The struggle to protect students鈥� privacy while making use of the data collected on them in school has for years been focused on the role of outside companies.
But while that debate has raged in Congress and statehouses across the country, K-12 school systems in more than a dozen cities and counties have quietly begun linking children鈥檚 educational records with data from other government agencies, covering everything from children鈥檚 mental-health status to their history of child-welfare placements and their involvement in the juvenile-justice system.
Proponents say that such intergovernmental 鈥渋ntegrated data systems鈥� can yield powerful insights that promote a more holistic understanding of children鈥檚 experiences. They point to an emerging track record of the information being used to improve policy, service delivery, and program evaluation.
Take, for example, Allegheny County, in southwestern Pennsylvania. After learning that 14,450 Pittsburgh public school students鈥攎ore than half the district鈥攈ad been involved in county human-services programs, officials there have worked to analyze the experiences of homeless students and children in foster care. They鈥檝e initiated a new cross-sector effort to combat chronic absenteeism. Child-welfare caseworkers will soon receive weekly email alerts when children in their caseloads get suspended or miss multiple days of school. And district officials or school counselors and social workers could get similar notices when one of their students shows up in a homeless shelter, runs afoul of the law, or is moved from his or her child-welfare placement.
鈥淚t鈥檚 been transformational in understanding how a community, school districts, and other child-serving government agencies can come together to support kids,鈥� said Erin Dalton, the deputy director of the county鈥檚 human-services department鈥檚 data-analysis, research, and evaluation office.
The administration of President Barack Obama and the U.S. Department of Education are both part of a growing national push for those kinds of data-sharing arrangements. But clearing the legal and technical hurdles to create such systems is difficult.
Turning the resulting data into better policies and fresh practices is even harder.
And the privacy concerns associated with integrated data systems鈥攊ncluding potential breaches, the creation of inaccurate or misleading profiles, and possible stigmatization of children鈥攁re immense.
Because of privacy concerns, schools have in some places declined to participate in integrated data systems altogether, or agreed to share information only on a project-by-project basis.
But now, even concerned privacy advocates acknowledge that schools鈥� full participation in integrated data systems is likely permissible under federal law.
A big reason for the shift: The U.S. Department of Education has over the past four years helped develop and articulate a complex legal rationale that districts can use to justify their involvement under the country鈥檚 main student-data-privacy law, known as the Family Educational Rights and Privacy Act, or FERPA.
The key is that schools 鈥渉ave to conceive of these integrated data systems as a multistep process,鈥� then justify their involvement in each step using a different exception written into FERPA, said Michael Hawes, a statistical privacy adviser at the department.
69传媒 may share identifiable education records with an external warehouse for linking with other administrative data under the so-called 鈥渟chool official鈥� exception to FERPA, Hawes said. That provision allows a district to share educational records with third parties as part of outsourcing a service that it lacks the capacity to perform itself.
Then, schools may authorize approved researchers to analyze 鈥渄e-identified鈥� versions of those linked records under FERPA鈥檚 鈥渟tudies鈥� or 鈥渁udits and evaluation鈥� exceptions鈥攂oth of which are frequently cited by districts when sharing student information with researchers for discrete projects.
Written agreements must spell out detailed terms. Each research project must be determined to have an instructional benefit for the participating schools. Districts remain ultimately responsible for making sure the education records they disclose are adequately protected.
In Allegheny County, Pa., that means strict security protocols. Access to the full integrated data warehouse is typically limited to fewer than 10 individuals, most of whom are IT workers at the human-services department, said Erin Dalton, the deputy director of the county鈥檚 human-services department鈥檚 data-analysis, research, and evaluation office.
Participating districts have little or no visibility into individual children鈥檚 linked records; typically, only aggregated and 鈥渄e-identified鈥� information has been shared back to schools and other agencies by the department.
And while human error has led to a few minor issues鈥攕tudent or client information being faxed or emailed to the wrong person, for example鈥攖here have been no significant breaches dating back well beyond the six years that schools have participated in the Allegheny County warehouse, Dalton said.
