A major reason for the absence of studies on alcohol brand use among youth is the lack of an established methodology to collect such data. There are hundreds of major alcohol brands so researchers have assumed that it would take too long to collect such data, making brand research costly and impracticable. Yet, by using a combination of carefully crafted skip patterns, piping questions (using responses from previous questions on brand use to elicit more detailed information on alcohol consumption patterns for the identified brands), and internet forms that include lists of brands with check boxes, we developed a survey instrument that assesses alcohol brand preferences within a reasonable time frame.
This pilot study was designed to determine our ability to use a pre-existing internet panel to administer the survey to a national sample of underage youth and obtain a valid assessment of their alcohol consumption.
To conduct our survey, we utilized a pre-recruited internet panel developed by Knowledge Networks, the only U.S. company that maintains an internet panel (the Knowledge Panel®) that was created using a national probability sample. The company recruited households to its Knowledge Panel® sample through a combination of random digit dialing (RDD) and address-based sampling (ABS), which involves probability sampling of addresses from the U.S. Postal Service’s Delivery Sequence File [16
The Knowledge Networks internet youth panel provides high survey completion rates because of the ongoing relationship between the youth and the panel staff. To ensure adequate representation of panelists across race/ethnicity, telephone numbers from phone banks with higher concentrations of Blacks and Hispanics are over-sampled. To ensure adequate participation across levels of socioeconomic status, subjects agreeing to participate in the panel who do not have internet access are given WebTV and internet access and training for free.
Previous research has validated the alcohol data derived from adults in the Knowledge Networks internet panel. Heeren et al. [17
] compared the results of an alcohol survey conducted through Knowledge Networks with results from the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC). Estimates of current drinking were similar to those from NESARC, demonstrating that the Knowledge Networks panel is a less expensive, viable alternative to telephone and in-person surveys for assessing drinking behavior.
Knowledge Networks recruited 108 youth ages 16–20 from its existing internet panel to participate in the study by sending an email invitation. The invitation did not disclose that the survey was related to alcohol consumption. Those who agreed to participate were provided a secure link to access the study site.
The initial screening question asked respondents to report on how many days out of the past 30 they had consumed at least one drink of alcohol. A drink was defined as a 12-ounce can or bottle of beer; a 5-ounce glass of wine or champagne; an 8.5-ounce flavored malt beverage; an 8-ounce alcohol energy drink; a 12-ounce wine cooler; 8.5 ounces of malt liquor; 1.5 ounces of liquor (spirits or hard alcohol), whether in a mixed drink or as a shot; and 2.5 ounces of cordials or liqueurs, whether in a mixed drink, a coffee drink, or consumed on their own.
Respondents who had consumed at least one drink of alcohol in the past 30 days were provided with an online consent form, which described the details of the study, risks and benefits, and the procedures in place to protect the confidentiality of their responses. Participants who provided informed consent completed the internet-based questionnaire, which ascertained the alcoholic beverage brands they consumed within the past 30 days. After completion of the survey, a $25 gift was credited to the panel member’s account. The protocol was approved by the Institutional Review Board of the Boston University Medical Center. A Certificate of Confidentiality was obtained from the National Institutes of Health to help protect the confidentiality of the panelists’ responses.
Because this was a pilot study, with funding for surveys of only about 100 subjects, we used a consecutive sampling process, enrolling the first 100 adolescents who responded to the email invitation and were found to be eligible after they completed the screening questionnaire. A total of 1,028 email invitations were sent out. It took just one week to recruit the desired sample. During that week, 360 respondents (35%) completed the screening questionnaire: 108 were qualified (i.e., had consumed at least one drink of alcohol in the past 30 days) and completed the survey. We exceeded the desired sample size of 100 because closing the survey does not throw out subjects who are already online.
