Reflexivity, positionality, and disclosure in HCI

Are you an HCI researcher thinking about including a positionality statement, possibly for that CHI paper you’ve been working on? Here are some thoughts, mostly drawn from a paper I wrote: Embracing Four Tensions in Human-Computer Interaction Research with Marginalized People. Apologies for brevity and the occasional tupo.

tldr; yes, you should include a positionality statement — even if your paper does or does not involve marginalization — that covers values, ethical considerations, biases, and, only if YOU want to, disclosure of parts of your identity.

A very light background

Reflexivity and positionality are concepts from feminist standpoint theory that encourage identification of a researcher’s values, biases, backgrounds, and beliefs, thereby acknowledging that the researcher is in no way an objective, neutral party (Moss et al., 1999). Sweet puts it really well, stating:

“Questions of reflexivity ask us to consider who we should listen to and why, how to place actors’ ideas in a larger field of power, questions about our own relationship to actors’ theories of the world. Reflexivity asks us to approach our work with epistemological unease because we are always at risk of reproducing categories that reify power” (Sweet, 2020, pg 924).

One way I distinguish the two is that reflexivity asks “who is the researcher?” and positionality adds “who is the researcher in relation to others?”

Reflexivity, positionality, and disclosure in HCI

There are and have been established conversations in the world of sociology that tie practices of reflexivity to matters of power and relationships, both with participants and with audience members. This discourse has begun to take off in HCI. As an example, one paper I really love is titled “Out there”, in which Taylor calls for HCI researchers to examine what is going on “in here” to challenge ourselves in how we build, design, and make decisions for those who are “out there” (Taylor, 2011). It is beautifully written, and I would highly recommend it to anyone. But back to my point — I came across three calls for HCI to take up reflexivity and positionality practices.

Bardzell and Bardzell, 2011, pg 683

As part of their feminist HCI methodology, the authors identify “Researcher/Practitioner Self-Disclosure. Practice should involve a disclosure of the researcher’s position in the world, her or his goals, as well as the researcher’s position in her or his intellectual and, to an appropriate extent, political beliefs’’ and

“Reflexivity. Research should be characterized by ongoing self-questioning about whether the research is delivering on its ambitions to be feminist, improve human quality of life, and undermine rather than reinforce oppressive social structures, etc.”

Erete et al., 2018, pg 69

As part of their call to action, the authors write about self-reflection: “Self-disclosing information about aspects of our identity and positionality, and potential impacts to the design research process, also helps improve the transparency and understandability of our research [citing Schlesinger et al., 2017].”

Schlesinger et al., 2017, pg 5421

The authors’ fourth recommendation for their Intersectional HCI is to “Provide an Author Disclosure” and make the following points:

1) authors should think critically about why a disclosure section might be valuable while also acknowledging that disclosure will not always be a good thing for the author,

2) authors should not disclose if doing so puts them in danger of harm,

3) authors who are in less marginalized positions should practice reflexive disclosures to normalize them, with an awareness that this will not solve problems of bias completely.

They conclude this section by giving alternatives to identity-based disclosures (e.g., privilege, ethics, or values). Put succinctly, they write that “providing ethical considerations or including information about the author’s cultural context helps clarify who the we each paper mentions is and how that influences the research.”

Positionality != identity disclosure

There are a range of calls for reflexivity, positionality, and disclosures in these examples, and importantly, none of them are wrong in their particular takes on how to do them. To each of their points, these practices would facilitate a really cool way of thinking about how we do research. What is difficult is how the community has taken up calls like these. In our TOCHI paper, we found that:

“Experiences of the HCI researchers in this study indicate confusion around when it is appropriate to ask authors to be reflexive and what authors should include in positionality statements. While we asked participants to talk about their experiences with reflexive or positionality statements, many spoke about requests for identity disclosure. These experiences indicate that while there is a theoretical difference between reflexivity and positionality, this distinction does not reliably carry over to current practices [emphasis added]” (Liang et al., 2021).

In this paper, we uncovered some really valuable points to consider when thinking about disclosure statements and an expectation of including them. For one, not all disclosures have the same consequences. Disclosing that I am 26 years old is relatively inconsequential compared to disclosing that I am [redacted]. (I’ll let you fill in the blank here. I’m not specifying to prove my point!)

What is complicated is that sometimes identity-based disclosure statements are helpful, especially when it comes to research with marginalized people — but they are often entirely based on assumptions about parts of the researcher’s identity. A quote from our paper reflects a struggle in assuming: “Because if it’s someone who is cisgender, I do feel like we have the right to ask them why [they study trans populations]. But then you can never assume that someone’s cisgender, you just can’t.” (Liang et al., 2021, pg 20). Here, I completely get the hypothetical urge to know whether a cis person is studying trans people, because, well, cis researchers have not always been kind to trans folks, to put it mildly. What is great about this quote is that the person who said it is being reflexive by calling out their assumptions. In doing so, they help us remember that normative assumptions (e.g., white, straight, cisgender) can be really harmful for the other person if our guesses are wrong. Authors should feel free to disclose parts of their identities if they want to, but a lack of disclosure cannot be met with potentially false assumptions about who they are.

Looking across all of this, here are my thoughts on a few common questions I have seen floating around:

Should I include a positionality statement?

Yes, but that doesn’t necessarily mean they have to include an identity disclosure (unless you really really think it’s important). You can talk about privilege, ethics, power, values, all of the above. Positionality can look like a lot more than just identity disclosures. Considerations of power along a spectrum of privilege and oppression can be done without getting too specific (e.g., the axes of oppression from Morgan, 1996).

