While target market segmentation is one of the most fundamental and celebrated concepts in marketing, targeting at-risk ethnic minorities with products considered harmful is simultaneously viewed as unethical, bordering on criminal. At the same time, it appears that minority wine consumers may feel ignored by wine marketers, distributors, and producers, despite the fact that ethnic minorities (African Americans, along with Asians and Hispanics) are the fastest-growing market segment in the United States. We characterize this phenomenon as a case of the market segment ‘peering through the looking glass (a reference to Alice in the classic tale Alice's Adventures in Wonderland, and her desire to be on the other side of a mirror or looking glass)’. This case reports on the use of focus groups to investigate our hunch that minority populations, specifically those of African descent, found few messages or images that spoke to them. Our results suggest a perceived dearth of imaging depicting or seeming to relate to African American populations. We also discuss our strategy of applying the results of the focus groups to develop a survey instrument for a larger, quantitative study in the future.
After reading this case study, students should be able to
- Discuss why qualitative research might sometimes be used to launch quantitative research (mixed methodology)
- Describe and undertake content analysis
- Apply recruiting techniques to effectively engage minority respondents and participants for qualitative research
- Recognize how to use the scientific method to follow up on anecdotal evidence
Project Overview and Context
Marketing Wine to Ethnic Consumers
Although wine marketing has traditionally focused on the ‘typical’ US wine consumer—generally identified as White, middle-aged, and well educated—this characterization is no longer accurate because there is no typical wine consumer (Campbell, 2009). As a result of changing US demographics (specifically, the rise of the multicultural majority–minority as well as the aging of the population due to increased longevity), consumer preferences and buying motivations are being reexamined across a variety of product and service categories. As we began examining the consumer preferences of minority groups in the United States, it appeared that the needs of the growing Hispanic population were being addressed. However, we found no prior research that concentrated on the needs and desires of the increasing population of African American wine consumers.
Lack of Qualitative Studies in Wine Marketing Research
Wine research, particularly wine marketing research, in peer-reviewed journals is largely based on quantitative survey research. For example, British, Australian, and Hispanic wine drinkers, consumers from New Zealand, and generation Y and millennial consumers all have been studied using predominantly quantitative and occasionally qualitative methodologies. Again, since scarce literature was available regarding minority wine consumers, it seemed to us that focus groups would be the best way to address unasked research questions because of their exploratory nature.
Our project was designed and presented at a conference where the second researcher (Sandra) took notice and expressed interest in the project. Nuances of the qualitative nature of this project warranted the combined efforts of both researchers. The initial researcher (Rhonda) is a non-Hispanic Caucasian; the second researcher, Sandra, is African American. Rhonda has previous wine marketing research experience, while Sandra has professional experience and training in conducting focus groups for market research and advertising firms. Their shared interest in wine, combined with their various strengths, established a solid foundation for securing participants in the focus groups. Based on the review of the transcripts, having an experienced moderator who belonged to the target African American population appeared to make the participants more comfortable in their responses.
Timeline and Discussion Guide
First, a timeline for the research project was developed (Figure 1). Soon thereafter, the discussion guide was developed. Given that we had no literature to guide us, we had begun an outline of the discussion guide from the very inception of our research planning. We knew that we would have to prompt basic conversations regarding the preferences, needs, and desires of the participants. We built the guide by taking what was known about a different minority group (Hispanics) and reformulating some of those same questions for our participants. After we were fairly confident that we had formulated enough questions to create a baseline for this group—but not so many questions that participants couldn't respond within 2 hours—we distributed the guide for review by experts (wine industry professionals and educators knowledgeable in the field). We selected individuals with demographic characteristics similar to those of the target population, but not people who would ultimately participate in the formal focus groups. Based on the feedback we received, the discussion guide underwent minor adjustments—we deleted some questions, added constructs, and changed phrases—before implementation.
Figure 1. The research timeline.
