Methodological Research on “Sensitive” Topics

Randomised Response

Lee (1993) introduced a distinction between “desensitizing” and “dejeapordizing” techniques for asking survey questions about sensitive topics. The former helps to foster disclosure by an interview respondent by creating a climate within the interview conducive to frankness. The latter typically attempts to minimize potential risks from disclosure by breaking the link between a respondent’s identity and the specific response(s) made by that respondent. Relevant techniques include the use of anonymisation strategies and the insertion of random noise into the data collected. While this insertion might occur during data post-processing, an important class of techniques seeks to inject indeterminacy directly at the point of data collection. A specific, and well-favoured, technique is that of “randomized response” (Warner, 1965). Although there are a variety of randomised response models of greater or lesser complexity, the procedure typically involves a respondent choosing to answer a sensitive or an innocuous question that has been assigned with a known probability. The response is given, however, without it being divulged to the interviewer which specific question is being answered.

A technique that perhaps owes some of its popularity to its intriguing and counter-intuitive character (you get the answer to a question without knowing what has been asked), randomised response and its variants attained small but noticeable representation in van Meter’s original review, and continues to attract interest.

It should be apparent on even a moment’s reflection that the method necessarily involves a number of trade-offs. The protection against self-incrimination given to the respondent is bought at the cost, for example, of decreased statistical efficiency, difficulties in assessing covariation with relevant variables, and operational complexity. Conceivably, too, there might also be psychological resistance to giving affirmative responses to whatever question is mandated by the randomisation procedure. Our sample indicates that over the last decade work has been devoted to ameliorating some of these difficulties through the development of randomised Fox response models that are more efficient and flexible than the models initially developed (Fox, 2005; Fox and Meijer, 2008; Lensvelt-Mulders et al., 2005; Ostapczuk et al., 2009). The comparison of randomised response to other strategies for eliciting sensitive information has also been apparent (see, for example, Lara et al., 2004).

Online Methods

Technological developments have often had profound, if not always anticipated, effects on research practice. One thinks, for example, of the impact of audio recording on the emergence, growth and popularity of in-depth interviewing methods (see Lee, 2004), or the way in which the development of telephone interviewing encouraged survey research on previously difficult-to-reach groups because it reduced considerably the costs of screening the wider population (Nathan, 2001). In recent years, it has been the Internet that has proved to be a major agent for the transformation of research methods. As Lee, Fielding and Blank put it (2008: 18):

The Internet provides psychologists with a space for experimentation that extends far beyond the walls of the conventional laboratory, and a traditional reliance on undergraduate research subjects. Survey researchers value the potential reach of the Internet and the level of control it gives them over cost and the particularities of data collection. Ethnographers of varying disciplinary hue, meanwhile, find in the Internet a plethora of virtual field sites ripe for the novel understanding of interaction, culture, community, and identity, and are attracted by “the apparent abilities of the Internet to sidestep, transform, highlight or reinvent some traditional political formations, identities and inequalities” (Hine, 2005: 242).

For those seeking to research topics that are deemed for whatever reason to be “sensitive”, online methods present interesting opportunities and challenges. For some of the survey researchers represented in our sample, the use of Web-based techniques has provided access to hard-to-reach populations, helped to overcome at least some kinds of self-selection bias, and offered a degree of anonymity that can enhance reporting of even quite sensitive matters (Mangan and Reips, 2007). As Mangan and Reips point out, however, these benefits are partly offset by limits on the obtaining of detailed information and problems controlling bogus participation. Internet surveys can have advantages when time and resources are in short supply (Shields, 2003). Use of a Web-based survey enabled Shields to overcome resource constraints in her study in an ethnically mixed school district in a large sparsely populated region of the United States. The protocol Shields developed included a substantial open-ended component that, in her judgement, provided qualitative data rich in detail and allowed pupils to disclose information about sensitive matters such as racial discrimination.

One important recent development is that the focus group has moved online. An advantage of this is that it creates the potential usually not possible with traditional methods to collect data from populations that are widely dispersed geographically (Oringderff, 2004). Reviewing the benefits and challenges of online focus group methods for children with chronic health conditions, Nicholas et al. (2010) noted a degree of respondent preference for face-to-face data collection but pointed to gains in efficiency and topic focus, and the fostering of participation in cases where face-to-face contact presented difficulties. While traditional focus group methods trade on the additional interpretive gains available from observation of non-verbal behaviours, text-based online focus groups forgo those gains. What such groups provide in relation to sensitive topics is anonymity, the ability to record and analyse virtual interactional dynamics, and, once again, the efficiency gains associated with access to machine-readable data (see Franklin and Lowry, 2001).

