Using Teacher Data to Drive Education Reporting
In the Minneapolis Public Schools, nearly two-thirds of the district’s enrollment are students of color. Additionally, 65 percent of the district’s more than 35,000 students qualify for free and reduced-price meals. Beth Hawkins, a reporter for the MinnPost, had a hunch that the best-paid local teachers were working in the wealthiest schools, teaching white students. But this was just a guess, and her colleague at the nonprofit news site, data editor Tom Nehil, wanted to see the numbers.
Reporters often assume they already know the story, Hawkins joked in front of an audience of Education Writers Association members at an October seminar on the teaching profession. On the flip side, Hawkins said, data journalists — whom she’s used to working with at the nonprofit news site — tend to doubt there even is a story.
Armed with a list of every district teacher – complete with date of hire, type of license, the building they teach in, the demographics of the workforce and of course, salaries – the journalists paired up to discover whether students in poor schools were getting the district’s lowest-paid (and least-experienced) teachers. Their analysis revealed that this was, in fact, the reality.
The city’s best-paid teachers did cluster in schools that serve the wealthiest families, and that have large concentrations of white students. The story made waves in Minneapolis, especially at the school board level.
For journalists thinking about using data-driven reporting in their education stories, here are a few tips from the MinnPost team’s “How I Did the Story” session at the Detroit Seminar:
Start with one question
Hawkins and Nehil knew they wanted to follow the money, so their story began with this query: Are the best-paid teachers concentrated in the city’s wealthiest schools? Given the amount of data they had, the team could have tried to determine whether the city’s highest-paid teachers are in fact the best educators. However, “we very intentionally did not go the experience and effectiveness route,” Hawkins said. “We felt that was a separate argument.”
Sticking to one question will help focus your reporting and help you determine which parts of your data set you should be using and why. A finished story that makes one argument is also less likely to overwhelm data-shy readers.
Clean up your data
The problem with data, Nehil explained, is that it’s rarely ever ready to use when you first receive it. The MinnPost’s initial scatter plots didn’t prove much of anything because the results were “all over the place,” Nehil said. There were some wealthier schools that had highly paid teachers, but there were also some poor ones with expensive staff. The plot was “a salad of wealth and poverty,” Nehil explained, with no observable patterns.
After removing part-time teachers from the data and taking out some specialty schools (which have expensive and uncommon programs), the data became a lot clearer. The key with cleaning up the data, though, is to be transparent so that readers know how it has been presented. In the final story, the reporters stated what wasn’t included in their analysis and made their data set fully available.
Getting access to documents and determining how to read and present them can take time, so allow yourself at least as much time as you would for a normal story. Data journalists are fond of saying that data, like people, have to be interviewed extensively before a story can emerge. The time spent cleaning data, asking questions, and going over graphs is “just another form of reporting,” Hawkins said.
Enlist some outsiders
Just as it’s always a good idea to fact-check a narrative story for accuracy, it also pays to have another person check your math. The MinnPost team took their results to several outside experts including an education policy statistician – and asked them to give their early work a look-over. If experts are reluctant to give their advice, allow them to do so off the record.
Identify any assumptions
Though it ended up being true that Minneapolis public schools with a larger share of students from low-income families tended to have lower-paid, less-experienced teachers, the reporters realized that this didn’t necessarily mean these teachers were the district’s worst. For example, a young teacher makes less money because she’s had fewer years on the job, not because she’s not as good at it. That’s why the reporters were careful to say upfront that the disparity their analysis revealed raises questions as well as answers them.
Watch what happens
Pay attention to how people interpret your findings, whether they agree with your methods, and even whether your work leads to political action. After MinnPost published this story, the school district responded by running its own numbers. Those results showed a less dire disparity, but the reporters quickly found out that the district used a different method of calculating averages.
Nevertheless, the original story did make a significant impact. District staff have since made efforts to address inequities by moving toward a weighted student funding system where dollars should more closely follow students to schools.