korean-owned businesses which attract Korean customers from the surrounding area [
24], 52% of businesses in City A were Asian-owned in 2012. In City B, 48.2% of the population (91,719 total) was Asian, with 40.3% identifying as Vietnamese; 23.2% were Latino and 24.5% non-Hispanic White. A Vietnamese commercial district is centrally located in City B, but spreads to surrounding communities [
24]. Of all businesses in City B, 54.2% were Asian-owned in 2012. In City C, 78.2% of the population (333,268 total) was Latino, with 72.6% identifying as Mexican; 10.6% were Asian and 9.2% non-Hispanic White. In City C, 31.8% of businesses were Hispanic-owned and 18.2% were Asian-owned in 2012. Thus, City A can be said to represent large Vietnamese and Korean populations; City B, a large Vietnamese population; and City C to represent a large Hispanic/Latino population.
Sampling Frame. To be included in this study, a retail outlet had to be either a vape or tobacco shop as defined above, and located within one of the three study cities. There is no definitive way of identifying ENDS retail outlets in the cities included in the sample. Thus, two methods were used to create a list of vape shops and tobacco shops located within the three study cities. First, Internet searches using the terms “vape shop,” “tobacco shop,” and “e-cigarettes” were conducted using Yelp and Google Maps websites, a method similar to online search strategies used in a previous study [
14]. Second, a windshield survey of the three communities was conducted to verify stores identified through Yelp and Google Maps, visually identify vape and tobacco shops which may not have Yelp reviews or be listed through Google Maps, and confirm which tobacco shops sold ENDS. This was especially important to identify small tobacco shops which may not have social media following. Based on an established windshield survey methodology [
26], researchers drove through each city to confirm whether stores identified through Yelp and Google Maps were still open and to identify vape or tobacco shops not found on Yelp or Google Maps. Once identified, researchers stopped at each business to determine whether the store met the inclusion criteria; if a store met the inclusion criteria, the store name, address, and whether it was a vape shop or a tobacco shop were recorded. Stores listed on Yelp or Google Maps that did not meet in the inclusion criteria or had closed down were not included in the sampling frame. The list of businesses created by the Internet search and windshield survey constituted the sampling frame; all shops identified through the two methods that met the inclusion criteria were included in the study.
Sample Description. A total of 68 retail outlets – 50 tobacco shops (74%) and 18 vape shops (26%) – were identified and included in the study. Of these, 44% were in City A (7 vape shops, 23 tobacco shops), 25% in City B (5 vape shops, 12 tobacco shops), and 31% in City C (6 vape shops, 15 tobacco shops). Just over half of the interactions (56%) involved female simulated customers, while 44% involved male simulated customers.
Measurement
Instrument. The data collection instrument used to document simulated shopping experiences was developed based on marketing claims identified from content analysis of e-cigarette retailer websites [
27] and Camel Snus magazine advertisements [
28]. Marketing claims about the benefits of smoking e-cigarettes were listed in a table on the data collection form which, after the simulated customer interaction, the research assistant used to record the claims made by salespeople. In addition, the data collection instrument included field notes, where the research assistant could record other notable features of the interaction.
Training. Simulated customers were one graduate and five undergraduate student research assistants, ages 19–23 years, who received formal training for the study. The training covered the purpose of the study, study methods, review of the data collection instrument, and role play. The forms were pretested in shops outside the study area prior to use as part of field training. Research assistants were sent into four vape and tobacco shops in non-study cities, where they practiced engaging in simulated customer interactions and recording marketing claims using the data collection form. A debriefing training was held after the field training to identify emergent questions and unforeseen issues, and to streamline the data collection protocol.
Variables. Thirteen different claims made by salespeople during simulated customer interactions were recorded using the data collection instrument. Claims were preselected for inclusion in the study based on previous studies which examined marketing claims made by e-cigarette manufacturers and retailers [
27,
28]. These included claims that e-cigarettes: 1) help one quit smoking, 2) are healthier, 3) can be used in more locations, 4) can be used anytime and anywhere, 5) do not generate second hand smoke, 6) are less expensive, 7) are friendlier to the environment, 8) are cleaner, 9) are more fire safe, 10) have no offensive odor, 11) are more socially acceptable, 12) are “cooler”, and 13) come in a wider variety of flavors, as compared to conventional cigarettes. In addition to these positive claims about the benefits of smoking e-cigarettes, corresponding negative claims also were recorded (i.e., e-cigarettes do
not help on quit smoking, are
not healthier, etc.). Thus, claims were recorded using 26 different variables, 13 representing positive claims and 13 representing corresponding negative claims. Simulated customers also maintained field notes on any discussions not captured by the thirteen different claims which occurred during simulated customer interactions or on the general nature of the interaction.
Scaling. The outcome for this study was the total number of positive claims about e-cigarettes made by the salesperson during the simulated shopping interaction. Each of the 13 positive claim variables was coded as “0” if the claim was not made, and “1” if the claim was made, during the interaction. A summative scale totaling the number of positive claims made was the outcome variable for this study, with a possible range of 0–13.
Procedure
Data collection. For each of the three cities, two research assistants, one male and one female, were assigned to serve as simulated customers. The research assistants were bilingual in Korean for City A, bilingual in Vietnamese for City B, and bilingual in Spanish for City C. Within each city, the simulated customers were randomly assigned to the retail outlets included in the sampling frame. Simulated customer interactions took place on weekdays during business hours, between 9:00 am and 8:00 pm.
For each simulated shopping interaction, the research assistant entered the retail outlet and announced to a salesperson that he or she was interested in learning more about e-cigarettes. After making this introductory statement, the research assistant then asked the specific question, “Can you tell me more about e-cigarettes?” After the salesperson responded, the research assistant followed up with the question, “What is the difference between e-cigarettes and conventional cigarettes?” Once the salesperson responded, the research assistant thanked the salesperson and exited the shop. The research assistant then completed the data collection sheet to document which of the 13 marketing claims had been made, and which, if any, had been countered, by the salesperson.
Analysis. Data analysis was conducted from January 2016 to March 2016. Univariate analysis included frequency, mean, and standard deviation. Bivariate relationships were tested using independent samples
t-tests and oneway analysis of variance (ANOVA). Post-hoc analysis was conducted using Tukey’s honest significance difference (HSD) test. Multivariate analysis consisted of multiple linear regression to examine factor
s related to the number of marketing claims made; predictors included type of shop and city. A preliminary automated quantitative content analysis [
29] of field notes from all simulated customer interactions was conducted using Atlas.ti 7.0 qualitative data analysis software [
30]. Field notes were copied verbatim into a digital format and imported into Atlas.ti. The Word Cruncher function in Atlas.ti was used to create a frequency list of words mentioned in field notes. Words were ordered based on the frequency of mentions; common words such as “the”, “to”, and “and” were removed from the analysis.