全球智库信息集成服务系统

Higher education data are now more widely available than ever before. In addition to institution- and program-level data, nearly every state has, or is in the process of developing, a student-level longitudinal data system that can follow students from kindergarten through college.

Despite having an abundance of data, it can still be difficult to understand how well state colleges serve students. The production and dissemination of new data have focused on the needs of potential students and their families, which is less useful for state policymakers. Further, higher education data tend to conflate institutional quality with student characteristics. Institution-level metrics fail to account for differences in students’ academic preparation before college and their families’ financial situations, both of which affect outcomes.  

How Can We Better Measure an Institution’s Student Outcomes?

Researchers at the Urban Institute worked with policymakers in Virginia and Connecticut to understand their data needs in measuring college performance and how we can meet them:

• Use student-level data to measure how well colleges serve their students. Adjusting data to account for such characteristics as academic preparation, family income, race, and ethnicity can help policymakers understand which institutions and programs have the greatest impact on students and allocate resources accordingly. The data suggest academic preparation explains a large part of the gap in graduation rates across four-year institutions, whereas race and ethnicity are more significant factors at two-year institutions.
• Assess how to build accurate program-level completion metrics. Developing program-level graduation rates is complex. To build accurate metrics, colleges may need to use multiple years of data or push students in four-year institutions to declare a major earlier. Further, policymakers must consider how to count students who change majors but still graduate from the institution.
• Create an index of earnings measures to better assess postenrollment outcomes. In-state wage data, the most common earnings information today, tend to be lower than national data because they cannot follow students across state lines. Using multiple measures of earnings, we show how the interpretation of earnings depends on when earnings are measured and for which students. The best way to understand student outcomes may be to use multiple earnings measures.

What Can Adjusted Student-Level Data Tell Us about Equity Gaps?

Adjusting student-level data for student characteristics such as family income and college readiness can help researchers and policymakers identify the causes of inequity in college graduation rates between minority students and white and Asian students.

The research findings suggest that much of this graduation gap is the result of students’ experiences before they enter college. Gaps in college readiness and financial circumstances before matriculation account for an estimated 60 percent of the racial gap in graduation rates at four-year colleges in Virginia, with similar results in Connecticut. Racial segregation between colleges, which cannot be explained by admissions criteria, account for an additional 30 percent and 15 percent of the graduation gap in Virginia and Connecticut, respectively.

To address these inequities, we must adjust and contextualize student-level higher education data. With better college performance measures, researchers can produce more relevant and actionable information for policymakers.