COVER STORY: Numbers Crunchers
Concept by Mark Bartley/CCWeek
C O V E R S T O R Y
Promise, Pitfalls of Big Data Dominate Innovations Conference
By Paul Bradley, Editor, Community College Week
DALLAS — When contemplating higher education’s struggles in collecting and effective leveraging data, it’s useful to take a look at Amazon.
Anyone who has ever ordered from the online retailing giant is familiar with the Amazon experience. Buy an item, or even take an online glimpse at it, and you’ll quickly get a list of similar items that could pique your interest. With its state-of-the-art, lightning-fast data analytics, Amazon can look at your buying history, analyze your interests and deduce the kinds of items you’ll buy. It’s the key to Amazon’s success, central to the company’s business model.
Higher education collects and disburses reams of administrative data: enrollments broken down by age and gender; persistence and graduation rates; student progress and attainment rates; the percentage of students who arrived on campus in need of remedial education, to name just a few.
But community colleges collect and analyze precious little of the kind of data that can be used to directly advance the sector’s overarching goal: propelling greater numbers of students to graduation or the acquisition of a useful workforce credential. Colleges are slow to get the data where it needs to be — into the hands of faculty and students. Data that can help students in real time is lacking.
That was the central message of a featured session at the 2013 Innovations conference sponsored by the League for Innovation in the Community College. The discussion was guided by Mark Milliron — former League president and CEO and now chancellor of Western Governors’ University — who used the Amazon analogy to illustrate how higher education has been slow to leverage the kind of data and technological tools that are fundamentally reshaping retailing, health care and social networking.
The bulk of how higher education handles data focuses on easily attainable information, he said. That’s fine for academically talented students and elite universities, he said. For community colleges, it just won’t do.
“If we want to serve the harder students, we have to get the harder data,” he said. “We have to move from easier to harder data and get it to the front lines faster.”
The Innovations conference, held at the sprawling Hilton Anatole hotel, attracted more than 1,200 community college educators from the around the country. They were able to attend dozens of workshops focusing on new ideas and practices in learning and teaching; workforce preparation; research, assessment and accountability; sustainability; open educational resource and five other areas.
But while the workshops were divided into ten “streams,” two topics seemed to dominate the conference: data analytics and massive open online courses. MOOCs are the latest big thing in higher education, potentially reshaping teaching and learning.
Effective use of data might hold even more promise, creating an ability to track students on a case-by-case basis and design individualized interventions. Big data analytics could allow colleges to identify which students are at risk, in real time. Administrators could look at groups of students and see which techniques or resources work best with which students. Students themselves could identify their strengths and weaknesses and improve their course selection and hone their career paths.
Colleges currently focus mostly on administrative data. That’s a little like unfolding a road map after a journey has already gone awry, Milliron said. You can certainly figure out what went wrong, but by then it’s too late.
Louis Soares, a senior fellow at the Center for American Progress, delivered one of the keynotes at the conference. His research at CAP includes community college reform, post-traditional learning demographics and technology-based innovations in higher education.
He said deep data analytics hold the promise of improving outcomes through the creation of individual educational pathways. Data collected through course selection, financial aid applications, participation in study groups and other digital interactions can go well beyond traditional quizzes and tests in revealing whether a student is on the correct educational path. The problem for colleges is packaging all that information together into a coherent whole, he said.
“We are only now beginning to learn how to use that data,” Soares said. “We haven’t been able to unpack and repack the data to look at groups of students or individual students. We don’t yet know how (data analytics are) going to change the college experience.”
The changes could be profound, allowing students to customize their own educations.
“It will really call into question things that have been in place for 150 years, the value of a baccalaureate degree and why it’s structured that way,” he added.
Many colleges already have embraced data-based decision making at the administrative level. They base course offerings, for example, on labor market information rather than anecdotal evidence.
But colleges don’t collect enough data on student learning, said Nicole Melander, chief technology officer for Achieving the Dream. ATD is in the midst of a decade-long mission to improve community colleges through data-driven reforms. Melander said her work has shown that while colleges have lots of administrative data ripe for analysis, learning data is much harder to come by.
The potential for collecting learning data through online and hybrid learning models and digital tools used in classrooms is vast, said Charles Thornburgh, founder and CEO of Civitas Learning, a new company that strives to use predictive data analytics to improve academic outcomes and keep college students in school. The company is working with a network of colleges to build an extensive cross-institutional data set on student learning.
Civitas hopes its work will reveal which resources work best for which students and what kind of innovations improve student learning. The company believes that the smart use of technology can help identify which students are at risk of dropping out. It can alert educators about which courses and degree paths are contributing most to attrition. It can aid students in creating individual learning plans based on their strengths and interests.
But the difficulties for colleges in using deep, rich data run deep. The barriers are many, Thornburgh said.
“Education is among the top ten sectors in terms of volume, but substantial systemic barriers in using it persist,” Thornburgh said. “The data that we know about and talk about is quite limited. It’s data exhaust. It’s not consumable or actionable.”
Translating mountains of data into insight — and transforming those insights into action — is difficult, he said. Data from various closed systems must be aggregated and analyzed. Data scientists and educators must work together to glean insights and recommendations and deliver them in a consumable format to faculty, students, and staff.
Some colleges feel hamstring by the Family Educational Rights and Privacy Act (FERPA), which poses severe restrictions on the kind of information institutions can collect and share. Other institutions have a kind of data anxiety, fearing that relying too much on data will remove the human element from the educational process. Smaller colleges often don’t have the institutional research capacity to crunch the numbers.
It’s critical that colleges overcome those barriers, said Diana G. Oblinger, president and CEO of EDUCAUSE, whose keynote address opened the Innovations conference. Just as technology has reshaped how people buy products and plan travel, so it must change higher education, she said.
“It’s no longer about us being destinations,” she said. “It’s about creating pathways to possibilities.”
“For education to get better, we can’t keep doing things the same way.”