The discussion guide was divided into three sections based on our study objectives. The first section focused on gaining insights into the participants' attitudes, opinions, and interests pertaining to wine consumption. The second section aimed at collecting information regarding factors that influenced wine consumption and purchases. The third section identified strategies to generate recommendations to industry players to reach African American wine consumers. As in other qualitative studies that include a quantitative component, each sub-group in this research also completed a brief questionnaire regarding demographic information, wine consumption preferences, and wine purchase behaviors to produce a thorough profile of the sampled population.
Focus Group Execution
Focus groups can be held in a variety of appropriate locations, such as a facility expressly devoted to conducting interviews and focus groups. These are often equipped with audiovisual equipment to record sessions and one-way mirrors through which participants can be viewed by the client or researchers in another room. Other settings include libraries, hotel rooms, client offices, and even the moderator's home. For our research study, three focus group sessions were conducted at a professional focus group venue. The facilities included a waiting room set up with beverages and snacks supported by hostesses, as well as equipment operators for technological assistance.
Our three focus group sessions were held in March 2013. A discussion guide organized by study objectives ensured a practical structure for introducing topics during the discussion. It also helped clarify group comparisons during the data analysis phase (Krueger & Casey, 2009). Discussion guide topics were developed by combing the literature, as meager as it was, and exploring what was unknown about this ethnic group, primarily:
- attitudes, opinions, and interests of African Americans pertaining to wine consumption
- factors that influence wine purchases
- strategies that might be implemented by the industry to reach African American wine consumers
No actual wine was consumed or offered during the focus group sessions.
Each focus group numbered from 6 to 12 participants; each cohort was chosen based on hypotheses that gender and age might yield nuanced responses. The first group was all male (n = 6), the second group was mixed-gender (n = 12), and the third group was all female (n = 11)—for a total of 29 participants. This study, like most focus group research, took a purposive sampling approach in order to match a representative sample as closely as possible (Wimmer & Dominick, 2006). However, recruiting in minority populations can prove to be challenging for a variety of reasons. First, researchers who are not a member of the particular minority may be apprehensive, or even fearful, of the group due to stereotypes and prejudices (Sheikh et al., 2009). Also, depending on the geographical layout of the region, facilities may not be within easy reach (such as on a major bus route) or readily accessible (in an unfamiliar part of town). For this study, additional participants were identified by a method known as snowball sampling. Initial recruits supplied names and contact information for people they knew who fit the study requirements.
In order to qualify for the study, participants had to
- have reached the legal drinking age of 21 years,
- self-identify as Black or African American,
- have wine consumption/purchase experience,
- provide informed consent.
Focus group participants were provided a US$75 stipend to compensate for their time and travel expenses, which also motivated respondents to participate.
The three focus groups were conducted in Columbus, Ohio, commonly identified as Middle America or the Crossroads of America, with a reputation as the representative city of US demographics (Brandau, 2011). The sessions were audiotaped with the respondents' written consent and professionally transcribed verbatim for the purpose of content analysis. Sandra used her experience as a trained moderator, skilled in group dynamics and leading focus groups, to conduct the discussions and debriefing sessions. As moderator, she ensured that each member of the group took part. She also assisted with data and content analysis, paying special attention to responses from participants who might jeopardize internal validity.
A content analysis approach guided the data analysis portion of the research. Krueger and Casey's (2009) procedure for debriefing questions was explicitly followed so that the researchers could minimize reliability threats by adhering to a recognized system. The initial process involved recognizing common themes, important points, anything surprising and/or unexpected, helpful quotes, and similarities and differences between or among groups.
Following the initial theme identification, which each researcher chronicled independently, investigators worked with individual paper copies of the transcript. Each line of the transcript was numbered, and each group's verbatim responses were printed on different-colored 8.5 × 11-inch paper: males were printed on blue paper; mixed group were printed on tan paper; and females were printed on pink paper. Materials used in this stage also included scissors, colored marking pens, flipchart paper, and tape (Krueger & Casey, 2009).