Furthermore, Allegheny County officials stressed, when it comes to using linked administrative records that have not been scrubbed clean of identifying markers to offer supports and interventions to individual students, nothing is done without consent.
The department is the legal guardian for many students in the county. Under FERPA, the department has extensive rights to access the education records of children in foster care and other court-ordered placements.
For all other students, parents or legal guardians must give their signed approval.
鈥淭here鈥檚 a strong feeling among parents and citizens that we want to minimize the amount of data collected and shared among agencies, not maximize it,鈥� said Leonie Haimson, a New York City parent and activist who was instrumental in the takedown of inBloom, a $100 million effort to build an educational-data warehouse that ultimately closed over privacy concerns.
More Data Than Ever
Without question, the collection and sharing of data about children is on the rise, in both the private and public sectors.
69传媒 have been flooded over the past decade with digital devices, software, and apps, all of which generate massive amounts of information about students鈥� behavioral and learning habits. Lawmakers and school officials have struggled to keep up.
At the same time, all 50 states have developed statewide longitudinal-data systems that contain anonymous information on students in preschool through college. Nineteen of those systems now link educational data to workforce information, and at least 30 link education records with records from at least one other public agency.
Cross-agency integrated data systems, meanwhile, go even further. Most are local, operating at the county or city level, often with a university partner playing a key role. Most focus primarily on the 鈥渁dministrative鈥� data held by various government agencies, often with a focus on human-services records. All contain far more sensitive information than what is generally collected by an app, school system, or state education department alone.
The linked records contained in integrated data systems are stripped of any identifying information about a specific child before being made available to researchers for analysis. Some systems, however, have begun pushing the envelope by seeking parental consent to use identifiable individual-level information to target supports to specific children.
All told, 11 states and about two dozen cities and counties have established integrated data systems, according to an ongoing survey by a coalition of nonprofit groups led by the National Neighborhood Indicators Partnership. Among the established sites, 14 cities or counties and five states include the personally identifiable education records of individual K-12 students, according to the survey.
Federal support for that movement intensified not long after Obama took office in 2009.
鈥淭here was a recognition by the Obama administration that we were collecting huge amounts of data in every program, but it was incredibly compliance-oriented, with each program looking only at its own activities and not learning anything,鈥� said Kathy Stack, who played a key role in the rise of 鈥渆vidence-based decisionmaking鈥� during her tenure in the federal Office of Management and Budget.
The Pittsburgh Model
The federal government has helped foster that growth by issuing technical and legal guidance to public agencies and by dangling federal dollars for initiatives that use linked administrative records.
Major philanthropic groups such as the John D. and Catherine T. MacArthur Foundation and the Annie E. Casey Foundation have also supported the work, giving millions of dollars to both individual integrated data systems and to an entity known as Actionable Intelligence for Social Policy.
Based at the University of Pennsylvania, AISP, founded by professors Dennis Culhane and John Fantuzzo, serves as the hub for a network of 12 integrated data systems around the country that focus primarily on research.
Allegheny County鈥檚 system is widely regarded as among the most advanced.
Educational involvement in the initiative there began in 2009, when the Pittsburgh public schools agreed to provide to a data warehouse maintained by the department of human services a wide range of information about its students, including name, home address, grade point average, standardized-test scores, attendance records, and involvement in special education programming.
The resulting revelation that nearly half the city鈥檚 students were also involved with human-services agencies was 鈥渁 turning point in our relationship with the district,鈥� said Dalton of the human-services department.
In November 2012, the memorandum of understanding between the Pittsburgh schools and the department was updated to allow the district greater access to human-services records. Similar agreements were soon signed by 17 additional area school systems.
Superintendent Alan Johnson described the impact that joining the county data-sharing partnership in 2012 has had on the 4,000-student Woodland Hills district.
