3. Where the university is
an app on your phone a police officer your employer your physician your coach
a security camera a shop assistant your landlord your instructor your isp
your gym a restaurant you frequent your facebook friend your counselor And more...
4. Where a Spectrum of Student Data Maybe Collected
Social
Media
Location
Data
Application
Data
Financial
Aid
Family
Data
MCard
LMS
Wearables/
Biometrics
4
Deidentified/
Aggregated
5. So What?
So U-M like any other organization wants to leverage big data and
data science to
● Transform teaching and learning
● Create personalized learning opportunities and pathways
● Engage in earlier and better interventions
● Provide for great mentoring and advising
● Lead to better student outcomes
6. But?
Will there be
● Loss of student privacy
● Curtailment of student growth and development
● Increase in student conformity
● Muting of free expression and intellectual curiosity
8. 1890. Recent inventions ... call attention to the next step which must be taken for the
protection of the person, and for securing to the individual ... the right to be let alone.
(Samuel Warren & Louis D. Brandeis, 1890)
1967. Privacy is the right of individuals to control, edit, manage, and delete information
about themselves, and to decide when, how, and to what extent information is
communicated to others. (Alan Westin)
1977. Building and maintaining an enduring, intimate relationship is a process of privacy
regulation. (Irwin Altman)
What is Privacy?
Privacy Defined (?)
8
11. Value of Privacy
Civil Liberties Freedom of thought, speech, expression
Freedom of social and political activities
Freedom of association
Limits authority
Individual
Liberties
Promotes individuality
Ability to grow and change
Autonomy and control over self
Allows for ability to reveal as much or as little about self
Ethics & Respect Promotes ethical and respectful treatment of others
Compliance FERPA, HIPAA, Common Rule, GLBA, COPPA, Red Flags, International laws,
etc…
12. Privacy Protection Challenges
In a big data world of pervasive data collection the ability to be “let
alone” is impossible
Traditional privacy protections may not work:
● Notice/Awareness: Informing an individual when data collection is taking place
● Purpose: Data collected for one purpose is only used for that purpose
● Consent/Choice: Allowing the individual to consent for their personal data to be
used for the purpose provided in the notice or allowing an individual to opt-out of
data collection
● Access/Participation: Allowing the individual to review, correct, update their info
● Integrity/Security: Keeping the data secure and accurate
● Enforcement/Redress: Mechanisms to challenge data accuracy and outcomes
12
13. Learning Analytics Privacy & Ethical Concerns
Not just a difference in scale of data
● Aggregation of anonymous or de-identified personal data can result in the creation
of identifiable data
● Data may be used beyond the intended purposes for which it was originally
collected
● The widespread collection, analysis, and sharing of data may run counter to the
words or spirit of existing privacy policies or the institutional culture
● Perception (or reality) of surveillance can change adversely change student behavior
● Traditional privacy protections & processes may not work.
● Predictive data does not always mean accurate data.
● Risk of human element losing out to algorithms
● Profiling can be made easy
17. Big Benefits from
Learning Analytics
U-M is a leader in the new field of learning
analytics, which offers great potential for:
● Personalized learning
● Individualized support
● Improved student learning outcomes
● Improved advising/mentoring
● Improved graduation rates
Big Responsibility
● Representatives from the Office of the
Registrar, Academic Innovation, and
Privacy Office reviewed numerous privacy
and ethical frameworks, as well as
emerging learning analytics principles and
frameworks.
● The result is a set of proposed U-M
learning analytics guiding principles.
● Principles will inform actions.
● Commitment to transparency, shared
governance, and community engagement.
Student Data, Student
Services, Student
Outcomes
● We collect data about students to provide
them services, but these data are largely
not combined with other data.
● Examples of data collected includes:
○ LMS data (Canvas)
○ Application data
○ Financial aid
○ MCard
○ Location data...And more
● Additional data, including some that is
available publicly, could be collected in the
future (e.g., social media postings,
LinkedIn profiles, biometrics).
Privacy & Ethics
Concerns
● Data may be used beyond the purposes
for which it was collected.
● Aggregation of anonymous or de-identified
data can result in identifiable data.
● Traditional privacy protections may not be
feasible.
● Predictive data does not always mean
accurate data
● Algorithms will make decisions, not
people.
● Profiling can become easy.
● Lack of student awareness.
● Practices may run counter to existing U-M
Privacy & Ethics
Proposed U-M LA Guiding Principles
Respect
● Ensure learning analytics is for the good of the learner, their institution, and improving higher
education.
● Ensure algorithms do not replace human interaction.
Transparency
● Disclose to students that learning analytics is a legitimate institutional interest
and that certain student data will be made available to appropriate parties for research and
pedagogical improvements.
Accountability
● Create and enforce policies and processes that appropriately ensure security, privacy,
quality, and proper stewardship of student data.
Empowerment
● Use personally identifiable information to provide services
that inform or benefit the specific individual whose data
is used.
● Provide students visibility and insight into collected data,
as well as the ability to question data accuracy.
