YouTube Generated Video Clips as Qualitative Research Data: One Researcher’s Reflections on the Process

YouTube Videos as Data

In preparing to utilize YouTube videos as research data, I began by considering the ethical principles of the practice. Per the definition used by my university’s institutional review board (IRB), I understood that no human participants were being included in the study. My efforts fell under the category of research work exempt from IRB review as it involved “the collection or study of existing data . . . if these sources are publicly available” (Office of Responsible Research Practices, The Ohio State University, 2006). Given these circumstances, I initially confidently concluded that use of the publicly available information should not pose ethical dilemma—at least not in the strictest, IRB endorsed definition of the word. But, as I unpacked the meaning of “ethical” and explored its boundaries within the context of publicly available social media information, the simplicity of that conclusion dissipated.

A search through YouTube’s privacy standards alerts users to the ways in which Google (YouTube’s parent company) agrees to use and protect personal information. It does not impose limitations on how individuals within or outside of the platform may use the data made available via the platform. YouTube’s (2017a) “Community Guidelines” establish “a few ground rules to keep YouTube safe and fun for everyone.” The four points included in the guidelines are “respect the YouTube community,” “don’t cross the line,” “we review your reports carefully,” and “join in and have fun.” Under the “don’t cross the line” subheading, there is a link to YouTube’s copyright center. From there, one is directed to another linked page about fair use, where the word research is mentioned only once and as follows: “Different countries have different rules about when it’s okay to use material without the copyright owner’s permission. For example, in the United States, works of commentary, criticism, research, teaching, or news reporting might be considered fair use” (YouTube, 2017b). While these indications of the expectations of video uploaders and viewers did not provide me with much direction, it also did not discourage me from using the YouTube data as a part of my dissertation research. It did, however, alert me to the fact that I would have to continue to trouble this idea if I wanted to honor my research commitments to interacting with participants in a humanizing way.

Research literature addressing the usage of YouTube and other social media data has increased in availability since I completed my dissertation. Although I did not have access to all of the publications I consider here while I dissertated, I include them now because they contribute to the current conversation and to my ongoing consideration of this issue. Several authors (Caron, Raby, Mitchell, Théwissen-LeBlanc, & Prioletta, 2017; Giglietto, Rossi, & Bennato, 2012; Highfield & Leaver, 2016) point to the lack of consensus among researchers about how to ethically evaluate the usage of information that is publicly available on social media platforms. The degree to which writers problematize the matter of utilizing YouTube video postings for research purposes falls along a continuum. At one end, researchers using such information treat it as a set of nonhuman documents or quantifiable data, including little to no pontification about the ethics tied to the data source (e.g., Adami, 2009; Choi & Behm-Morawitz, 2017; Gooyong, 2011; Porter & Hellsten, 2014; Rotman, Golbeck, & Preece, 2009; Xu, Park, & Park, 2017). Although they do not always include detailed information about the ethical considerations influencing the choice, some other researchers report having received permission from the individuals whose videos they used in their studies (e.g., Frobenius, 2014; Harley & Fitzpatrick, 2009). Others at least briefly discuss the ethical perspectives associated with the data type, but ultimately discern without much trouble that their usage of the data is appropriate (Berger, 2012; Caron et al., 2017; Laurier, 2016; Misoch, 2014). As justification for their choices, these authors often cite the following: Those who share their videos on YouTube do so with knowledge that the resulting uploads are available to anyone with Internet access, and YouTube users have access to settings that will render their videos unsearchable by unknown others if they so desire. On the other extreme of the continuum, some researchers transparently grapple with the unsettling uncertainties that exist when working with social media data (e.g., Lomborg, 2013; Longhurst, 2009; Markham, 2012; Morrow, Hawkins, & Kern, 2015). I will rely upon several of these works along with the PiH theoretical framework as I reflect upon my dissertation process in this article.

PiH Research

In thinking about and choosing to display my dispositions as a researcher, I take up Kinloch and San Pedro’s (2014) conceptualization of research as PiH. From this perspective, I consider qualitative research to be the work that takes place in “the space between listening and storying” (p. 21). Within this space, my own ideas are influenced by, meshed with, and separated from ideas shared by others. This framework guides ways “to listen to ourselves and interact with others as we conduct research,” a process during which the researcher’s role shifts between “listener, learner, advocate and participant” (p. 22). Importantly, Kinloch and San Pedro do not see these interactions as being limited by space or time. Within their proposal of PiH, the traditional understanding of a dialogic relationship is reworked. For Kinloch and San Pedro, the reader of their chapter is engaged in conversation with them via the text as a static representation of a glimpse into their thought processes; the reader enters into a space of interaction with the authors through the act of reading and thinking about and merging their own thoughts with those being shared. Kinloch and San Pedro see conversation as the result of one individual honoring another by engaging with that person’s thoughts and allowing that person’s ideas to spark one’s own engagement, thus producing understanding. The authors identify sites of knowledge construction as those places in which “we walk through the world with other people” (p. 29) or, in my case, as we share the virtual space of the world wide web.

