• Background
• Goals and Design of the Feasibility Study
• Proposed Redesign of IPEDS
• Challenges to Implementing a UR System
• Conclusions

This report examines the feasibility of implementing a student unit record (UR) system to replace the student-related components of the Integrated Postsecondary Education Data System (IPEDS). The feasibility study was initiated by the National Center for Education Statistics (NCES), a part of the Institute of Education Sciences (IES) within the U.S. Department of Education (ED), in response to growing interest within the postsecondary education community for more accurate measures of net price and graduation rates, especially measures that take into account institutional mission and student mobility. This interest parallels a growing congressional desire to hold postsecondary institutions accountable for student outcomes.

This discussion of the feasibility of a UR system at the federal level is occurring within the context of the development of other UR systems for students attending postsecondary institutions. Unit record systems are maintained by most colleges and universities to track registration for courses, academic performance, degree and certificate completion, financial aid, and other purposes. A number of states began to develop UR systems in the mid-1980s and use UR data for analysis and program evaluation. Today, 39 states have at least one student UR system. A limitation of state UR systems, however, is that most do not include data on students attending private institutions, or students who leave an institution and transfer across state lines.

Many governmental and other organizations also maintain UR systems on specific groups of students. For example, the National Student Loan Data System (NSLDS) within the office of Federal Student Aid (FSA) compiles information on all recipients of federal student loans, including verification of enrollment by academic term. In addition, the National Collegiate Athletic Association (NCAA) collects UR data on 1,800 institutions with Division I, II, or III varsity athletic programs, and about 2,800 colleges and universities currently contract with the National Student Clearinghouse to perform enrollment verification and other services using student UR data uploaded from member institutions.

At IES/NCES, IPEDS is the core postsecondary education data collection program, designed and implemented to meet its mission to report on the condition of postsecondary education in the United States. IPEDS is a single, comprehensive system that encompasses over 10,000 institutions whose primary purpose is to provide postsecondary education (including roughly 6,700 institutions that have Program Participation Agreements with ED for Title IV federal student financial aid programs and are required by statute to report to IPEDS). IPEDS collects institution-level data in the areas of enrollment, program completions, graduation rates, faculty, staff, finances, institutional prices, and student financial aid. The use of aggregate data has some limitations in comparison with UR data, such as the inability to track the academic progress and experiences of individual students, and therefore to study the longitudinal enrollment of different types of students.

Despite its comprehensiveness, IPEDS cannot measure many of the evolving trends in postsecondary education that are necessary for sound policy decisions. The current IPEDS framework cannot accurately capture changing enrollment and completions patterns in the postsecondary education sector, especially given increasing numbers of nontraditional students, and cannot describe the prices various types of students face after financial aid is taken into account. To do so, it would be necessary to collect accurate student-level information on persistence systemwide (i.e., regardless of institution and nationwide), multiple enrollment, part-time enrollment, transfer, and attainment. It would also be necessary to collect student-level information on prices and financial aid, in order to calculate net prices that take into account the individual circumstances of each student. By its very nature, a UR system would enable the collection of data that would lead to more accurate estimates of these variables. In addition, a UR system would allow the development of a whole range of new measures, such as net prices for specific groups of students, graduation rates that take into account institutional missions, persistence rates that consider student mobility and a systemwide perspective, measures of enrollment patterns for nontraditional students, and time to degree by field of study.

In exploring the feasibility of a UR system, the study attempted to investigate whether such a system could be constructed technically and effectively, given the knowledge about UR systems already accumulated at the state and institutional levels. In addition, the feasibility study tried to explore whether such a system should be developed by the federal government. To do so, the study solicited input on several dimensions, including privacy and confidentiality, institutional burden, coordination, technical issues, and timing.

As part of the feasibility study, three Technical Review Panels (TRPs) were designed to gather feedback and ideas from different perspectives related to the study, and included representatives from the following groups: (1) states, state systems, private systems, and private associations of colleges and universities; (2) institutions, particularly institutional researchers and registrars; and (3) other stakeholders, including the national postsecondary education association community, federal agencies, units within ED, and vendors such as administrative information system developers. In addition, the contractor developed an architecture and flow of operations for a proposed student UR system, as well as a list of potential data elements that might be collected under such a system.

