Can Software that Predicts Performance Help Kids Learn?
In acquiring a maker of wireless education technology, News Corp. may help validate the market for assessment tools.
In the 1990s, two professors at the University of Oregon developed a series of one-minute tests for elementary readers called Dynamic Indicators of Basic Early Literacy Skills, or DIBELS. The professors claimed that their tests and software could predict whether a child was on track to master reading at grade level by year’s end or was likely to fall behind.
Many reading instructors objected that the tests seemed to have little to do with reading comprehension. Instead, they measured skills such as rapid reading and pronouncing nonsense syllables. But the creators cited research demonstrating a strong correlation between performance on those tests and on longer, standardized exams.
DIBELS began catching on in districts across the country, eventually being adopted in 43 states. For the past several years, the drills have been available on handheld testing devices made by Wireless Generation, a New York-based company that has become a big supplier to New York City’s million-student school system.
Now, education analytics suddenly look poised to become a big business. In November, News Corp. agreed to acquire Wireless Generation for $360 million, and chairman and CEO Rupert Murdoch cited the technology as a way to tap into a giant but underdeveloped market. “When it comes to K through 12 education,” he said in a press release, “we see a $500 billion sector in the U.S. alone.”
That’s the total spent on teacher salaries as well as educational products and services. The emerging market for educational analytics is only a small slice of that, but it is already being cultivated by some of the world’s largest software vendors. Microsoft, IBM, and SAS Institute have all adapted their business intelligence software to help measure and predict the performance of K-12 students. The same programs that gather data about how companies serve customers and markets is now being applied to evaluate how schools serve students and meet larger educational goals. Microsoft cites early success analyzing student performance in the Charlotte-Mecklenburg School District in North Carolina. Meanwhile, SAS says its Educational Analytics Suite helped improve classroom performance in the Liberty Public School District in Missouri. In Houston, the 100,000-student Cypress-Fairbanks district is using IBM’s Cognos analytics to help administrators keep students in every class on track to pass an important annual state exam.
But since it often takes years for schools to change practices and adopt new technologies, the market has been taking off slowly. That’s why the News Corp. acquisition can be seen as validation that classroom use of predictive analytics constitutes a real business opportunity. “We’ve come a very significant way,” says Roland H. Good III, the University of Oregon education professor who, with Ruth Kaminski, developed DIBELS.
In a growing number of districts, analytic tools are in use daily. In New York City, a few classrooms are using a system that analyzes each day’s performance and creates a customized lesson plan for the next day that combines in-class, small-group, and online learning. On a larger scale, administrators are creating data warehouses that can be mined to determine which students are in danger of dropping out.
Still, many teachers and education specialists are skeptical. Some reading specialists complain that DIBELS encourages children to read fast without comprehension. Kenneth S. Goodman, a University of Arizona professor emeritus and an expert on reading, blasted the approach in a book, The Truth About DIBELS. “The focus on improving performance on DIBELS is likely to contribute little or nothing to reading development and could actually interfere,” he wrote. Other critics worry that frequent testing will stigmatize children who do badly.
But proponents say that’s a misuse of the tests. Larry Berger, president of Wireless Generation, says that analytics help produce customized lessons for each student. He notes that the cycle of assessment, analysis, assignment to groups, and lesson creation repeats every 10 days, so that teachers can keep tailoring lessons individually. The software enables almost endless customization, he says.
In some cases, these predictive assessment tools are challenging older ways of doing things. New York City used the technology while developing the curriculum for its experimental School of One program. The city now supports online distribution of open-source early reading materials and lesson plans, in competition with traditional textbooks. The Wireless Generation software can quickly assess each student’s progress every day, and it uses a learning algorithm to create the next day’s customized “playlist” of recommended activities. The list may include instruction by a teacher, online video training, collaborative group work, or one-on-one tutoring delivered live or online.
Some charter schools are big believers in using analytics to shape instruction. Mosaica Education, a New York-based operator of 80 schools around the United States and overseas, requires teachers to administer the tests every Friday to assess how well students understand the material they’ve been studying. Michael J. Connelly, the company’s CEO, says most students take them online in the morning, and teachers study the results over lunch.
The assessment tells the teachers which students should be grouped together for small practice circles on particular issues. And because the results can be viewed online, principals in Mosaica schools can monitor teachers throughout the year and offer counseling if a class is slipping.
There’s a certain irony in the way these kinds of data-driven tools are being used. In the past, administrative computing was attacked for treating all students as “just another number.” But now, says Wireless Generation’s Berger, it’s all about customizing the curriculum, by finding “a predictive way to figure out what’s the right thing to teach that particular student.”
Become an MIT Technology Review Insider for in-depth analysis and unparalleled perspective.Subscribe today