Using Kinloch and San Pedro’s perspective as guidance, an essential component of what I consider to be quality, worthy research is respect for the contributions that all involved in a research project have to offer. I seek to establish and maintain trustworthiness in my qualitative work by upholding the ethical standards outlined by San Pedro and Kinloch (2017) that include “willingly center[ing] the realities, desires and stories of those with whom we work” and “situat[ing] their stories in relation to our stories, lives and research projects in humanizing ways” (p. 374S). Given these methodological commitments, I had deep interest in interacting with the YouTube content in different ways than I may interact with anonymous document data such as Census forms or de-identified school records. In observing the intimate musings or passionate declarations of the YouTubers whose videos I viewed, I could not help but to see humanity—a unique though also commonly familiar striving to make sense of the world we cohabitate with others—as the perceptible energy fueling the posts. I felt ethical and methodological obligation to retain the participants’ humanness in my analysis, to interact with the video content while keeping the human conditions of their uploaders at the forefront of my mind.

Standing upon the principles put forth by PiH, in the absence of individuals with whom to engage in traditional two-way dialogue, I committed to engaging with the stories shared through YouTube in humanizing ways. I listened to the stories, watched them be told, transcribed them, read them, reread them, re-watched them, immersed myself in them, allowed the data to talk to me, and eventually also talked back to the data. Even given these efforts to honor the humanity of the storytellers, I still experienced doubts about characterizing them as “participants.” I thought it was overreaching to title the generators of the YouTube content as participants (considering they were actually “participating” in my project unbeknownst to them). With acknowledgement of this semantic challenge, I settled on referring to them as informants. I believe this term captured the position the video generators occupied—the space between being a participant with whom I could enjoy real-time reciprocal interaction and existing as a document in the traditional sense.

Deciding upon the terminology that most accurately connoted the nature of the data was only the first of several challenges I encountered in working with this data source. In the remainder of this article, I explore some of the tensions that surfaced as I managed the pressures of holding simultaneous allegiances to my research goals and to the integrities of my informants who were not aware that they were informing me. In particular, I focus on the following two tensions: (a) becoming acquainted with my informants and (b) navigating informant “attrition.” I conclude with a brief summary of the methodological lessons learned from grappling with and working through these instances of astriction.

Tension 1: Getting to Know My Informants

I utilize here the word tension (Charmaz, 2007; Gergen & Gergen, 2003) to describe my evolving relationship with the dissertation data because, though my commitment was at times stretched and strained, it was never broken. The first of such tensions came with the process of attempting to build a foundation of trust between myself and my research informants. Kinloch and San Pedro (2014) endorse the power of listening as a part of this process. The authors note the desire to be listened to and the act of genuinely attending to another’s stories as foundations of what can be human connection that generates knowledge. They describe engaging in observation through listening as a prerequisite to conducting research “with and alongside others” instead of doing research “for a purpose or for people” as such situations are significantly less likely to demand the valuing of participants as human beings (p. 24).

I relied upon intentional listening as a means for becoming a researcher whose efforts my informants could trust, even though the distance from which I was seeking to work alongside them was great. Knowing that my informants would not have the opportunity to evaluate me, but also remaining committed to thinking of my work as a Project in Humanization, I felt an increased pressure to ensure that I acted in ways that would engender trust. I imagined my informants coming alive off of the computer screen and watching me as I paused and rewound their videos over and over, scribbled notes about the stories they shared, turned the volume in my headphones as loud as it could go so I could re-listen to a mumbled uttering, and clicked on other videos on their channels in a quest to add to my understanding their of perspectives. In spite of these strivings, I sometimes became uneasy with the fact that I was both judge and jury in terms of assessing the trustworthiness of my methodological intentions. Was I unrightfully assuming too much power as a part of the research process?

