Case
Abstract
In this case study, we explore the challenges associated with researching a stigmatized topic (women’s experiences of smoking during pregnancy) and in doing so trace “behind-the-scenes” of our research journey. This case study is based on a collaboration between the two authors and formed part of the first author’s PhD research. We start by describing the project and how the problems of recruitment and data collection arose. We then focus on three methodological strategies that offered more creative options for us to fully explore women’s experiences of stigma: (1) moving to online spaces; (2) utilizing multiple sources of data for qualitative analysis; and (3) changing the frame of the study through discourse analysis. We show how together these strategies allowed us to consider how initial “problems” in the research (e.g., slow recruitment) could be seen as stigma in action.
Learning Outcomes
By the end of this case, students should be able to
- Identify the methodological challenges involved in doing research on stigmatized topics
- Understand the benefits of presenting your research at conferences and consulting with peers
- Explain the value of reflecting on the various stages of the research process and how this can enhance an analysis
- Assess the utility of discourse analysis and the opportunities offered by this approach
- Describe the value of including multiple sources (e.g., media accounts) of data in an analysis
This case study explores the challenges associated with researching a stigmatized topic: women’s experiences of smoking during pregnancy. In what follows, we trace “behind-the-scenes” of our research journey and outline three methodological strategies that enabled a creative and rich analysis of women’s experiences.
Project Overview and Context: Encountering Problems in Researching Women’s Smoking During Pregnancy
B.W. enrolled in a PhD with an interest in how people make sense of occupying contradictory social positions or engaging in contradictory behavior. Having never been pregnant or smoked cigarettes, her interest in smoking during pregnancy was largely reflective of her assumption that there was a contradiction associated with smoking during pregnancy and her further assumption that this contradiction might produce stigma. As a young, White, educated, heterosexual, able-bodied, and cis-gendered woman, B.W. started this project having to question the basis of her own anti-smoking views. Having a mother who has smoked on and off during her life (including through her pregnancy) and living in a country where anti-smoking views are widely accepted meant that negotiating her own views (and those of people who are important to her) about smoking was an ongoing task achieved through reflection and peer-consultation. Reflection can involve questioning the aims, values, assumptions, power relations, and theoretical position of the research and researcher(s). Peer-consultation can range from supervision with colleagues or mentors to presenting the research at conferences in order to seek feedback from others.
B.W. decided to focus her first PhD project on the ethics of stigmatizing women who smoke during pregnancy and the practical question of whether this stigmatization actually made it harder for those women to stop smoking. To foreground women’s views, B.W. made the decision to begin by interviewing women who smoked during their recent pregnancies. The aim of these interviews was to gain an understanding of women’s experiences of smoking during pregnancy, specifically whether, and in what ways, they experienced and responded to stigma.
Recruitment
Researching such a sensitive and morally charged topic meant that B.W. needed to balance a position that was neither pro- nor anti-smoking but that still conveyed her non-judgmental position and allowed women to feel safe and heard. To deal with this, and prepare for the interview study, B.W. designed recruitment posters that included an attractive and positive photograph of a pregnant woman alongside a statistic (“Did you know that 17% of women say they smoke during their pregnancy?”). Her intention was to normalize smoking during pregnancy and to convey her non-judgmental position. The poster also made it clear that women’s views and experiences were important to this study.
B.W. approached seven different women’s health and medical centers to help with advertising the interview study. Disappointingly, only three of these seven agreed to advertise the poster described above. The other three refused, and the fourth requested an additional ethics approval—although the study had already received ethics approval from the University. To diversify sampling, B.W. also approached four online pregnancy forums to advertise the interview study. However, only two of these forums agreed to advertise, with one forum explaining their refusal on the basis that smoking during pregnancy is a “very sensitive” topic that they “try to avoid” because “discussions inevitably deteriorate and become abusive.”
In addition to the difficulty in gaining support to advertise, recruitment for the interviews was notably slow. Indeed, over a period of 5 months of recruiting (April-August 2011), a total of 13 women volunteered to be interviewed. Two of these were face-to-face pilot interviews that were conducted with women known personally to B.W. and who had smoked during their recent pregnancies. The remaining interviews were made up of 10 telephone interviews and 1 email interview that transpired over the course of three consecutive days. It is worth mentioning here that women were offered the choice of either face-to-face or telephone interviews. Telephone interviews were introduced as a way of reducing the inconvenience associated with participating (i.e., many women have caring or paid-work responsibilities) and also to allow the recruitment of a geographically diverse sample. It is notable that all participants without previous connection to the interviewer chose to be interviewed via telephone or email. This may have usefully provided some distance for participants, given the nature of the topic.