Should I include a positionality statement even if my study doesn’t involve marginalization or marginalized people?

Yes, adding context in the form of your reflections is always valuable. It helps the reader better understand where you are coming from, and how you arrived at whatever contribution you are making.

I am reviewing a paper that involves marginalized people. Should I request the authors include a positionality statement?

You can, if you give some thought as to why you want one and how having that information might change your opinion, for better or worse.

Here’s another quote from our study that highlights an important factor to consider for disclosure statements: audience and, more specifically, reviewers because they have the power to block or allow studies from getting out in the world.

“There’s a level of encouraging self-disclosure because it can actually encourage further trust in the research. So you say from my perspective, I am, for example, maybe I’m a queer woman, so […] I bring these new perspectives to this research that are really beneficial. But then there’s also the problem where not everyone reading that paper actually thinks that’s a benefit […] there is that potential again of people who are just biased and then they immediately think this paper is garbage because they think the researchers are too close to the problem” (Liang et al., 2021, pg 19).

I have heard of numerous accounts of anec-data where people have been asked about their identities in their reviews, and I experienced requests about my identity in the TOCHI paper. So, if you are finding yourself fixated on knowing a person’s identity, you need to be honest with yourself and consider the possibility that you are just trying to scratch a nosy itch. Once you acknowledge that possibility, you are then free to explore how having that information would affect the contributions of the paper. What is complicated, however, is that we might have very good reasons for requesting this information. I certainly have wanted to know more about an author while reviewing a paper. So, we put together this table of questions to think deeply with:

What are some good examples of positionality statements?

Starks et al. and Keyes et al. provide two examples of how to describe the authors’ positionalities without specifically naming their identities (Starks et al., 2019, pg. 291; Keyes et al., 2020, pg 32). In Starks et al.’s work in progress, I appreciate reading the three questions they worked through, because it gives me a handy peek into their process. Keyes et al. include a collective and individualized position statement that shows how the authors grappled with the limitations of their analysis and how they reason through them.

What I admire most about these statements is their open-endedness. They do not set out to be definitive in the way that we have come to expect from an academic paper. Providing insight into how the authors are still unsettled in their positions helps me as a reader think about how my own position might affect my work, without the pressure to have all the answers. This, in turn, pushes me to consider how my involvement in research and other peoples’ lives is not neutral, but one that is shaped by who I am instead.

My goal with this essay is to steer HCI researchers away from falling into “reflexivity theater” in which we disclose parts of ourselves solely as a way to perform for others. (I first saw this phrase from a tweet of Alba Villamil’s and I think it is brilliant and you should give Alba a follow https://twitter.com/albanvillamil/status/1427650675966550022?s=20). I believe that being thoughtful about our engagements with our work and then choosing whether or not to disclose that thought process would move us towards a more equitable way of conducting research. Wouldn’t that be lovely?

A very special thanks to Emily Tseng for giving feedback on this and for being my friend. Also to my advisors, Sean Munson and Julie Kientz for their guidance throughout the TOCHI paper.

References

Bardzell, S. and Bardzell, J. 2011. Towards a feminist HCI methodology: Social science, feminism, and HCI. Conference on Human Factors in Computing Systems — Proceedings (2011). 675–684. DOI:https://doi-org/10.1145/1978942.1979041

Erete, S., Israni, A. and Dillahunt, T. 2018. An intersectional approach to designing in the margins. Interactions. 25, 3 (May-June 2018), 66–69. DOI:https://doi.org/10.1145/3194349.

Keyes, O., Peil, B., Williams, R. M., & Spiel, K. (2020). Reimagining (women’s) health: HCI, gender and essentialised embodiment. ACM Transactions on Computer-Human Interaction (TOCHI), 27(4), 1–42.

Liang, C.A., Munson, S.A., and Kientz, J.A. 2021. Embracing Four Tensions in Human-Computer Interaction Research with Marginalized People. ACM Trans. Comput.-Hum. Interact. 28, 2, Article 14 (April 2021), 47 pages. DOI:https://doi.org/10.1145/3443686

Morgan, K.P. 1996. Describing the Emperor’s New Clothes: Three Myths of Educational (In-)Equity. In Diller, A., Houston, B., Morgan, K.P., Ayim, M. and Morgan, K.P. 1996. The Gender Question in Education. Routledge.

Moss, P., Jones, J.P., Nast, H.J. and Roberts, S.M. 1999. Thresholds in Feminist Geography: Difference, Methodology, Representation. Economic Geography. (1999). DOI:https://doi.org/10.2307/144254.

Schlesinger, A., Edwards, W.K. and Grinter, R.E. 2017. Intersectional HCI: Engaging identity through gender, race, and class. Conference on Human Factors in Computing Systems — Proceedings (2017). 5412–5427. DOI:https://doi-org/10.1145/3025453.3025766

Starks, D. L., Dillahunt, T., & Haimson, O. L. (2019, June). Designing technology to support safety for transgender women & non-binary people of color. In Companion Publication of the 2019 on Designing Interactive Systems Conference 2019 Companion (pp. 289–294).

Sweet, P. L. (2020). Who Knows? Reflexivity in Feminist Standpoint Theory and Bourdieu. Gender & Society, 34(6), 922–950.

Taylor, A. 2011. Out there. Conference on Human Factors in Computing Systems — Proceedings (2011). 685–694. DOI:https://doi-org/10.1145/1978942.1979042