In reading this report, it is important to keep in mind that any redesign of IPEDS to develop a UR system would require legislative authorization through amendments to the Higher Education Act (HEA) and funds would have to be appropriated by Congress to implement the system.

If authorized and funded, the proposed UR system would replace the student-related components in the current IPEDS collection—Fall Enrollment, Completions, Student Financial Aid, and Graduation Rates—as well as the price-of-attendance variables collected in the Institutional Characteristics component. The UR system would be designed to include all of the variables necessary to replace those components and calculate institution-level estimates for the Peer Analysis System (PAS). The collection process for nonstudent-related components in IPEDS would remain the same.

It is difficult to describe exactly what the UR system would look like before the design process is undertaken. Such a process would involve numerous TRPs and input from campuses, university systems, and state coordinators, particularly from states with UR systems. Generally, the UR collection system would be designed to collect individually identifiable data through files that are submitted electronically by institutions. The files would be used to calculate institutional summary totals for each school, with information about enrollment, completions, graduation rates, financial aid, and price. Four types of files would be submitted:

• Header files: These data provide individually identifiable information such as name, Social Security Number (SSN), date of birth, address, race/ethnicity, and gender that are attached to an individual student's record. These files would be required at least once for every student. New header records would be submitted as needed to document any changes in these key data.
• Enrollment/term files: These data include program information such as number of courses and credits attempted, major field of study, start and end dates, and attendance status. The files would be required three to four times a year, and institutions would be allowed to upload files more frequently if they wished.
• Completions files: These data include information on degree completions and the date of completion. The files would need to be uploaded at least once per year.
• Financial aid files: These data include information on financial aid received from federal, state, and institutional sources. Information on price of attendance would also be included with the financial aid file. These data also would need to be uploaded at least once a year.

For each submission of data, the IPEDS keyholder at an institution or coordinating agency would submit data electronically through the IPEDS collection system, similar to the process that exists currently. After submission, NCES would review the data to make sure they are consistent within the file and with prior submissions. Schools would work with the IPEDS Help Desk to match all records, and any that do not match would have to be resolved. The UR data would then be summarized in online institutional reports, which would also be checked for consistency, before the keyholder "locks" or finalizes the submission.

The UR data would then be moved from the collection system to the permanent database storage system. The full UR database would only exist in this permanent storage area, which would not be accessible via the Internet and would be subject to high IES/NCES levels of protection for confidentiality and security. Ultimately, aggregate estimates would be calculated from the full UR database and moved to the PAS, where they would be stored as institution-level data.

Individually identifiable data would remain within the permanent storage system. The only allowable redisclosures of individually identifiable data would have to be specifically authorized in the HEA legislation, including

• Enrollment verification for the National Student Loan Data System (NSLDS): The UR system would be used to verify enrollment for students who are receiving federal student loans. Currently, this verification is being done either by institutions themselves, or by organizations such as the National Student Clearinghouse.
• Verification of subsequent enrollment to the IPEDS keyholder: The UR system would be used to redisclose individually identifiable data back to the initial keyholders and to state/system coordinators, in order to give something back to institutions. Data on the subsequent enrollment of students who left the first institution in the previous year would be redisclosed to the keyholder, including the institution of subsequent enrollment, date, attendance status, attainment, and date of attainment.¹
• Record mismatches: During the process of data collection for the UR system, mismatches between data records and other types of edit failures would have to be resolved. This would involve sending individually identifiable information back to the IPEDS keyholder. These types of edit failure resolutions would be essential to the data integrity of the database.

Other uses of the data would not involve the disclosure of individually identifiable student information. For example, while ensuring the confidentiality of the data, NCES could generate aggregate reports for the Office of Postsecondary Education (OPE) using the UR data (e.g., to generate aggregate measures of persistence, transfer, and attainment for various types of federal student aid recipients, such as those attending on a part-time basis). It would also be possible to add new derived variables to the PAS, used by institutional researchers and other analysts. Each of these derived variables would be reviewed for potential disclosure risks prior to their release on the PAS. Such variables could include new definitions of net price; new measures of graduation rates that better take into account the missions of postsecondary institutions and the mobility of students across institutions; new definitions of time to degree, including transfer calculated for various fields of study; variables that describe enrollment by field of study and program length; and completions by field of study.