Initial Challenges
Together, the impersonal medium of the interviews (mostly via telephone), the sensitive nature of the interview questions, and B.W.’s lack of experience in conducting interviews meant that B.W. found it difficult to draw participants out in the interviews and encourage them to articulate their own stories in detail. To some extent, B.W. felt she had failed to fully capture women’s experiences of smoking during pregnancy and collect in-depth accounts of this marginalized experience. Indeed, not only were women opting for more anonymous forms of engagement when given the option, but the interviews themselves were notably short, lasting between 10 and 20 min.
Despite these issues, B.W. and her primary PhD supervisor were able to analyze and write up the interview study (Wigginton & Lee, 2013). She applied thematic analysis drawing on Virginia Braun and Victoria Clarke’s (2006) six-step approach to answer the question, “How do women experience and respond to the stigma associated with smoking during pregnancy?” This analytical method allowed B.W. to pay attention to the content of women’s interviews and broader thematic patterns. While this interview study was B.W.’s first qualitative project, which meant she had a lot to learn, thematic analysis was a useful (and flexible) methodology for B.W. to gain her confidence with coding and writing up an analysis.
Moving the Research to Online Channels
Following the publication of the interview study, B.W. was considering her next research project and was convinced it also needed to be qualitative in order to best hear women’s stories. However, she was concerned about the significant recruitment challenges she had faced in the interview study, and more generally, her time constraints (i.e., 3 years is the expected timeline for PhDs in Australia). In reflecting upon the interview study, B.W. was able to recall three important take-home messages: (a) most of her participants mentioned that they saw her advertisement on the online pregnancy forums, (b) participants had overwhelmingly preferred more anonymous (or removed) forms of participation, and (c) that it was relatively easy to advertise online, particularly because one Australian pregnancy forum has a section designated to research. With this in mind, B.W. designed an online survey with open-ended questions in order to explore women’s experiences of seeking information and support about smoking during pregnancy in both online (pregnancy forums) and offline contexts.
Recruitment
In an attempt to humanize the research and further clarify her non-judgmental position on the topic, B.W. developed positive and friendly advertising material that contained personal photographs (of her and her dog) with a clear emphasis on women and their experiences. Although the survey may be considered less intrusive than interviews, B.W. felt that it was still important to represent the research in ways that were deliberately non-judgmental. Again, B.W. was seeking to recruit women who were recently pregnant and smoking. The survey consisted of open-ended questions related to whether (and why) women went online for information and support regarding their smoking during pregnancy, whether (and why) women disclosed to others online and offline about their smoking, and more generally, to their online experiences of support and their interactions about smoking (including how these compare with offline experiences).
Advertising for the online survey took place via prominent Australian online pregnancy forums and began in March 2013. B.W. identified 13 possible online forums to advertise the study. However, once again, she encountered significant obstacles in access. Only seven forums allowed her to post her advert. Some of these forums requested she obtain moderator approval (which she did, on a number of occasions).
Conferences as Opportunities for Feedback
By July (2013), B.W. was due to attend the International Society of Critical Health Psychology (ISCHP) conference in the United Kingdom. She had planned on presenting the preliminary findings from this online survey. However, by July, she had been actively recruiting for 4 months and again was struggling to recruit women to complete the survey. Only 38 women had started the survey, with most only completing about 40% of the survey—despite the survey only taking women around 10 min to complete.
In preparation for the conference, B.W. began re-visiting the idea that maybe there was a problem with how she was approaching this topic, since despite her deliberately non-judgmental advertisement, recruitment was still a challenge. At this time, B.W. was also playing around with a more conceptual idea of whether there was a narrative of smoking during pregnancy and whether that was a central reason why women’s accounts were brief and participation was limited. However, she felt unequipped (methodologically and theoretically) to broach this conceptual topic in her conference presentation. Instead, she decided to focus her presentation on the recruitment challenges and her move to online research that offered participants the anonymity they seemed to want. Although B.W. felt a sense of vulnerability in publicly describing all these “problems” associated with her research, she decided that she needed to consult others about this research problem and that this conference, being supportive of qualitative research and reflexivity, was going to be the most appropriate context to do so.