Finally, online methods potentially create possibilities for citizen participation in the design and execution of research that might encourage democratisation of the research process (Lee et al., 2008).

Ethical Governance

It has long been recognised that research on sensitive topics poses particular, sometimes intractable, ethical difficulties. Many of the issues involved, and discussions of their resolution, are represented in our sample. One broader contextual issue, however, that was of particular concern to a number of writers relates to the increasing problematics that now surround ethical governance. In many countries, systems of ethical review traditionally focused on biomedicine have increasingly been extended to the social sciences with requirements for mandatory prior ethical review by human subjects review boards, ethics committees, and the like. The extent to which this extension of bureaucratic surveillance improves the ethical practice of social scientists has been somewhat contested. The issue is particularly important in relation to putatively “sensitive” topics since the potential for the “chilling” of research which is “difficult” or controversial might be high. Herdman (2000) discusses in some detail a study in which researchers, research participants, and members of the ethics committee that had approved the study were all at loggerheads, a situation exacerbated by sensationalist media coverage of the research findings. The researchers regarded many of the strictures placed upon them by the ethics committee as unworkable, given the qualitative nature of their research. In addition, the committee itself, most of whose members were medical professionals, responded negatively to criticisms of their role directed at the absence of social science expertise among committee members. While it is not surprising that mutual incomprehension can come to characterise governance systems in which biomedical professionals make judgments about social science research of a qualitative kind, the work of Gerrish and Guillaume (2006) suggests that survey researchers, too, can face difficulties with ethical regulation. In particular, they suggest, there is a need for the level of scrutiny received by proposals to be proportional to the potential risk involved.

Following on from this, governance bodies often make rather commonsensical assumptions about “sensitivity” that do not match the experiences of researchers routinely dealing with sensitive topics. Buckle et al. (2010) note that a degree of incongruity exists between the perspectives of research ethics boards and research participants. Whereas ethical review committees assume that the bereaved are inevitably distressed by being interviewed about their experience of bereavement, research participants often find such interviews to be beneficial. Much of this work underlines the importance of feeding back to bodies concerned with ethical governance empirical work on the operation of the governance system.

Emotional Demands of Research on Sensitive Topics

We have noted the emergence over the last decade of work on the emotional well-being of researchers. Strongly influenced by the feminist tradition, this work rejects a model of the researcher as detached and objective. It does, though, see personal engagement in the research process as potentially having emotional consequences for the researcher. These consequences can be all the more difficult where sensitive topics are being studied. For example, speaking of her research on battered women, Chatzifotiou comments: “By talking about their painful experiences of violence the interviewees relived the violent events in their entirety, and by listening to their traumatic stories I felt inevitably overwhelmed” (2000: 9.1). Melrose (2002: 347) speaks of her “anger, guilt, powerlessness, and frustration,” when listening to accounts of juvenile prostitution, and notes that these feelings often persisted well beyond the interview itself. A somewhat broader overview is provided by the work of Dickson-Swift and her colleagues (2007, 2008a, 2008b, 2009). Drawing on interviews with a sample of public health researchers in Australia, Dickson-Swift et al. map out the emotional challenges that face researchers when confronted during the course of a study with difficult and sometimes harrowing topics. Feelings of vulnerability, guilt, and physical and emotional exhaustion were commonplace while researchers struggled with issues relating to feelings of desensitisation, transference and counter-transference, and sought to locate the boundaries of their own responsibility towards alleviating the suffering of those studied. ³ Dickson-Swift et al. point to the importance of training, supervision, and support in helping researchers deal with the emotional consequences of researching sensitive topics. They also underline the importance of self-care; the need, for example, for careful scheduling of debriefing sessions and time away from the field as ways of avoiding emotional burnout. Work of this kind has been extended in interesting ways by Corden et al. (2005), who describe the exploratory use of professional therapeutic support by a team of researchers studying the financial implications for parents of the death of a child. The team availed themselves of the services of a psychotherapist who provided a support group for the researchers, as well as being available for consultation by individual researchers. In fact, it seems that the researchers drew in a fairly limited way on this resource, although they found it beneficial and enabling. As Corden et al. point out, the funding of such provision needs to be factored into research proposals. This, of course, might limit availability, but the approach itself is one which is likely to repay further investigation.