Big Data Invades the Classroom
Inside classrooms, more student data is being mined than ever before. Is the personalized learning it aims to support worth it?
Server farms of data and armies of analytics gurus have streamlined online shopping, helped drivers avoid traffic jams and dictated how much guacamole goes into burrito bowls. Now, algorithms are at work in schools, attempting to revolutionize learning. Software engineers tabulate every student keystroke and government officials report every test score, hoping to make education more personalized and more equitable.
The programs are constantly tracking your student, and kids’ academic behavior is constantly being analyzed.
But do all these metrics add up to a better, more personalized educational experience? And do the risks — privacy breaches; students being profited on — outweigh the benefits?
Like the never-ending data sets, the answer seems to be constantly changing: We don’t know. Not yet. Only with enough teacher support. Only for some students.
The one thing we do know: Your student will, increasingly, be interacting with big data.
The promise of big data
Educational analytics have long promised to deliver the holy grail of teaching: personalized learning. One of the hallmarks of the century-old Montessori movement, personalized learning is the concept of teachers addressing the specific needs and interests of individual students, rather than teaching the entire class the same predetermined concepts. In theory, students get more out of their education if it is tailored to their specific needs.
But customized learning requires planning individual lessons for every student, every day. Few teachers have the resources to do so. Enter big data — a term that generally refers to large data sets that can be analyzed by computers to reveal patterns, trends and associations — and online learning platforms, powered by that data.
Online educational nonprofits such as Khan Academy and CK-12 are already present in most schools, with their videos and lesson sets. For-profit companies such as Knewton and Pearson Education have assessments linked to lessons that collect student knowledge. Every answer clicked, every second between problems and every tap of the help button immediately recalibrate a student’s learning profile. Automatically, the next perfect piece of content, tailored to the student’s continuously measured need, is always delivered to each student. There is no needless repetition and no information given that students aren’t ready for.
Sounds tantalizing, doesn’t it? Just stick students in front of a computer and let the videos and analytics do the teaching. These platforms have led to an explosion of online charter schools, serving kindergarten through high school. Traditional high schools also purchase programs for credit recovery, letting students redo a failed class online.
Results have not been good.
The 2015 National Study of Online Charter Schools found that “students of online charter schools had significantly weaker academic performance . . . compared with their counterparts in conventional schools.” After spending a year of learning online, students are nearly a year behind their peers.
And it’s not just lower achieving students attending school online who have negative results. A recent study out of the Northwestern University School of Education and Social Policy found that “high-achieving eighth-graders who took algebra online performed worse than similar students who took the course in a traditional classroom.”
Clearly students need more than online content, no matter how personalized or data-driven it is. Blending online resources into traditional classrooms has been relatively successful, but systems to collect data, analyze student knowledge and personalize instruction are still in their infancy.
That is, unless your school is supported by Silicon Valley funds. Google, Facebook, Amazon and Microsoft, for example, are heavily invested in education analytics, and these large, data-driven companies, and some of their high-profile executives and philanthropists, have been generous contributors to some charter schools built on personalized learning.
Teachers versus data?
Charter and private schools are often touted as the champions of big data, but their leaders are careful to note that the teachers, not the data sets, drive powerful instruction. Summit Public Schools, a California charter and recipient of Facebook funds, has recently moved to Washington state.
“We believe in the power of a school community coming together. In no way should technology or data replace the power and magic of a teacher-and-student relationship,” says Summit chief regional officer Jen Wickens. “Teachers are supported with the 21st-century tools that we have.”
And they are well supported with tools. Facebook has donated software engineers to work with Summit teachers on a Personalized Learning Plan (PLP) dashboard.
“For us, personalized learning doesn’t mean just technology and data,” Wickens says. “The majority of the student day is spent making meaningful connections with teachers, mentors and other students.”
“Fully online schools tend to show very poor results, while personalized learning schools show very promising results. Summit Public Schools are personalized learning schools, not online schools,” says Robin Lake, the director of the Center on Reinventing Public Education.