An Opportunity for Collaboration
Given her interest in women’s health, M.L. was eager to attend B.W.’s presentation and to learn about this important but under-researched area of inquiry. She listened to B.W.’s description of the multiple recruitment problems she encountered, as well as the creative strategies she employed to overcome them. What began to develop for M.L. during the presentation was that this wasn’t a standard “problem” of recruitment but a function of a much deeper issue. As a discourse analyst, M.L. has been trained to pay attention to how the language we use shapes our identities and our possibilities for action. She began to imagine how women might struggle to account for this unaccountable behavior and wonder whether B.W. might be, in essence, asking women to “speak the unspeakable.” After the talk, B.W. and M.L. discussed these ideas at length (and got on famously!), and a plan was struck to explore the data together, but this time using a discourse analysis—a method that allows the exploration of meaning-in-action. In adopting a discourse analytic approach, the researchers could investigate the ways in which participants’ identities were positioned in the context of the research as well as how they positioned themselves in their accounts. Thus, this approach enabled an examination of how identity and stigma were enacted and managed in the data.
After some significant international travel (the conference was in the United Kingdom, B.W. lives in Australia and M.L. in Canada), the two authors (now referred to as “we”) met to revisit B.W.’s PhD data using a discourse analytic approach. This research meeting was preceded by B.W.’s immersion in readings about discourse analysis and her development of some important analytic ideas. Together, we explored what participants did (and did not) say and the strategies they used to account for their smoking in pregnancy.
Initially, we analyzed the interview data focusing on the question, “How do women account for smoking during pregnancy?” This was seen to be an important first step in our work together, in that if we were to explore whether smoking during pregnancy is indeed “unspeakable,” we first needed to understand what women managed to say about their experiences. In this way, we were interested in identifying the discourses women drew on to account for themselves and their behavior. In short, discourses offer frameworks of knowledge. Our logic was that by identifying the discourses that women drew on during their interviews, we could better understand the possibilities (and limits) for accounting for the practice of smoking during pregnancy and the implications this might have for women’s identities. This approach was also seen as appropriate because it allowed us to potentially explain the aforementioned recruitment difficulties and women’s reluctance to participate (or at least participate in ways that maintained their anonymity).
Following the publication of our first analysis (Wigginton & Lafrance, 2014), we became interested in how women’s accounts of smoking during pregnancy might vary according to the discursive context in which these accounts are produced. Put another way, are accounts of smoking during pregnancy contingent on the context in which they are produced? It was this question that led us to consider multiple sources of data for our next analysis.
Utilizing Multiple Sources of Data for Qualitative Analysis
We began our second analysis, this time drawing on both the interview and online survey data. Despite being collected in different mediums and over different points of time, both data sources included women’s responses to various questions concerning their experiences of smoking during pregnancy, others’ responses, their disclosure of their smoking, and experiences of support. It was during the early stages of our analysis that we noticed a participant had cited a public figure in response to the online survey question entitled “Can you tell us more about reading others’ posts about smoking? How did people respond?” In particular, this participant had recounted a recent public “incident” where a popular Australian radio host (Chrissie Swan) was photographed smoking privately in her car while 7 months pregnant. This was a very public example of the naming and shaming of a “pregnant smoker” in Australia, so it is unsurprising that a participant was referencing this in their responses.
After some searching, we found several pregnancy forum threads that were dedicated to discussing this particular public figure, her smoking, and the public’s reaction. In fact, this “controversy” had sparked a total of 46 news articles, to which Chrissie Swan formulated a public reply in a well-known Australian newspaper. It was her reply that prompted a divisive response among the Australian public (including in the online forums).
Since the letter was highly relevant to the topic at hand and discourse analysis is applicable to all forms of text, we decided to include this public reply to smoking during pregnancy in our analysis. However, this was not without some deliberation. In particular, B.W. was hesitant about including an account from a woman who had been able to quit smoking following her being “caught” and thus provided a very different account from the rest of the data set which included women who continued to smoke. However, M.L., with extensive experience in discourse analysis, was able to point to the analytical opportunities posed by a more rich and diverse data set. For instance, exploring how women who are able to claim being smoke-free implicates their (maternal) identities is one such opportunity. As a result, we decided to combine the three data sets (some public, some private) in order to answer the question “How do women manage the spoiled identity of the pregnant smoker?”
Changing the Frame of the Study Through Discourse Analysis
There are many forms of discourse analysis. Budds, Locke, and Burr (2014), in their SAGE case study, provide a clear description of the two main types of discourse analysis used in psychology (Foucauldian discourse analysis and discursive psychology), their strengths and limitations, and how these methods can be combined. We also combined these two forms of discourse analysis by examining the fine-grained particularities of participants’ talk (discursive psychology) and how these are influenced by broader social, political, and institutional forces (Foucauldian discourse analysis) to shape and constrain what can and cannot be said (Edley & Wetherell, 1997; Wetherell, 1999). From this perspective, participants are active producers of discourse but are also constrained by discourse. As a result, discourses privilege certain ways of understanding ourselves and the world around us and, by implication, produce taken-for-granted knowledge and ways of speaking about particular topics.
