With the growing demand for improving college completion rates has come a need for more thorough information about just how well or poorly colleges and their students are performing on a variety of measures. In a growing number of states, that data is being used to improve the number of students who finish their degrees. Some states and higher-education systems are even connecting data about individual students to the jobs they get after college, to determine the average wage earned by graduates of particular programs. Several nonprofit groups are involved in pushing states and institutions to dig deeper and connect the dots between data and policy, and the U.S. Department of Education provided $250-million through the federal stimulus bill to help states analyze what is working in education from preschool through college. But the process of collecting, reporting, and using that information to guide policy and practice turns out to be complicated by limited and inconsistent definitions of whom to count as students, by bureaucratic hurdles, and even by institutional resistance to accountability. The problem with using data to inform and improve completion and graduation rates is not an inability to collect relevant information. All but a handful of states and higher-education systems can track individual students in some way, through financial-aid data, for example. But one issue is that the academic performance and completion of most students in higher education isn't being counted. Much of information that public colleges are reporting on completion is only on full-time students who are enrolled for the first time--only a quarter of all college students. That leaves out the 40 percent of students who attend public institutions part time as well as all the students who transfer to another college. A related issue is that educators and states have struggled to find common definitions of who should be considered a degree-seeking student and even which students should be considered full time or part time. Students at private, nonprofit institutions also go uncounted by most states. Only 19 states follow the college outcomes of such students. Making policy based on data is also difficult within and across states because of bureaucratic hurdles. Many states have more than one system that collects data on college students, such as a coordinating or governing board or a state agency. In addition, states and higher-education systems make it difficult for researchers to study student performance and completion data because only 27 states will release that information, even when a request meets the federal requirements meant to protect sensitive information about individuals.