In 2011, six online learning institutions pooled their data to see if they could identify patterns related to student loss. This project grew to become the Predictive Analytics Reporting (PAR) Framework, a national collaborative that uses predictive analytics to improve student performance.

The PAR Framework will become an independent organization in December. "We believe the time is right to establish a not-for-profit practice improvement center that supports using learning analytics in the service of student success," says Beth Davis, co-founder and managing director of the PAR Framework.

The PAR Framework's membership consists of traditional universities; community colleges; for-profit, public, and private institutes; and online and blended learning institutes. Its data pool is growing, currently containing more than 2 million student records and 18 million course records.

Partner institutions receive localized predictive models from PAR that have been constructed from their own student data and normed to the national sample, and have access to benchmarking reports that help them see how they compare with cohort institutions on various dimensions of student performance.

Ellen Wagner, chief research and strategy officer for PAR, says that PAR is "actively linking predictions to interventions and then looking for measures of impact so we can see which interventions work well and under what circumstances."

Using predictive analytics to inform student intervention programs requires a cultural shift for higher education, however. "Higher education has a culture of authority-driven decision making," says Wagner. "Using data to support better decisions demands a culture of evidence-based decision making."