At Summit Sierra in Seattle’s Chinatown–International District, 10th-grader Sammie* explains the approach: “Each project has steps you work through. You work at your own pace, but you still have to keep up. You turn something in on the PLP, the teacher will go through it, make comments . . . and then you get a little green wrench.” She points to assignment boxes on her laptop. “These boxes will turn this really scary shade of red when they are overdue, so you get really motivated.”
There are benefits to the method, says her classmate Jackson*: “I’m independent and visual; I like seeing all the steps ahead of me and taking care of them.”
Both students say they also appreciate the differences in testing. At schools they previously attended, Sammie and Jackson say tests were given to the class and then everyone moved on. At Summit, students take a diagnostic pretest, and then focus only on objectives not yet mastered with the help of resources that teachers suggest via the PLP.
For students, this interactive learning map is the place to prove what they’ve learned and move ahead on new topics. For teachers, the PLP is a data-driven gradebook. With a quick glance at a screen, teachers can see exactly what their students know. Summit Sierra science teacher Camden Hanzlick-Burton says the visual data allows him to be more specific when deciding who and what to reteach.
“We can get more granular with the objectives for content assessments,” he says. “That’s where the data is really useful, because [then] we can dictate if [reteaching] is happening as a whole class, small workshop or after school hours. Technology is not doing anything that teachers haven’t already known. It’s not revolutionary, but more efficient.”
This personalization of the learning and reteaching process is something teachers always strive for. Teachers don’t need online platforms or computers to know students fail tests, but in order to figure out exactly what each student didn’t understand, a teacher would have to break down every question and create a chart to show which students understood which concepts. Many teachers do this by hand or know it intuitively. The PLP makes it automatic.
But does this data equate to better outcomes for students? Naturally, Summit says yes. Its website reports its California schools are ranked in the top 20 percent of public high schools in the state. The 2015–16 school year was Summit’s first in Washington, so no data has been reported to the state yet, but according to Wickens, Summit Sierra students outperformed the national average in both math and reading on the Measures of Academic Progress (MAP) assessment.
However, it’s hard to peg statistics solely on big data. Summit schools, which are public, receive Facebook resources, have the autonomy to hand-pick teachers and are responsible for only one-fifth the number of the students a traditional high school has. Also, comparing charter and traditional schools is problematic. Summit accepts students through a random lottery, but students still have to complete an application process that is separate from Seattle Public Schools. “We work very hard to secure a diverse and high-needs population,” Wickens says, “from door-to-door canvassing, tabling at community events, phone banking and presentations at food banks and low-income housing areas.” Summit Sierra’s population of free and reduced lunch students last year was 56 percent, compared to Seattle’s average of 40 percent.
While video footage could be a useful tool, many are wondering about the implications of students growing accustomed to constant surveillance.
Surveillance, screen time and profit
Summit schools aren’t the only places where software engineers crunch numbers and design dashboards for teachers. In 2013, Max Ventilla left Google, where he’d been working on personalizing search terms (aka how Google knows what you want before you do). Ventilla raised $100 million to start up AltSchool, a private pre-K–8 network. Whereas Summit has PLP, AltSchool students get a customized “playlist.”
Teachers at both schools use data and personalized lessons, but Ventilla’s quest to turn teachers into “data-enabled detectives” goes a little further: AltSchool records every single classroom moment through a surveillance system called AltVideo.
Carolyn Wilson, teacher and director of educators, explains that AltVideo allows teachers to travel back through their day: “We can go back to a lesson and see which students were hesitating, which ones wanted to speak up but didn’t, which ones looked distracted.” When children are learning to write, AltVideo allows teachers to analyze minute details, such as how close their eyes are to the paper.
“If I see something I want to think [about] more deeply later,” Wilson says. “I can bookmark a moment to look at after school or during my planning time, or I can take it across the hall to my colleague who may have a specialty in math or reading.”
According to Wilson, AltVideo is only used by teachers and the school says there are no engineers looking at videos to get aggregate data. The school will not provide details about where or how videos or stored.
Districts around the country are discussing student safety and protection from data breaches, and some have experienced data breaches already. The U.S. Department of Education, nonprofit groups and education consortiums have begun issuing best practices, rules and guidelines to safeguard student data and help districts navigate as they seek to bring more educational technology online. But contrary to being resolved, privacy concerns around personal data are likely just beginning to heat up.