As an example, expressions of smoking during pregnancy as “morally wrong,” “shameful,” “risky,” and “selfish” are culturally speakable—that is, they are taken-for-granted as the “appropriate” response to this behavior. This is because biomedical discourse has been built on medical knowledge about the harms of smoking and hence the riskiness of maternal smoking. In addition, dominant discourses of “good” motherhood portray women as selflessly absorbed in attending to the emotional and physical needs and wants of their child/children (including removing any potential health risks). As a result, women who partake in this practice are seen to be non-compliant with medical recommendations and either “ignorant” about the risks or “selfish” because they are willing to put their baby at risk. These constructions of women who smoke during pregnancy are possible because of the intersection of biomedical and good mother discourses, which provide the scaffolding of such portrayals.
Discourse analysis was a relevant method for our study because we were interested in women’s identity work. For instance, how do women construct their maternal identities in the face of their positioning as “non-compliant” and “bad mothers”? In discourse analysis, identity is conceptualized as dynamic and continually accomplished in and through interactions. According to Carla Willig (2000), identities also hold particular meanings and serve certain rhetorical or interactional functions, in that they do certain things in talk and have certain effects. It is from this perspective that we analyzed the three sources of data.
Stigma in Action
Through the process of applying a discourse analytic view to the data, we were able to re-conceptualize women’s management of stigma. Given that all three sources of data were implicitly addressing the notion of stigma, we were able explore how women’s identities are implicated in such discussions. From a discursive perspective, we were able to identify the “moral trouble” in women’s talk. In particular, we found that women immediately oriented to the notion that “smoking during pregnancy is problematic” and engaged in accounting work to “explain themselves” and their behavior.
Given our research question, through the analysis we paid attention to any identity work in women’s accounts. This led to an identification of two prevalent identities: “the smoker” and “the mother.” We were specifically interested in how women evoked these identities and what occurred in their talk when they did so (i.e., the rhetorical effects of these identities). For instance, throughout the data, we traced the ways in which women positioned themselves within a maternal identity and the discursive effects of such positionings.
In asking identity-specific questions of the data, we came to identify that women rarely evoked or stepped into a good mother identity. In fact, the only mention of a good mother came from Chrissie Swan (public media account) who claimed this identity following her successful cessation of cigarettes. Instead, women spoke about readily being positioned as bad mothers by others because of their smoking during pregnancy and, in their accounts, sought to defend this positioning. What emerged for us during this process was an overall inability for women to tell positive or political accounts of smoking during pregnancy—accounts that diverged from the “wrong,” “shameful,” and “selfish” narrative in which their only claim to a maternal identity was negative (the “bad mother”).
Reflecting on the research, we came to theorize the multiple challenges in recruitment as well as the details of the analysis as converging evidence of a concept we called “discursive silencing”—a process that renders stigmatized experiences untellable (Wigginton & Lafrance, 2015). That is, we shifted our view on the challenges in recruitment from a “problem” to an additional source of information and, in turn, were able to identify this as a discursive predicament. It was only in considering the entire research process as a whole (warts and all) that this overarching analytic frame came into view—one that we hope is useful for other research on stigmatized experiences. Part of the value for us was the application of discourse analysis which allowed us to view women’s stories as constrained by the discourses available to them—discourses which position smoking during pregnancy as an unspeakable and indefensible practice.
Conclusion
In this case study, we have engaged in a reflexive dialogue about our research journey, the challenges we encountered, and our methodological strategies for overcoming these obstacles. We showed how the problem of “slow recruitment,” short interviews, and limited participation in interviews and online surveys could be seen as informative with a different analytical perspective. In addition, we described how a timely media event could be integrated in the analysis to produce a diverse data set allowing for additional insight into the experience of interest. Our case study shows the value of reflecting on all stages of the research process and how instructive these reflections can be to the analysis.
Exercises and Discussion Questions
- Describe how the cultural acceptability of a topic may impact the recruitment and data collection stages of a research project.
- How do recruitment invitations position potential participants? For instance, how might the imagery and language used in the advertising of a research project influence participation?
- What did discourse analysis, as a method, offer this particular research project? What might it offer to your research project?
- How did discourse analysis change the researchers’ approach to the study of stigma?
- What are some benefits of using media accounts (like the public letter described here) as data sources? What are some drawbacks?
- What are some of the benefits of engaging in reflexivity (i.e., reflecting on the research, the researcher, and the topic)?