Researchers from the human-services department and elsewhere have identified previously unknown connections between bullying, the onset of health problems related to childhood asthma and diabetes, and school attendance and performance, Johnson said. The information has helped prod Woodland Hills school staff to look for underlying problems when they see an uptick in the number of students missing days or needing treatment for such conditions.
The county researchers have also analyzed suspension data to pinpoint the specific streets and blocks where high school students who frequently get in trouble live, Johnson said. That information has been used to target security and prevention resources and to initiate new collaborations with local law enforcement.
SOURCE: Allegheny County (Pa.) Department of Human Services
And Woodland Hills will be one of at least five Allegheny County districts to take part in the system鈥檚 new email-alert program, which aims to provide human-services caseworkers with more timely and easily accessible information so that they can better support regular school attendance.
鈥淲hether we like it or not, schools are now involved in all these domains of a child鈥檚 life,鈥� Johnson said. 鈥淵ou do a good job with that by learning as much as you can about the young people who come through your doors.鈥�
Privacy Concerns Abound
Similar success stories have emerged from integrated data systems around the country.
In New York City, for example, an analysis by the Center for Innovation through Data Intelligencee of integrated child-welfare and juvenile-justice records led to the realization that children placed in foster care at age 9 were far more likely to end up in trouble with the law than children placed in infancy or early childhood. Child-welfare officials are now targeting resources and supports accordingly.
Such examples are all well and good, said Khaliah Barnes, a lawyer with the Electronic Privacy Information Center, a Washington-based advocacy group. But integrated data systems raise a host of significant privacy concerns, she said.
For one, parents and students have no legal right to opt out when schools decide to share their education records with integrated data warehouses. (Identifiable student information cannot, however, be disclosed by the warehouse to researchers or others without parental consent.)
In addition, Barnes asked, do parents and students have the right to access their own information and challenge anything that is inaccurate? Could linked records be used down the line in unanticipated ways鈥攆or example, to prosecute truant students or their parents?
There is also good reason to believe, she said, that information about a child鈥檚 history of drug and alcohol treatment, or abuse in the home, or previous arrests will be just as likely to result in that child being stigmatized as being provided with smarter services.
As a result, Barnes concluded, 鈥渢here should be less intrusive ways to achieve this end goal.鈥�
Technical, Human Challenges
Privacy-related hurdles are not the only, or even the most, daunting challenge that public agencies face when trying to establish integrated data systems.
Significant technical capacity is also required to deliver educational and other data to a central warehouse, construct the algorithms that are typically used to match individual records from multiple systems, and establish the security protocols necessary to adequately safeguard the linked information.
Dozens of cities and counties around the country are setting up 鈥渋ntegrated data systems,鈥� in which schools share data with other public agencies in an effort to improve education and social services.
SOURCE: Ongoing survey by Actionable Intelligence for Social Policy, the Data Quality Campaign, the National League of Cities, and the National Neighborhood Indicators Partnership
And what鈥檚 truly hard, experts say, are the human challenges: getting and keeping the right people from the right public agencies around the same table and helping policymakers and practitioners make effective use of the insights that integrated data systems produce.
That is particularly true when money is scarce.
鈥淵ou only make data actionable when you put it into the hands of practitioners who have the resources, knowledge, and capacity to do something with it,鈥� said Diane Castelbuono, the deputy chief of early-childhood education in the Philadelphia school system. That district is currently weighing whether to re-engage with the city鈥檚 integrated data system, roughly five years after its regular involvement stopped following budget cuts and ongoing leadership churn.
Despite the barriers, it鈥檚 clear that integrated data systems are on the rise. At least 16 cities and counties, from Buffalo, N.Y., to Fresno Calif., are preparing to launch such systems on their own, according to the survey by the National Neighborhood Indicators Partnership and others.
鈥淟inkage of data both longitudinally and horizontally is on the rise,鈥� said Michael Hawes, a statistical privacy advisor at the U.S. Department of Education. 鈥淲e have been seeing it at the state level for awhile, and now we鈥檙e seeing it at the local level because of the richness of the data they have available.鈥�
