The issue of confidentiality in research design has been discussed for many years in the research community and is generally regarded as a hallmark of any study.  Researchers have (and do) approach the design process with the underlying belief that people will not respond or not respond honestly to our research requests unless they are convinced that all responses will be held in confidence.  And, because minimizing response error while maximizing the truthfulness of our outcomes are at the core of what we do, confidentiality has been an indisputable component of our research designs.

But does confidentiality really matter?  Is the integrity or quality of our research impacted by the degree to which research subjects are persuaded that their responses will not be divulged as their own?  The assumption is that there is an indirect relationship between the level of concern about confidentiality and the rate and quality of response (the higher the concern the lower rate and quality of response).  But is that true?

Various survey researchers have explored this question.  Singer, Von Thurn, and Miller¹ conducted a meta-analysis looking at studies dealing with confidentiality and its effect on response.  They concluded that the assurance of confidentiality has some effect when the research topic is deemed sensitive (i.e., stronger promises of confidentiality lead to “better results”) but, regardless of the level of sensitivity, “the effect of confidentiality assurances is not large.”  The authors further conclude that “elaborate assurances of confidentiality” can actually backfire by unduly raising respondents’ concerns about the survey.

Confidentiality plays as large, if not larger, role in qualitative research.  Even in the expanding world of online techniques, much of qualitative research is intimate in nature; revolving around a close interaction between researcher and participant.  Maybe because of this intimacy qualitative researchers are particularly careful to comfort discussants and interviewees with the understanding that their comments “won’t come back to haunt you.”  This is no more true than in employee research where corporate employees are typically fearful that their candid responses will jeopardize their jobs.  But are our efforts to allay these concerns improving the quality of our qualitative research?   Unfortunately it is difficult to say because there is so little examination of best practices in qualitative research generally and on this issue specifically.  In my own work I see anecdotal evidence that confidentiality (even among employees) does not necessarily reap more honest or heart-felt responses.

All of this has left me thinking: If there is no strong link between our guarantees of confidentiality and response rate or response quality – and if these assurances can actually backfire and have a negative effect on response – are we missing an opportunity?  That is, would our respondents and participants actually be more forthcoming if we dropped the promise of confidentiality and promised instead that each individual voice is being heard, that their opinion does count, and that their participation in our research really does make a difference?  Can we gain more from our research by emphasizing “you”?  These may be particularly fitting questions as the use of social media in research – bringing product/service providers closer to end-users – continues to evolve.

¹Singer, E., von Thurn, D., & Miller, E. (1995).  Confidentiality assurances and response: A quantitative review of the experimental literature. Public Opinion Quarterly, 59, 66-77.