Enabling the Student
Data Revolution
18. University of Michigan Approach - Guiding Principles
● Respect
○ Limiting access via request process; providing information back to student
● Transparency
○ Planning to update FERPA statement; create web materials for guiding principles
● Accountability
○ Collaborating across U-M (Academic Innovation, Registrar, U-M Privacy Officer)
○ Define Guiding Principles
● Empowerment
○ Exploring outreach, engagement, and education opportunities
● Continuous Consideration
○ Revisit models, positions, decisions in order to improve
19. U-M Approach - Emerging Tangible Outcomes
● Guiding Principles based on established and emerging privacy and
ethical frameworks (and considerable stakeholder input/review)
● Multiple internal and external presentations on guiding principles
● Building privacy considerations into U-M developed ed tech tools
● Ensuring data protection agreements are in place when
integrating external services into Canvas
● Research into student perceptions and attitudes around privacy
and learning analytics
● Updating the U-M FERPA statement (http://ro.umich.edu/ferpa/)
19
21. Proposed U-M Guiding Principles (full)
Principle Accounts for Responsibilities
Respect Respect for the rights and dignity of
learners
U-M will:
● Ensure Learning Analytics is for the good of the learner, their institution, and improving
higher education
● Ensure algorithms do not replace human interaction
Learners will:
● Participate through contributing data to Learning Analytics research and programs that will
benefit themselves and their fellow students
Transparency Transparency
Purpose Specification
Data Minimization
Use Limitation
Respect for the rights and dignity of
learners
U-M will:
● Disclose to students that learning analytics is a legitimate institutional interest and that
certain student data will be made available to appropriate parties for research and
pedagogical improvements
Learners will:
● Know what information is collected about them; how it is used; and who it is shared with
● Understand that they are part of a community that constantly seeks to improve approaches
to teaching and learning for themselves and other current/future students
Accountability Security
Quality and Integrity
Accountability and Auditing
U-M will:
● Create and enforce policies and processes that appropriately ensure security, privacy,
quality and proper stewardship of student data
Learners will:
● Have the ability to inspect and review their learning analytics data
● Have the ability to question the accuracy of that data
22. Proposed U-M Guiding Principles (full) - continued
Empowerment Individual Participation
Use Limitation
Respect for the rights and dignity of
learners
Learners will:
● Know what information is collected about them; how it is used; and who it is shared with
● Have the ability to ask to view their learning analytics data
● Have a defined process to question data accuracy
● Make choices about whether certain types of their personally identifiable data may be used
for LA purposes
U-M will:
● Only use personally identifiable information to provide services that either directly inform or
benefit the specific individual whose’s data is used or answer specific questions that will have
a concrete and measurable impact on improving teaching and learning at U-M.
Continuous
Consideration
U-M will:
● Always seek to revisit models, policies, decisions in order to improve
23. What to Do: Plan & Transparency
Plan
● Make sure appropriate stakeholders, such as general counsel, data owners, data
users, are engaged as part of the planning process
● Develop Guiding Principles
● Provide training and guidance to those accessing the data.
Be Transparent
● Provide info on how data is being collected, used, and how privacy is addressed.
● Communicate the benefits of data collections and usage.
● Be cognizant of the privacy controversies related to government and private-sector
data collection/usage practices
24. What to Do: Use R-E-S-P-E-C-T
Respectfully Use the Data
● Avoid identification. Use aggregate, anonymized or de-identified data as a first
choice
● Be judicious. Strive to limit your data collection to what is necessary for the sought
after outcome
● Respect the context. Avoid using data collected for one purpose for different
purposes unless you have appropriately planned and provided transparency
● Give Back. When feasible, provide analytical information back to the “subject.”
● Limit access. Not all data users need access to all the data. Develop processes that
only share the appropriate data sets
● Resist profiling/Keep the human element in mind. Do not allow data to entirely tell
the story or drive decisions
25. What to Do: Empowerment
Empower the “Subject”
● When appropriate and/or possible, ask for consent before collecting data
● When appropriate and/or possible, provide opportunities for the subject to opt-out
of some of the data collection or usage (opt-out should not be assumed to be
unworkable)
● Provide reasonable access for individual data review and correction
26. Consider 4 R’s*
Reuse Data collected for one purpose is used
for anotherRepurposing
Recombination Anonymous, de-identified, aggregate
date leads to re-identificationReanalysis
*M. Steinmann, S.A.Matei and J. Collmann, “A theoretical framework for ethical reflection in Big
Data research,” in Ethical Reasoning in Big Data, J. Collmann and S.A. Matei (eds), Springer series
in Computational Social Sciences
27. (some) Resources
● Frameworks
○ Asilomar Convention
○ Big Data Dialog - Data Scientist's Code of Ethics
○ Big Data Ethics Initiative - Unified Ethical Framework Part A: Unified Ethical Frame
○ Learning Analytics Community Exchange - DELICATE Checklist
○ Open University Analytics Student Data Principles
○ Student Data Principles
● White Papers/Articles/Presentations
○ 2016 UC Summit on Analytics for Institutional & Student Success (University of California, 2016)
○ Big Data: Seizing Opportunities - Preserving Values (White House, 2015)
○ Big Data in the Campus Landscape (Educause ECAR Working Group Paper, 2015)
○ Privacy in the Age of Big Data (NASPA, Summer 2015)
○ Taming "Big Data": Using Data Analytics for Student Success and Institutional Intelligence (Trusteeship
Magazine, 2015)