In an offline research situation and in line with a PiH approach, I would have begun my efforts by casually communicating with participants, becoming familiar with them while we built reciprocal rapport and connected with each other as humans. Although power is at play in offline research situations as well, participants can exercise agency within the relationship in ways not afforded to YouTube-based informants. Morrow et al. (2015) wrestled with subjectivity and power as they reflected on their evolving beliefs around online research ethics. Like myself, Hawkins (one of Morrow’s co-authors) ultimately chose to understand the YouTube platform as a “site of ‘public’ discourse,” and thus did not seek consent to work with the data available there (p. 555). Also like myself, Hawkins admitted a sense of discomfort about the choice as she reflected on it, noting, “Questions still plague me about this decision” (p. 555). Upon collaborative contemplation upon the idea and application of a feminist framework, the undeniably problematic nature of power dynamics in online research did not discourage the authors from doing work in this context; rather, they concluded that these dynamics must be continually explored and the researcher must commit to practices of reflexivity and transparency as a part of the process.

I engaged in a similar reflective process, sitting with the feelings of disquiet and second-guessing I experienced instead of dismissing them to be able to move forward with my work unfettered. I had to make a decision about whether or not I was okay with the power dynamics as they were and with my attempts to disrupt them through adoption of a particular theoretical framing that influenced my methodological choices. As a researcher, I had to determine how I would balance genuine care for my informants as humans with “the researcher’s rights to analyze and interpret data” (Lomborg, 2013, p. 29) while also “trusting one’s interpretive strengths and engaging in an act of interpretive authority” (Markham, 2012, p. 345). As Denzin and Lincoln (2011) remind us, “there are no objective observations, only observations socially situated in the worlds of—and between—the observer and the observed” (p. 12). Although not without uncomfortable spots, I settled on working in the space between listening and storying (Kinloch & San Pedro, 2014), a space I believed I could occupy—or at least continue to strive to occupy—in humanizing ways. Coming to this conclusion meant that I had to trust myself as a researcher and a person committed to interacting with others in trustworthy ways, but not to extend this trust in a manner that removed me from obligation to continue to trouble the power dynamics influencing my work. I remained committed to honoring the PiH practice of listening to learn and, as Schultz (2009) describes, listening to “gain deeper understandings of who people are and who we are in different and complex contexts we find ourselves moving in and out of” (p. 24).

One of the components of the YouTube data that influenced my commitment to its use was that the digital platform the site provides gives users the opportunity to share ideas and comments they may not choose to share in their offline worlds (Patterson, 2017). Thus, in the cases of at least some informants, I could witness a level of intimacy and sharing that we may not have achieved in a face-to-face interview. I was afforded access to very personal thoughts that people chose to make public for specific reasons. In keeping with my pledge to trouble thoughts that could otherwise be conveniently oversimplified, acknowledgement of this unique opportunity for gaining insight also raised for me questions about distinctions between public availability and privacy. Longhurst (2009) uncovered a similar tension around vlogger intent in viewing and analyzing uploads of mothers who had posted videos chronicling their pregnancies and deliveries. Although the information was publicly available at the discretion and desire of the mother giving birth, the author grappled with how the children featured in the videos may feel in the future to read her interpretations of their entrances into the world.

Markham (2012) also acknowledged the daunting task of reconciling privacy ideals in an online research context, noting that “cultural definitions and expectations of privacy are ambiguous, contested and changing,” thus further complicating conversations around how to protect something so illusively understood (p. 336). For her, the crux of the privacy issue is around exposing individuals to vulnerability or harm. As such, researchers must assume the responsibility of making educated determinations about how to address individual situations that arise. In my case, even though informants were not always explicit about their reasons for sharing, the idea that someone would record a video of themselves talking about their racial identity and then post it for the world to see was important to me; the significance of this act alone made me want to think more about this particular brand of sharing. Following Morrow et al.’s (2015) conclusion, I understood the online space of YouTube to be simultaneously public and private. Taking up this understanding meant that I had to be satisfied with not having an unambiguous answer as to how my informants envisioned the space, and I also had to be okay with the idea that their conceptualizations about the space were subject to change given specific circumstances. My responsibility to interact with my informants’ stories in ethical ways I deemed trustworthy existed outside of, and possibly in spite of, the perceived public-ness or private-ness of their videos.

Tension 2: Reconciling a Loss-of-Subjects

In offline research settings, ethical research practices dictate that those who volunteer to interact with the researcher do so with an understanding that they are able to withdraw participation at any time and for any reason. In qualitative research, this type of loss-of-subjects threat cannot be anticipated, but its potential effects upon the trustworthiness of subsequent data analysis must be considered (Tuckett, 2004). Working with YouTube data created a unique loss-of-subject scenario, causing a second source of tension within the project. The loss-of-subjects reconciliation process was guided by my ethical considerations as framed within PiH. In particular, PiH shaped the way I thought of the temporal nature of my interactions with my informants’ videos.