The next step was for each researcher to independently read and analyze all the transcripts. Flipchart paper was hung on the walls of the room (a hotel room for the first session and a university conference room for subsequent sessions). On the top of each page of the flipchart was a question from the focus group; all groups were asked the same questions. Taking the transcript in hand, each investigator independently answered the following questions:
- did the participant answer the question asked? If Yes—go to Q3. Don't Know—set aside for later review. If No—go to Q2.
- does the comment answer a different question in the focus group? If Yes—go to Q3. If No—goes in the discard pile (but isn't thrown away just yet).
- does the comment say something of importance about the topic? If Yes—tape it to the flipchart under the appropriate question. If No—goes in the discard pile.
- is it like something that has been said earlier? If Yes—start grouping similar quotes together. Now you are basically making piles/categories of similar things. If No—start a separate pile. You are constantly comparing and making decisions. (Krueger & Casey, 2009, pp. 119–122).
It did not take long for the flipchart pages to fill up and for themes to become apparent. The researchers grouped together quotes and comments that said the same thing, which led to several rearrangements because some comments were appropriate for more than one section of the discussion.
Some questions elicited a plethora of comments and suggested more than a single conclusion. For these questions, sub-categories were established independently by each researcher without any verbal exchange. The researchers could then continue to the next activity either working in isolation or sharing their full flipcharts. The next stage was to condense comments and quotes and write a descriptive summary for each question. Not every comment, quote, or idea made it to the flipcharts, as focus group participants often go ‘off script’ until the moderator brings them back to the topic at hand.
By the time the initial theme identification had been completed, other ideas had emerged for carrying out this process in future studies. For example, to facilitate traveling with the materials, the process could be carried out using word processing documents (Microsoft Word, for example) and color coding the comments to reflect individual themes.
At this stage, flipchart sheets were shared and comparisons/contrasts were documented across groups for each question. Points of similarity and disagreement all were duly noted. Although no actual interpretation takes place at this stage—findings are simply recorded—we did take note of the weight given to themes, comments, and quotes. This can be done by observing the frequency of comments (even though the mere number of mentions is not necessarily most important). On the other hand, specific comments that provide more detail, as well as ones that express emotion, may be weighted more heavily. Finally, we considered extensiveness—that is, how many people commented on a particular point (as opposed to the number of times the point was made).
Once the above process was complete, the researchers shared findings with each other and began the interpretation process. We did this by identifying key concepts and their meanings, grouping data with the same or similar meaning on a similar dimension. Sometimes the dimension we identified was completely new, having surfaced during the group discussion.
Inter-Rater Reliability and Other Threats
Inter-rater reliability tests are used to assess the degree to which different raters/observers give consistent assessments of the same phenomenon. This is always a concern in qualitative research, and researchers must pay attention to the inherent biases they bring to the analysis. Aware of an apparent predisposition to distrust content analysis in some academic circles, we wanted to be especially careful about how we managed our data and comments. We wanted our method to be as objective as possible in identifying the comments that would be emphasized and those that might be unauthentic. Calculations of high inter-rater reliability demonstrate consistency among the observations of comments, quotes, and themes that multiple coders provide.
Basically, there are three methods used to calculate inter-rater reliability: conventional, summative, and directed (Hsieh & Shannon, 2005). The conventional approach may be used in describing phenomena when existing theory or research is limited. In the summative approach, researchers identify and count particular words and phrases, not to imply importance but to gauge context. Directed analysis is often used when information that does exist would benefit from more robust and richer description. Because we were working with limited prior knowledge and were interested in understanding the context of our participants' comments, we used a combination of conventional and summative approaches. Summative approaches are often referenced by percent of agreement.
The percentage of agreement calculations may seem like a logical and straightforward basis for measuring inter-rater reliability. Researchers simply count the number of match-agreement pairs over the course of the observations; the method may be a little unsophisticated, but it is certainly easy to use. However, as more clearly explained by James Wood (2007), the downside of the percentage agreement approach is that it does not differentiate among random agreements (agreement by chance). Random agreements between raters may have no relationship but still reflect a high percentage of agreement, even though the agreements may have been based on completely different observations.
