Using Big Data to Improve Learning
Downtown Pittsburgh / photo: Rob Long
More city governments are using data to improve city services, efficiency, and transparency. Can “big data” do the same for education?
“Big data” is everywhere these days. You can’t go far without reading another article about the benefits of using data to improve our lives. I attribute the craze to Freakonomics, and its “answer anything with data” message. Whatever the source, as the costs of analyzing reams of data keep dropping, people are awakening to the realization that they have a gold mine at their fingertips.
As New York Times reporter Alan Feuer noted in “The Mayor’s Geek Squad,” cities are getting in on the game big time. New York City, for example, is using big data to make city services more efficient and transparent and to find new ways to serve residents by tracing the daily data breadcrumbs we leave behind as we eat, complain, drive, walk, turn down the thermostat, throw out the garbage, park the car, swipe our transit cards, or any of the other mindless, quotidian acts.
With the data they can track commuting habits to find bottlenecks and other issues. They can monitor parking tickets to figure out traffic flow. They can identify dangerous crosswalks (or mid-streets in the case of New Yorkers). They even found restaurants that were dumping grease into the city sewers, causing countless headaches and stopped up sinks.
“I think of us as the Get Stuff Done Folks,” the leader of the team told the Times. “All we do is take and process massive amounts of information and use it to do things more effectively.”
As cities get into the act, one wonders, what could “big data” do for education?
Quite a bit apparently—and several companies are launching several new tools. Richard Nieva reports at Online.Edu on one such tool, Desire2Learn, which he reports, has raised $80 million in funding. The tool does two things. It helps a student pick classes that he or she is likely to succeed in, and thus avoiding wasting time and money (in college at least) of taking a course you’re bound to flunk.
As Nieva writes:
It will, for instance, tell a liberal arts type how he will likely fare in an engineering class by scouring his past classwork (or high school transcripts if he’s a freshman) and compare his academic record to other students who have taken that class. Baker claims it can predict if a student will pass or not with 90 percent accuracy and even settle on his letter grade with 92 percent accuracy.
At lower grade levels, it gathers data on how a student is actually doing in a class “and spots red flags like a bad grade on a quiz, or, more subtly, rushing through an online assignment.”
The New York City public school system, the University of Arizona, the University of Memphis, and the Harvard School of Business have all signed on, according to Nieva.
Other tools help teachers tailor lessons to individual students. As student progress through lessons, the computer program sends information to the teacher about each student so he or she can customize the lesson to that individual learner. Without such instant insights, it “would be impossible to modify an in-person lecture to account for every individual student’s needs, but technology can help deliver a customized level of education.”
One tool, “CourseSmart,” writes David Streitfeld:
goes further by individually packaging for each professor information on all the students in a class — a bold effort that is already beginning to affect how teachers present material and how students respond to it, even as critics question how well it measures learning.
The program lets teachers know when “students are skipping pages, failing to highlight significant passages, not bothering to take notes — or simply not opening the book at all.” Streitfeld says, “Engagement information could give the colleges early warning about which students might flunk out, while more broadly letting teachers know if the whole class is falling behind.” But the teacher, or the department, or the school as a whole is not the end of the information chain: “Eventually, the data will flow back to the publishers, to help prepare new editions.”
Here in Pittsburgh, Vincent Aleven and his colleagues at Carnegie Mellon University are developing Intelligent Tutoring Systems. Darrell West, writing at the Brookings Institution, thinks that these tools can help instructors see the process of learning unfold for students.
These types of computer tutorials can evaluate problem-solving approaches and provide feedback along the instructional path. The system sends error messages if the student follows an incorrect approach and provides answer hints if requested by the student. Instructors can get a detailed analysis not just of whether the student reached the final answer correctly, but how they solved the problem.
And beyond the classroom, big data could help parents and others identify failing schools. Chicago School Select, for example, a recent winner of a Knight Foundation contest for improving how citizens and government interact, is a personalized decision tool for parents choosing from the myriad of public schools. This web application allows parents to check off what they want in a school and then rank and compare schools on the basis of those attributes.
But not all are so enamored. The collected mass of data on individual learners could follow them – and in some cases haunt them—for life. And data is inherently biased, despite claims otherwise. Data without design is just numbers after all, and humans are the ones still designing the data analysis tools. And as we know from centuries of science, human bias inevitably creeps into any design. Similarly, as Microsoft Research’s Kate Crawford writes at the Harvard Business Review blog, “Data are assumed to accurately reflect the social world, but there are significant gaps, with little or no signal coming from particular communities.”
No doubt more and better approaches to using big data are on the horizon as the younger generation takes the reins.
“Young people, because of social media, have always felt they’ve had a voice,” said Jennifer Pahlka, the executive director of Code for America, a volunteer group of techies that helps city governments write code for public projects.
“They’re coming from the assumption that government is a hackable system — an operating system that can be optimized. It’s in their DNA, and they just go and do it.”