The title of this post is joking, but is a reflection of my hesitancy to make strong recommendations about, well, anything education-related. But I’ll try to make a moderately strong case here. The quick back story is that Macy Burchfield, an undergraduate student at the University of Tennessee, Knoxville, led the analysis of data and the writing up of a study on students’ privacy in the context of K-12 schools’ and districts’ posts on Facebook. We finished analyzing the data this week, and completed a proposal for a conference; we shared a pre-print here: https://joshuamrosenberg.com/pre-prints/burchfield-et-al-2021-facebook-student-privacy.pdf
The punchlines are:
- We estimated that there are between 15.2 and 20.3 million photos of students shared by schools and districts
- We estimated that there are between approximately 150,000 and 1.15 million photos of students shared in ways that make it easy to identify them (by connecting their name with their face in the photo)
- These are strictly public posts; anyone (even without a Facebook account) can access them, and these photos on schools’ and districts’ pages may represent the largest unstructured but accessible collection of identifiable images of students/minors/children in the United States; if accessed with information about the location/name of the school and district and the time stamp, this data could pose a serious threat to (many) individual’s privacy
Yikes is the response.
I was so … taken aback by the findings that I jotted down some thoughts at the same time that Macy was submitting the proposal. These are for an audience of administrators/educational leaders/educators; maybe for a venue such as Educational Leadership. Anyway, those thoughts are below. I am sure I will shape them up (likely with Macy and colleagues), but thought to share these in case others had thoughts/advice to add on - as well as feedback on whether there is a serious threat from students’ photos being posted.
For many schools and school districts around the county, a Facebook page or Twitter account serves as a context for updating the community about events, highlighting the accomplishments of staff and students, and, during a pandemic, sharing essential information (Kimmons et al., 2021; Michela et al., 2021; Rosenberg & Nguyen, 2021). You may have liked posts from such pages or accounts yourself! We, too, have, considering these posts (and our interactions with them) to be a way to connect with our communities or children’s (or our own) school or school district, and others in it.
Our Research on Schools and Districts’ Use of Facebook
Over the past year, we have explored how schools and school districts use social media to communicate, and we have been impressed by the unabashedly positive and uplifting messages shared by the accounts associated with these K-12 educational institutions. At the same time, we noticed a pattern that gave us a moments pause: Many of these public posts included the names and faces of students. Moreover, because these were posts by schools and districts, they were identifiable photos that were also implicitly tied to a particular place. While privacy has been a focus of some research on social media use (e.g., Fiesler & Proferes, 2018), no research had investigated privacy in the context of social media use by K-12 educational institutions. We note that while any individual crafting a post is doubtlessly with the intent to recognize or even celebrate individual students—indeed, many identifiable posts were to announce students’ accomplishments—the effect could be that students’ privacy is reduced in ways small and potentially large.
How might this data be used in ways that are harmful to students or their parents and guardians? While we do not have any reason to think that there are individuals or organizations using identifiable posts of students for nefarious ends, there is the risk that photos of students linked to their name and school (or district) could be used to threaten individual students or groups of students now or in the future. Indeed, it is hard to predict how identifiable photos of students could be used; we already know that facial recognition algorithms can predict characteristics of people, such as their political identity, better than other humans can (Kosinski, 2021). It is not beyond the pale to imagine predicting other characteristics of students, including those that could be more damaging if predicted (correctly or incorrectly) and revealed than one’s political identity. Furthermore, it is widely-documented that Facebook creates “shadow” profiles for individuals depicted in photos, but without accounts (Quodling, 2018). Finally, it is possible that malicious actors could use such posts to target individual students (e.g., cyber-stalking). Thus, Facebook, along with others, may already be using data on students. In sum, the risks to individual students may be small, but the risks to a small number of students over time may be great, and difficult to predict at present.
How widespread is the potential issue? We found that there have been a remarkable 18 million posts on Facebook from United States schools and school districts. Importantly, 9.3 million posts included one or more images (Burchfield et al., 2021). In our research, we manually examined 100 randomly sampled posts containing photos (from the larger collection of more than nine million posts). For these, we counted how many photos that made students’ faces visible. Then, we sought to find out how many of these photos were named. In all, we detected 187 student faces in the 100 posts we analyzed; of those 187, five were identified in such a way that it was clear what the students’ name was (and vice versa; what the photo of the student who was named was).
While these numbers for the sample (of 100 posts) we analyzed are small, when we consider the entire set of posts, they became more notable. We carried out an inferential analysis and estimated that there are between around 15 and 20 million photos of students’ faces shared on the pages of schools and districts. Between around 150,000 and more than one million students’ faces are shared on these pages in ways that these students can be identified.
Implications of Our Research on Students’ Privacy on Social Media
What should we take away from these findings? The scale of Facebook is such that even small proportions of posts that identify a few students accumulate over time and across school and district pages to posting tens of millions of photos of students, possible identification of hundreds of thousands, and perhaps more than a million, students. These posts could represent the largest public collection of identifiable photos of students (and minors). Importantly, and highly disturbingly, these posts are accessible by anyone in the world—even individuals without a Facebook account who simply visit the page of a school or district. Moreover, given the nature of data such as that generated by Facebook, these posts are also likely to persist on the Internet for a long period of time.
It is a strange feature (and perhaps bug) of social media that posts, doubtlessly each written with only the aim of lauding a student (and allowing others in the school or district community to do the same), could mean that students’ privacy is placed at risk. In some ways, the blurred boundaries between the personal and professional on social media contribute to this issue: We are accustomed to sharing (identifiable) pictures of ourselves and our family members and friends through social media. The key difference is that the pages of schools and districts are not, like our own posts, restricted to those who we wish to see our posts; they are, perhaps strangely, available to anyone who wishes to see them. In this way, K-12 institutions’ use of Facebook represent a form of context collapse, the phenomenon of social media encouraging individuals to communicate to and with multiple audiences (i.e., friends and colleagues at work) via the same medium, breaking down historical and traditional barriers between interacting across these different audiences (Marwick & boyd, 2011; Vitak et al., 2012).