Cohen's kappa addresses this issue. Similar to Pearson's product–moment coefficient (Pearson's r), Cohen's kappa measures consistency among raters. A kappa of 1.0 indicates perfect agreement between two raters; conversely, a kappa of −1.0 indicates perfect disagreement. A kappa of 0 simply indicates a random level of either agreement or disagreement. In other words, kappa represents the level of correlation between the scores of two raters.
Although kappa can be calculated by hand fairly easily, SPSS was used to run kappa for this study. Most statisticians prefer kappa values to be at least .60, and often higher than .70, in order to claim a good level of agreement (Pallant, 2007; Wood, 2007).
We approached this study with some trepidation because we mistakenly believed that qualitative research is a long, tedious, and arduous process. Had there been an appropriate quantitative methodology to employ, we are likely to have adopted it out of sheer fear. However, preliminary readings clearly suggested that the best way to conduct this exploratory study was by means of interviews or focus groups. At first, this project was to be conducted by a single researcher who would conduct individual interviews. Talk about time consuming! One of the lessons of this project is the importance of networking with others to share research ideas and interests. That is what led to the collaborative effort and the switch to focus groups as a means of collecting data.
The old saying ‘Many hands make light work’ was certainly true of this undertaking. Collaborating with others enables researchers to combine resources and greatly increase the likelihood of achieving the goals of the study. For example, Sandra had prior experience as a professional moderator and contacts that helped in the recruitment effort. Rhonda had access to transcription services and funding—a seed grant through Kent State University—that supported the data collection process. Without these combined resources, it certainly would have taken much longer to gather the focus groups and collect the essential information. By being open to others and sharing ideas, skills and resources can be brought together much more efficiently. We don't pretend to know the average time it takes to complete qualitative studies of this nature, but we were surprised and pleased to complete our efforts in about a year.
Regardless of the constraints that may be associated with qualitative research, the richness of the results cannot be ignored. Ultimately, we intend to take what we learned from this research and develop a quantitative study on a much broader scale. Without the deeper responses and insights from the focus groups, we would not know where to begin. For instance, we learned that across race, class, age, and occasion, this sample viewed wine as socially acceptable. They tended to view as a common denominator, the beverage that brings people together. Comments such as these from focus group members amplified that sentiment:
It's a conversation starter.
[W]ine is non-intimidating … especially in a mixed group of company.
At the same time, wine is generally perceived as a drink for Caucasians. The focus groups in this study observed that wine marketing efforts have been aimed primarily at White consumers. Even on the service and sales side of the industry, no one could recall a time when they had been served wine by an African American sommelier, distributor, sales representative, wine tasting room employee, winemaker, or other industry professional. Because African Americans generally have a stronger ethnic identity than other groups (Ting-Toomey et al., 2000), it makes sense that wine marketers elicit positive responses from African American wine consumers who ‘see people who look like me’ in media advertising. The overall perception was articulated by one study participant as follows:
The other thing you don't see enough in wines [retailing] if you think about all the wine tastings you've been to over the years, it's a very white industry. You don't have that many black representatives that many black people pouring wine. Definitely not many wineries owned.
Many of these valuable insights may not have been obtained from research based on the original plan (individual interviews). These and other findings will drive future research in an effort to gain more generalizable data to help the industry better serve this market segment.
Exercises and Discussion Questions
- Why was the methodology of this study changed? What was it changed from? What was it changed to? Do you think the change necessary? Why or why not?
- In this case, we discuss some of the nuances associated with the qualitative nature of collecting data from a minority segment. What were some of the concerns that were described, and why was it necessary to address them?
- Given that there was little previous research to help develop the discussion guide, what do you think of the research questions that the investigators developed? What would you have asked?
- Focus group respondents in this case were given an honorarium for their participation. Was this incentive necessary? Why or why not?
- Based on the explanation of snowball recruiting tactics used in this case, what other recruitment tactics might you consider using?
- In reviewing the timeline, do you think the project was completed in a timely manner? Why or why not?
- What two conventions are associated with measuring the consistency of inter-rater reliability? Which one is the preferred measure? Why?