Ethics of Analysis: Learning Analytics and Academic Libraries
By: Kelly Bilz
Since getting into the swing of the new semester, something that has been on my mind is the role of the library in students’ success. Georgia State University (GSU) has taken this role very seriously, implementing a novel method to make sure they can help students who are struggling, namely, by using learning analytics. This was a massive undertaking on GSU’s part, using data from students (over 40,000 of them), identifying risk factors (800, to be specific), and having “interventions” based on alerts when students were falling behind. Listening to Vice Provost Timothy Renick describe the outcomes of using learning analytics, GSU has had some undeniable successes. Other universities and higher-education institutions have taken note, including academic libraries, and they have begun to invest in learning analytics themselves.
Learning analytics was listed as one of 2018’s top trends for academic libraries by the Association of College and Research Libraries. Learning analytics, the ACRL says, means “tying the use of library materials and services to student performance measures, such as GPA and retention rates.” However, the ACRL notes that learning analytics has drawn criticism from privacy and intellectual freedom advocates.
Collecting data in a library setting isn’t new: circulation statistics, assessment measures, and user surveys are not only useful, they’re critical for enhancing the library’s services, as well as communicating the library’s value to stakeholders. So what makes learning analytics so different?
First, there are concerns about collecting and using user data in an ethical way. Kyle M. L. Jones and Dorothea Salo, in their article, “Learning Analytics and the Academic Library: Professional Ethics Commitments at a Crossroads,” raise the question of students’ data when it comes to negotiations with content vendors: “it (learning analytics) creates conditions where institutions and content vendors may use data as a bartering chip in contract negotiations.” Jones and Salo also argue that learning analytics is a threat to intellectual freedom, as it “negatively affects the conditions necessary for the free pursuit and dissemination of ideas by tracking and influencing behaviors.” Student might be more hesitant to research what they are curious about, and doesn’t that completely undermine the library’s mission?
Data collection, in fact, is only half of the story. The next step of learning analytics is predictive analytics, including a process called nudging. Nudging means “prompting individuals to modify their behavior in a predictable way (usually to make wiser decisions) without coercing them, forbidding actions, or changing consequences,” according to this article from the Educause Review. There’s a theory around nudging that predates learning analytics, and it’s distinct from its counterparts, “shoving” and “smacking,” which are coercive or punitive. The idea behind learning analytics is that, if students show signs of falling behind, they can somehow be “nudged” into, say, meeting with an advisor or going to a tutoring session. For libraries, students can be “nudged” to meet with a reference librarian or use any other of the library’s services.
The point of learning analytics is to influence students’ behavior—for their own benefit. (Using their powers for good and not for evil, one might say.) As one professor put it in an article from The Chronicle of Higher Education, “If students know that the teacher can pull all the data, they might put in more effort.” The logic in that statement may be sound, but it’s reminiscent of Orwell: Big Brother is watching, and he wants you to put in a bit more effort. Also, your grades in O-Chem are slipping; have you tried going to the library?
With learning management systems like Canvas and Blackboard moving much of the classroom into the digital landscape, the classroom has become less of a lecture hall and more of a computerized panopticon where students are constantly surveilled. Zoe Fisher, who gave an excellent rebuttal of the practice at the 2018 California Academic & Research Libraries conference, argued that learning analytics “conflates data tracking and surveillance with library assessment.” Jones and Salo, as well as Sarah Hartman-Caverly, in her book chapter “Our ‘Special Obligation’: Library Assessment, Learning Analytics, and Intellectual Freedom,” also recognize learning analytics as a form of surveillance. Even if the data is effectively anonymized, students still know that they’re being watched, and that doesn’t make the library–or the university–a safe space for inquiry.
One of GSU’s biggest achievements from using learning analytics was gains among students from underrepresented groups. But April Hathcock traces data collection/analysis to the history of colonization and enslavement. For GSU, learning analytics could have, as they claim, provided support for minority students, but in most cases, data collection targets–and threatens–minority students the most. Hathcock also raises the point of students’ consent to this extensive data collection. Hartman-Caverly, similarly, refers to the methods of collecting students’ data as “surreptitious.”
I’m a student myself, so this issue weighs heavily on my mind. Is learning analytics used to change the library, or is it used to change students’ behavior? Is the end objective improved service, or control?
Fisher’s critique includes the following quote from Canadian researcher Lise Doucette: “I always ask, ‘What would you do if the results were opposite of what you expected? What if library use was correlated with NEGATIVE student outcomes? e.g., the more students used the library, the worse their grades were?’” These questions point out the biggest problem with analytics: if there was a negative correlation, what toll would it take on the university’s view of the library, or the library’s budget, or students’ trust?
On the other hand, if it only confirms the library’s assumption, what use does learning analytics have as an assessment tool? If it’s not to improve services, and if it’s such a threat to ethics, what are we doing all this for? Perhaps we should have a clear answer before putting the profession on the line.
Kelly Bilz is a graduate student from Kentucky pursuing her MLIS with a specialization in academic libraries. She works in her university’s Special Collections as well as the local history department of a public library. Kelly first heard about intellectual freedom in her Information in Society course and has spent the time since arguing with her friends about intellectual freedom in algorithms, ethics, and institutional integrity. Because she is passionate about history and the cultural record, Kelly is interested in how intellectual freedom affects access to genealogical records and ethical collecting practices in archives. In her free time, Kelly enjoys listening to podcasts (especially Ear Hustle) and watching old movies (like Lady from Shanghai). Find her on LinkedIn.
Nice post Kelly. A follow up to your line “Even if the data is effectively anonymized…” that a lot of the library learning analytics datasets I’ve seen are NOT anonymized (even if the study authors believe that they are); all the data to back this up is in this study: http://doi.org/10.7710/2162-3309.2268. This means that there are currently practical security issues with learning analytics in addition to the ethical concerns you mentioned.