Technically, UR could be done at most institutions in the long term, after investment of time and financial resources. This can be inferred from the fact that 39 states have compiled UR systems in some form; thousands of postsecondary institutions already submit UR data electronically to private organizations; and postsecondary institutions that are Title IV participants are required to upload information on federal aid recipients to the FSA. Nonetheless, in feedback from institutions, states, associations, and other stakeholders, it is clear if a UR system is legislatively authorized, certain concerns must be dealt with and resolved in the design phase of implementation.

Privacy and confidentiality

Concerns have been raised about student privacy and the confidentiality of individually identifiable student data under a federal UR system. ED, IES, and NCES have always taken seriously the importance of safeguarding student data, but a UR system raises questions about students' rights to withhold or control personal information. This is particularly the case for students who do not receive federal student aid. However, these students benefit indirectly² from federal student aid funds, which support all programs, and benefit directly from state appropriations at public institutions and the tax-exempt status of private, not-for-profit institutions. Additionally, data on nonaided students are a critical element to compute graduation rates, retention measures, and other indicators. Information on nonaided students would be necessary in order to compare these measures with information on students receiving student aid.

In addition to misgivings about student privacy, there are practical, technical concerns about unauthorized access to the data by hackers and identity theft. This is particularly true given the proposal to use SSNs as one of several personal identifiers that are necessary for matching student records. The use of SSNs would be essential to a UR system to accurately link together student information on financial aid, enrollment, and completions, as well as records from various institutions. Enrollment verification for the FSA already includes the use of SSNs as a student identifier. An additional measure of enrollment intensity at the start of each term (such as full- or part-time) would also be collected to satisfy FSA requirements.

Despite these concerns, IES/NCES is well suited to protect the data, given the strict limits of the legislation regarding data confidentiality under which it operates. IES/NCES legislation protects the privacy of individuals, making wrongful disclosure a Class E felony punishable by up to 5 years in jail and a $250,000 fine. NCES has experience in working with individually identifiable data through its various sample surveys, and has created the structures and procedures necessary to prevent unauthorized disclosure of such data. In fact, there are no cases where individually identifiable data collected by NCES have been wrongfully disclosed by an employee, a contractor, or a restricted licensee, or of cases in which hackers have breached IES/NCES firewalls. If collected, the data would be technologically protected and secure, and would not leave NCES unless allowed by law. Under the Patriot Act, the Attorney General and the Department of Justice could conceivably obtain access to UR data in order to fight terrorism. Students on whom data are held would be able to "opt out" of the redisclosure of subsequent enrollment information.

Institutional burden

The additional burden of a UR system can be divided into two categories: initial implementation and subsequent operations. The burden of initial implementation is expected to be higher than the costs of subsequent operations. A field test would be necessary in order to make sure the system works, to anticipate and address problems that would be encountered, and to develop all necessary features in the system prior to implementation. About 1,200 to 1,500 institutions would be required to participate in the field test and report using both the old and new IPEDS collection system. Although NCES would make every effort to notify selected institutions early, participating institutions would need to make changes in their reporting systems within a relatively short time frame, depending upon the desire of Congress for an implementation schedule.

In the full-scale implementation, many institutions would need to upgrade information technologies and assign staff to comply with new reporting requirements. Staff would need to be trained in the use of these systems and the details of reporting procedures. Some institutions would need to rely on vendors to provide upgrades to existing software, build their UR extracts, or pay for changes to legacy information systems. These additional activities would likely increase software costs. Obtaining historical GRS files for all cohorts in the first year would present a burden (although these same files are needed now to calculate the GRS locally). The initial burden on small institutions is likely to be relatively high, unless the institutions are part of a larger system or state association.

The additional costs of subsequent operations under a UR reporting system are expected to be lower than the costs of initial implementation. Keyholders would need to coordinate with offices on campus to gather data, run internal checks to make sure data make sense, submit data to NCES several times per year, and work with the IPEDS Help Desk to reconcile record mismatches and discrepancies in data. Some mismatches of records could be difficult to resolve, especially if there are numerous records.