Our goal in this research is not, at all, to critique or to point out flaws in the social media practices of K-12 institutions. Indeed, we have been uplifted by how schools and districts used - and use - social media, particularly during the COVID-19 pandemic, when we have seen the many ways districts have continued to provide (and communicate about) the services they provide. Our aim, though, is to prompt reflection and changes in how institutional social media platforms are used, as we discuss next.
What Schools and Districts Can Do to Protect Students’ Privacy on Social Media
What can and should schools and districts do? First, we recommend that schools and districts take small steps to protect the privacy of students. This can be done by taking care when posting photos of students that are easily identifiable (i.e., portrait/yearbook photos and photos that clearly depict most of students’ faces). Also, when posting photos, especially portrait photos, posting only students’ first name (and not their last name or grade level). These would be best done in conjunction: Not posting photos in which students’ faces are clearly identifiable and using only students’ first names.
In all, we urge schools and districts to be cautious when it comes to posting identifiable information about students on social media. While it is likely not posing legal issues (or resistance from parents), there are privacy threats to students that may be serious now or in the near present. Moreover, an ethical perspective suggests that students - minors - may not yet fully understand how their likeness is being used and shared by their educational institution, and we should be cautious given the responsibility we have to protect and care for them.
We also recommend some reflection and potential changes at the level of how parents and guardians are notified about how their childrens’ names and photos may be used on social media. At present, media release forms are common; as are policies asking parents to “opt-out” (by notifying the school or district) of their child’s name or photo being shared in web, social media, and other materials. First, we recommend that districts use media release forms, instead of a system that puts the responsibility on parents or guardians to opt-out. Requiring this is likely a greater administrative burden for schools and districts, but perhaps that burden can emphasize the gravity of the situation (and the potential risks) involved. Second, we recommend that districts share information with parents, students, and the wider community about the privacy risks associated with social media use; such communications can establish the norm of not sharing highly identifiable information through official school or district pages.
Finally, we note that our study (and therefore any recommendations we would like to make) do not pertain to individual educators’ uses of social media. While some of the issues we have discussed are comparable between schools’ and districts’ social media use and the use of social media by individual educators, the risks are smaller, in our view, because the reach and audience of individual educators is, in many cases, smaller than schools and districts–and it is more challenging (for us and others) to collect this kind of information from individual educators at scale. Nevertheless, some of the recommendations we have made likely apply to educators sharing news and updates from their classrooms via social media, and we also urge some critical reflection and careful planning pertaining to these uses of social media.
Social media is here to stay, and we see this as a potential boon to educators and educational institutions seeking to communicate with parents, students, and the community more effectively using tools that many individuals are already using. Our research has deeply impressed upon us the power of social media to both support schools’ and districts’ efforts to continue to provide services (instructional and other) to students, and to communicate about these services, especially during times of crisis. Our research has illustrated that there are potential risks associated with social media use related to posts about students; particularly, posts including identifiable photos. With millions of photos, and hundreds of thousands (and potentially millions) of identifiable photos that are both connected to particular locations (and times), the risks to students at present and in the future are notable. This is especially the case given the high degree of social media activity by educational institutions that is entirely public; accessible to so many in the community, but also those who may wish to cause harm.
So, as you view a post by your school or district, continue to connect. Our aim is to prompt reflection, especially on the part of those responsible for posting from their schools or district’s social media page. As technologies and modes of communications advance, issues beyond the benefits of these tools also emerge, with privacy being perhaps a key one for us as educators to consider going ahead.
Burchfield, M., Rosenberg, J. M., Thomas, T., Borchers, C., Gibbons, B., & Fischer, C. (2021). Is student privacy “quick and easy”? Investigating student images and names on K-12 educational institutions’ Facebook postings. http://joshuamrosenberg.com/pre-prints/burchfield-et-al-2021-facebook-student-privacy.pdf
Marwick, A. E., & Boyd, D. (2011). I tweet honestly, I tweet passionately: Twitter users, context collapse, and the imagined audience. New media & society, 13(1), 114-133.
Michela, E., Rosenberg, J. M., Sultana, O., Burchfield, M., Thomas, T., & Kimmons, R. (2021, April). “Life will eventually get back to normal”: School districts’ Twitter use in response to COVID-19. Presentation at the American Educational Research Association Annual Meeting, Orlando, FL.
Kimmons, R., Rosenberg, J., & Allman, B. (2021). Trends in Educational Technology: What Facebook, Twitter, and Scopus Can Tell us about Current Research and Practice. TechTrends, 1-12.
Kosinski, M. (2021). Facial recognition technology can expose political orientation from naturalistic facial images. Scientific reports, 11(1), 1-7.
Rosenberg, J. M., & Nguyen, H. (2021). How K-12 school districts communicated during the COVID-19 pandemic: A study using Facebook data. In XXX (Eds.), Proceedings of the 11th International Conference on Learning Analytics & Knowledge (pp. 1-4). Association for Computing Machinery.
Quodling, A. (2018). Shadow Profiles-Facebook Knows about You, Even If You’re Not on Facebook. The Conversation, 13. https://theconversation.com/shadow-profiles-facebook-knows-about-you-even-if-youre-not-on-facebook-94804
Vitak, J., Lampe, C., Gray, R., & Ellison, N. B. (2012). “ Why won’t you be my Facebook friend?” strategies for managing context collapse in the workplace. In Proceedings of the 2012 iConference (pp. 555-557).