The internet-based survey instrument was developed to assess brand-specific alcohol consumption among underage youth. The list of assessed alcohol brands was generated using two main sources. The first source was the complete list of alcohol brands measured by GfK Mediamark Research & Intelligence (GfKMRI) in its Survey of the Adult Consumer, a written survey of a representative sample of approximately 10,000 U.S. adults. GfK MRI’s survey, which ascertains the prevalence of use of various consumer products, inquires about the past six-month and past seven-day consumption of 90 beer brands and 81 wine brands and the past six-month and past 30-day consumption of 17 flavored alcoholic beverage brands and 132 spirits brands.
The second source was a list generated by TNS Media Intelligence, which is an advertising industry standard source that monitors advertising occurrences and expenditures in more than 300 national periodicals. The survey instrument included every alcohol brand advertised in the magazines during any of the years 2001–2006, except for wine and champagne. Due to the extensive number of advertised wine and champagne brands, it was sufficient and more practical to include only brands advertised in 2006.
Alcoholic energy drinks are not included in the GfK MRI list. To identify these brands, we performed internet searches and used an extensive list compiled by the National Association of Attorneys General as part of an ongoing investigation into the marketing of these beverages. Finally, we included all alcohol brands reported by participants in the preliminary pilot study that were not on our initial list.
The final survey instrument included 61 brands of beer, 81 brands of wine or champagne, 19 brands of flavored alcoholic beverages (including flavored malt beverages, alcopops, wine coolers, and malt liquor), 35 types of mixed drinks, 38 brands of alcoholic energy drinks, 14 brands of bourbon, 3 brands of brandy, 8 brands of cognac, 9 brands of gin, 19 brands of rum, 15 brands of scotch, 13 brands of tequila, 30 brands of vodka, 8 brands of whiskey, and 27 brands of cordials and liqueurs. In total, the instrument assessed 380 brands of alcoholic beverages. Brand extensions were ascertained as a single brand category if they are typically advertised together (e.g., Absolut flavored vodkas); they were ascertained separately if they are typically advertised separately (e.g., Budweiser, Bud Light).
For each category of alcohol, the respondents checked off which specific brands they had consumed during the past 30 days. If a specific brand was not listed, then respondents entered the name, giving as specific a name as possible. After identifying the brands they had consumed in the past 30 days, the respondents reported the number of days during the past 30 that they had consumed each brand and how many drinks of each brand they usually had on a day when they drank that brand.
Respondents also reported on how many days out of the past 30 they had consumed five or more drinks in a row (that is, within a couple of hours). Next, for each category of alcohol, they selected all of the brands, either alone or in combination, which they remembered drinking in the past 30 days on those occasions when they drank at that level.
The two main data analysis questions were as follows: (1) Of underage youth (ages 16–20) who consumed alcohol in the past 30 days, what proportion reported having consumed each type and each brand of alcohol? (2) Of underage youth who consumed five or more drinks in row, what proportion used each type and brand of alcohol during these drinking episodes?
Knowledge Networks applied statistical weighting adjustments to account for selection deviations and to render the sample representative of the underlying population. These weights accounted for the different selection probabilities associated with the RDD- and ABS-based samples, the oversampling of minority communities, non-response to panel recruitment, and panel attrition. Post-stratification adjustments were based on demographic distributions from the Current Population Survey (CPS) conducted by the U.S. Bureau of the Census.
Validation of Survey Findings
Since there has been no previously published survey of youth alcohol brand preferences, there is no true gold standard to which we can compare our findings. Instead, we can validate our findings against GfK MRI’s Survey of the Adult Consumer, a written survey of a representative sample of approximately 10,000 U.S. adults used to ascertain the prevalence of use of various consumer products, including past 30-day consumption of 132 spirits brands. Since GfK MRI defines adults as age 18 years or older, their findings on types of alcohol used can be applied as a validity check on our own findings for the 18- to 20-year-olds in our internet panel. We previously obtained data on past 30-day use of various types of liquor among 18- to 20-year-olds from the 2007 Survey of the American Consumer [18
], and we used these data here to validate our survey findings.