Education researchers across the country are always in search of more data. They all have access to resources like the National Student Clearinghouse (NSC), but each state has different models of collaboration and data sharing. A recent example in Colorado showed how state bureaucracy can sometimes be at odds with academic research.
Against that background, the Michigan Education Data Center (MEDC) serves as an example of best practice. And the teams at the Education Policy Initiation (EPI) are willing to share their experience with other institutions around the country.
Data architect Kyle Kwaiser went on one such mission earlier this year, spending three days at University of North Carolina (UNC) at Chapel Hill. The visit was initiated by Steve Hemelt, who did his post-doc research at U-M, working with Brian Jacob and Michigan education data.
“He knew that we had matured in our relationship with the state, and was interested in, I wouldn't say replicating, but establishing something of a similar nature in his research program and those of some of his colleagues out of UNC Chapel Hill,” says Kwaiser.
Over the three days he met with education researchers, research center directors, and graduate students to give an overview of the foundational aspects of managing the research data and the workflows required.
The most immediate aspects of what they were looking for was: how MEDC is processing those data? What kind of data sets are being built? What kind of practices go beyond just the code to ensure the documentation and the quality of the research output?
Hemelt and his collaborators also wanted to learn more about how EPI works with the Michigan Department of Education, so that they can try to establish a similar relationship with North Carolina's Department of Public Instruction.
They also invited in a number of directors and associate directors from different education-related organizations around the state.
“I think one of the goals was just to get folks together, and directly informed about what's going on in the state of Michigan, so they can start to see what could happen if they collaborated and started sharing data among systems,” Kwaiser says.
The MEDC is unique in the scope and the span of the data, accumulated over 20 years and including information from pre-K all the way to post-secondary.
Kwaiser feels it is a part of EPI’s mission. “We've always been quite open with our practices and our code, which is actually quite a valuable resource. Having figured out how to process data, for example, multiple student records in a given year down to a single record per student, which types of variables to build out in terms of what will fit well into regression models or analysis workflows, is the type of transferable knowledge we accrued over years of experience.”
He also invited Heather Handley, the longitudinal data unit manager of Michigan’s Center for Educational Performance and Information (CEPI) to share how the state agency handles and integrates NSC data, how they select what they send, how they manage things year to year, and how they build their reports.
Overall, the trip was a success. "While it's not possible to replicate everything we are doing at MEDC because each state, like North Carolina, has slightly different data systems., it definitely helps that there are examples for them to work from."