It is very difficult, at the conceptual stage, to make cost estimates with any degree of precision. Costs would differ widely among postsecondary institutions, depending on whether they are in state UR systems, whether they currently upload to organizations such as the National Student Clearinghouse, whether they use local or proprietary administrative information systems, and the extent of their information technology and institutional research capability. There would be a decrease in burden after the initial implementation of a UR system, as postsecondary institutions would no longer need to track and maintain records on GRS cohorts for 6 years or fill out the current IPEDS student-related components.

If a UR system were implemented, it would be important to take into account these various issues during the design phase of implementation so as to minimize institutional burden. There are different ways to offset the cost and burden of a UR system. One funding mechanism, Administrative Cost Allowances (ACAs), is used to help defray the cost of administering federal student aid programs.³ A similar funding mechanism could be put in place for a UR system.

Technical issues

Technical issues were also raised as potential challenge to the implementation of a federal UR system. The proposed system would include the creation and maintenance of a database of millions of student records, with new records added every year. In addition, the system would require the uploading of large files from postsecondary institutions to NCES, using multiple forms of security to protect against unauthorized disclosures of data. NCES currently has most of the hardware and software necessary to implement a UR system, including current equipment used in the web-based IPEDS collection as well as servers capable of storing large amounts of student data. One necessary addition would be database storage, to be located offline in a secure site and protected by physical and software firewalls.

There would likely be greater technical challenges for postsecondary institutions, with the extent varying among the registrar, institutional research, and financial aid offices, which sometimes utilize different and incompatible information systems. Institutions using both legacy and proprietary student information systems would need to make software conversions or updates. For the smallest schools, an Excel template could be provided to collect data and generate the data file needed for submission. Although the technical issues could present a problem, these schools currently find a way to do uniform reporting for FSA financial aid eligibility and NSLDS loan deferment.

The proposed UR system would also use XML⁴ technology for the submission of data files to NCES, although it is likely that ASCII files would be accepted in the early years of implementation. Some postsecondary institutions have already adopted XML and are using it in their exchange of data with other organizations. On the other hand, many institutions do not currently use XML and training would be required on the use of this technology. Nonetheless, the FSA has already mandated that institutions begin submitting data to the office using XML by 2005-06.

Coordination

Coordination of the flow of information presents a multitude of challenges in implementing a UR system. For example, a UR system might not work well within the existing IPEDS structures in some states. Most state systems are based on specific census dates. If multiple header and/or enrollment files need to be submitted at different points in time to capture total enrollment, this would involve a change in workload for both institutions and systems. Special TRP meetings should be held during the UR design phase in order to leverage existing UR systems whenever possible in order to meet federal and state/system requirements and needs. This will prevent unnecessary duplication of effort and reporting, and ensure that any federal UR system maximizes the lessons that have been learned through years of state UR reporting.

Timing

In implementing a UR system, the timing of data collections would have to be addressed. If a UR system were authorized in 2005, a field test would then be administered in 2006-07, followed by full-scale implementation in 2007-08. The project timetable is designed to yield data relatively quickly while avoiding potential problems associated with an expedited time frame. A phased implementation could also be considered to provide additional time to address problems during implementation. To respond adequately as part of the field test, it might be necessary for institutions to examine the utility of their administrative information systems for the purposes of producing UR extracts and to address some of the burden issues mentioned above such as training and staffing. Early notification for the selected institutions would be crucial for the institution's ability to respond in a timely and accurate fashion. It is possible that NCES could draw the sample of institutions immediately after legislative authorization to allow selected institutions almost a year to prepare.

Since the UR system is based on individually identifiable records, it must comply with the Office of Management and Budget (OMB) requirement for collecting race/ethnicity data with a two-question format. A byproduct of the UR system is that schools that have not yet implemented this change will need to do so to meet OMB Statistical Policy Directive No. 15, Race and Ethnic Standards for Federal Statistics and Administrative Reporting.

