Achieving big data privacy in education

While big data and learning analytics could be promising in educational assessments and individualizing student learning, as educators, we may also pay attention to the problem of students' privacy in using this technology. In the K-12 context, students are mostly minors and it is an ethical obligation to protect their rights and privacy. The law also requires that the students' educational records be confidential (see 20 U.S. Code §1232g). Students may also be more vulnerable to various harms -- physical, psychological, etc. So while learning the potentials of big data, I did some research in big data privacy and want to recommend this article (Reidenberg and Schaub, 2018) to my peers. The following is a brief summary of the article.

Reidenberg and Schaub (2018) summarize some risks of big data on student privacy:

1. Inappropriate use or disclosure of private information

2.  Surreptitious monitoring of every mouse click and page load can create fear and put the students in a stressful situation as if their every move is under surveillance. 

3. Biased data or algorithms (with their complexity and invisibility) may disadvantage certain learners or undermining individuality. 

4. By charting a path for learners, big data in education may curtail the opportunities for self-discovery and leave very little space for learners to chart their own paths. 

5. Data ownership and consent.

After pointing out these problems, the authors discussed possible safeguards to these risks. Their suggestions include:

1. Technical mechanisms to assure transparency about data collection, processing, and use.

a. Effective Consent. The authors argue that the terms of service agreements and privacy policies are not enough. They suggest the short-form notices and privacy indicators be integrated into the learning system's user interface. Students and parents need to have the options to opt-out and control the granularity of data used. 

b. Accountability for analytics algorithms and algorithmic decision making. The authors suggest that the learners, instructors and other stakeholders need to be able to inspect the assessment and decision making, receive validation of the legitimacy of the automated decision and intervene when the decisions are problematic. 

c. Securing and protecting learning analytics data as sensitive data. The authors point out that with the mass amount of data taken based on profiles, there is the potential to re-identify the individual students. They also suggest that encryptions, de-identifications and ensuring appropriate access controls. 

2. Organizational safeguards: limited and appropriate access, an assessment to make sure that the educational benefit weights over the harm, and minimizing the harm.

3. Legal safeguards: 

a. Defining "educational record" and "legitimate educational uses"

b. Pre-requisite assessment of educational and privacy impact.

c. Criteria for public procurement.

After reading this article, as well as other articles on the possibilities and the benefits of big data, my attitude towards big data changes from positive to mixed. While big data has such strong functions in making predictions and providing more "holistic" assessments comparing to test scores, it also troubles me because of the exposure of privacy, and the complexity and the invisibility in the algorisms. While the authors put more emphasis on the technical mechanisms, my concerns are more about the legal and policy aspects. Comparing to the speed of tech development, law-making and policymaking may take a longer time. And while it is not difficult to move a technology from one country to another, the policymaking and the law system are vastly different across countries. From this perspective, the privacy issue in big data in education may be more complex.