Technically speaking, the YouTube platform is a space within which individuals can upload their videos, not a source from which they are to be downloaded. Although the task can be completed using third-party software, at the time of publication YouTube does not provide options for permanently downloading videos from its site, giving the uploader (and the YouTube site managers) ultimate control over video access. Utilizing the platform as it was intended to be used—as a streaming service and not a video depository from which I could download and re-catalogue data—translated into another opportunity to explore and to honor the public/private nature of the online space. Determination to interact with my informants’ stories in a trustworthy manner dictated that I interact with those stories within (the constraints of) the medium via which they had chosen to share them. As I identified videos of interest, I created a database of their hyperlinks. As the next step in the data collection process, I transcribed the portions of the videos that specifically discussed experiences that took place in educational contexts, per the larger goals of my dissertation work.

After the first round of transcribing, I often returned to the videos to ensure transcription accuracy or to glean additional insights from the informants by observing spoken and body language communication in concert. This process worked well, allowing for more rich interpretations or adding needed clarity, until I clicked on a link that connected me to a page providing the following automated message: “This video is unavailable. Sorry about that.” Initially I responded with confusion as I had only just recently successfully viewed the video. Upon further investigation, I found the video was, indeed, no longer accessible. This discovery—and the finding that 12 of 51 videos fell into this category—led to another series of tensions in navigating this data type.

In an offline research situation, though participants are afforded the right to suspend or terminate study participation at their discretion and without the necessity of an explanation, I have not yet had the experience of someone dropping from a study with such unexplained abruptness as I experienced through the viewing of this automated message. Caron et al. (2017) recounted a similar instance when they found—“disappointingly” so—that a YouTuber whose post they had included in pilot study analysis had made their YouTube channel private before the authors were able to return to it (p. 55). Experiencing similarly disheartened emotions, I wondered what right I had to continue using transcriptions of the video data as I bombarded myself with a barrage of questions: If I thought of the newly inaccessible videos as consenting participants who had withdrawn participation, did I have legitimate access to information gathered before they decided to suspend participation? In an offline instance, wouldn’t I make this decision on a case-by-case basis, informed by knowledge about why a participant wished to discontinue with the study? Thinking within a PiH perspective that called upon me to consider the humanity of my informants, could I reasonably assume that I would be able to remove the influence of a past interaction on my current understandings because the person with whom I shared the interaction wished it to be?

Considering that the data were, in fact, document data and not human participants, could I apply the understanding that copies or transcripts of original data would continue to be usable if the original documentation was suddenly made unavailable in the case of fire or some other disaster? Did reducing the videos to document data to serve my own purposes violate my larger commitments as a researcher? This may have been the one question for which a PiH lens pointed to a clear answer, in this case, an emphatic affirmation that it would indeed be a violation of my commitments. The others, however, were not so cut and dry. As these questions swirled, I became aware that the answers I settled upon would need to honor both my commitments to my informants—to treat them with their humanity in mind—and to my research project—responding to my guiding questions using an identified set of data.

Again, Lomborg’s (2013) and Markham’s (2012) ideas about balancing researcher authority and ethics of navigating online data spaces came into play. One realization that greatly contributed to my decision as to how to move forward was about the randomness of the timing. Although a PiH approach drives me to welcome and embrace the products that result from critical listening and engagement, it also spurs me to appreciate these interactions because none of them are promised. We can only create spaces that may inspire engaging interactions; we cannot plan for the spontaneity of the specific interactions themselves. A formerly viewable YouTube video could become unavailable under one of three circumstances: It could be removed from the site altogether by the user; it could be re-categorized as private by the user meaning it now required a specialized hyperlink for access; or it could be removed from the site by Google based on an issue of copyright or violation of community content standards. Unfortunately, I had no way to anticipate when or why a particular video would be removed from the site. Thus, I was forever at the mercy of the serendipitous timing I chose for viewing a particular video.

As approximately 300 hr of video are uploaded to the YouTube site per minute (Smith, 2017) and approximately 32% of uploaded videos are taken down within 24 hr (Krum, 2016), the video pool to which I had access at any given time was constantly being repopulated. Realizations about the randomness that characterized access to the videos made me more confident in the decision to use transcripts even from videos that had been removed from the site. Taking further consolation from a PiH perspective, interactions are not temporally limited; the thoughts of one can continue to influence the thoughts of another long after the thoughts were first shared and engaged with. Stresses around the limits of my access to the data in a way that would be most convenient to me as a researcher fell away as I came to appreciate every interaction I had with informants as an opportunity to listen to and have my own thoughts sparked by the thoughts of other beings. Appreciation of the informants’ stories in this way renewed my commitment to interacting with the data as a project in humanization.