Another important issue is operational—how to time data collection schedules, while minimizing conflicts with other reporting schedules. The proposed UR system likely would collect enrollment records once per term. However, some institutions do not have standard terms; for example, courses may be offered on a rolling basis or on 6-week terms. Institutions could choose to upload data more frequently, especially for the purpose of enrollment verification for student loan programs. It would be necessary to find a method of specifying a whole range of flexible term reporting options, perhaps by asking institutions to document all possible term sequences using the IPEDS Institutional Characteristics component.

Degree and certificate completions would likely be collected with only one file per year, although institutions with several commencement periods might wish to submit multiple files over the year. In some cases, awards are recorded months after the relevant students have stopped attending institutions; degree dates then reflect the date the degree was awarded rather than when the degree was finished. In designing the timing of data collections and the periods of reference for the data, it would be useful to align the completions data with the enrollment data necessary to calculate graduation rates so that completions records can be matched to comparable enrollment records.

Student financial aid information also would likely be collected in only one file per year. Data submitted in an academic year would be from the previous year's award cycle. It would be important to time the collection of financial aid data so that it does not conflict with the institution's aid packaging period, which is the busiest time of year for financial aid offices. In addition, the treatment of summer sessions varies by institution, especially regarding whether summer sessions would follow or lead the submission of an annual data file.

All of these timing issues would be addressed during the design phase of UR implementation, should a UR system be authorized. In the proposed UR system, collection schedules would not need to be on a uniform schedule, but rather could be geared to a schedule that works best for individual institutions. In other words, institutions with different calendars or financial aid packaging schedules could submit data to NCES on different cycles.

As this report has outlined, a central question for a UR system is "Could it be done?" Have the information technologies and infrastructures at the campus and state levels matured, could the current IPEDS web-based reporting system be adapted to a UR system, and would there be adequate technical and legal protections in place at IES/NCES? The report has addressed some of the technical and system problems associated with the design and development of a new IPEDS UR system. At the technical level, a UR system could be done at most institutions given time for implementation.

The feasibility study also addressed the "Should it be done?" question, providing a framework for the discussion of issues inherent in this question. These issues constellate in several areas of concern—privacy, burden, coordination, technology, and timing—which would need to be addressed and resolved in the design phase of a UR system, should policymakers decide to authorize and fund such a system.

Finally, the feasibility study outlined areas of federal interest: better information for informed consumer decisions, including the improved calculation of net prices; and more accurate measures for institutional accountability and program effectiveness, including enrollment, persistence, transfer, and attainment rates by program of study. Policy-makers would be able to monitor in real time federal student aid programs (such as Pell Grants) and variations in aid packaging. The study also has attempted to highlight some potential benefits to institutions, researchers, consumers, and other users of NCES data.

The study did not attempt to address every challenge or make recommendations about how each aspect should be addressed. Nor did the study document specific organizational positions regarding the obstacles a UR system might face. Rather, it provided a framework for policymakers to understand the potential costs and benefits of a UR system as they discuss whether it should be considered.

The central defining question of the feasibility of a UR system in IPEDS is not a "could" question. It is a "should" question, asking whether the federal government should develop a system that is based upon individually identifiable information about enrollment, financial aid, and attainment. This system would, for the first time, give policymakers and consumers much more accurate and comprehensive information about postsecondary education in this country. Some of the benefits of a UR system include the collection of new data that would measure the success rates of students at institutions to which family and federal student aid monies flow, provide more accurate consumer guidance, and improve federal programs that support those families and students. In addition to benefits, the feasibility study found a number of significant issues that would need to be overcome before a UR could be implemented, including objections about student privacy, confidentiality of data, new institutional burdens, coordination within and of institutions, and timing issues.

¹Redisclosure of student information to the original institutions could take place over a longer time period if this was decided by future design TRPs and NCES.

²Tuition at these schools is probably lower than it would be if they were not the beneficiaries of tax-exempt status and state appropriations.

³Institutions currently receive over $150 million in ACAs, which is provided to help cover the cost of administration of federal programs such as Pell Grants and campus-based aid.

⁴XML is a "markup language," or mechanism for identifying structures within a document or data file. It employs tags to identify data elements, thereby facilitating the seamless exchange of data. In other words, it allows users to describe data and deliver it across a network, through the creation of common records across disparate databases.