Another concern: While video footage could be a useful tool, many are wondering about the implications of students growing accustomed to constant surveillance. Others are concerned about the amount of screen time students are getting at school. California charter Rocketship Education has drawn criticism for its daily practice of placing 100 elementary students in a room with 100 laptops. One adult oversees the “learning process” by enforcing hours of silence as the students work through educational content and learning games.
Critics are also asking who profits.
Screen time giants such as Google, Facebook and Microsoft are walking the line between philanthropy and entrepreneurship, leaving many wondering if education reform should be in the hands of corporations. The political action committee working to save charter schools in Washington receives most of its funding from Connie Snyder Ballmer, the wife of Steve Ballmer; Reed Hastings, a cofounder of Netflix; and Vulcan Inc. (owned by Paul Allen).
While many companies right now are donating their time to develop these tools, and not profiting by it, the fact that millions of children are relying on technology to learn certainly doesn’t hurt Silicon Valley. There is big money to be made in educational analytics: The education app market is expected to grow 35 percent in the next three years, and the education gamification market is expected to grow 65 percent in the next three years. Big data is vital to both of those markets.
Equity issues: Who’s being helped?
Video cameras, online platforms and data sets are not exclusive to charter and private schools. Like the No Child Left Behind Act, December 2015’s Every Student Succeeds Act (ESSA) dictates that every district in the nation report a dizzying number of statistics on everything from suspensions to tardiness to grades on biology exams. These numbers are broken down by gender, race, special education/language status and family income level. (Data for every Washington state school can be found on the Office of the Superintendent of Public Instruction website, under the Research and Reports tab.)
Education data is often helpful. Test scores can illuminate gaps that exist due to socioeconomic differences, so districts can try to address them. Aggregate data is used to compare programs, set goals and measure progress. Susan Canaga, the program manager for data governance at the state’s Office of Superintendent of Public Instruction (OSPI), spends her days ensuring that the state is using quality data to help people make good decisions, often analyzing numbers through an equity lens. When looking at graduation rates, Canaga says, “We can start connecting and learning from districts that are having success. We have identified some districts that were positive outliers and interviewed them to find out what they are doing.” Districts with the highest graduation rates are currently pairing up with and mentoring other districts.
However, there are pitfalls. The need for assessment data requires students to complete a yearly battery of tests, some necessary for graduation. The negative effects of standardized testing include skyrocketing stress levels; the elimination of social studies and the arts as teachers focus more on tested subjects; teachers losing autonomy in their classrooms; and schools being judged on their test scores. This last factor is particularly problematic because test scores correlate strongly with socioeconomic factors. When government offices and online resources such as GreatSchools rank schools based on test data, they are essentially ranking schools on the wealth of their student populations. Concerned over the emphasis on standardized tests, many families have opted out, and some teachers have criticized the current testing landscape as well.
In the quest to manipulate data, schools often focus resources on groups of students who are likely to improve overall test scores. Teachers spend hours identifying kids who didn’t meet a benchmark last year, but were close. Those will be the students who will get more teacher attention and resources, at the expense of the highest or lowest achievers.
Constant surveillance seems like a high price to pay for the hope of better educational outcomes. In a world of WikiLeaks, cyberattacks and security breaches at major financial institutions, this wariness is not unfounded.
In addition to increased screen time, online school failures, tension over standardized tests and questions about ethics, privacy is a concern. In 2014, Google rolled out a free Apps for Education platform, and thousands of public teachers signed their students up. In 2015, the Electronic Frontier Foundation (EFF), a nonprofit with the goal of defending civil liberties in the digital world, filed a complaint with the Federal Trade Commission claiming that Google was “collecting, maintaining, using and sharing student personal information.” When students are logged into their Google Apps for Education (GAFE) accounts, data about other, non-educational sites those students visit is collected and used to enhance their behavior profiles. In a 2015 blog post, Google countered that “schools can control whether students . . . can use additional Google consumer services — like YouTube with their GAFE accounts.”
Many districts do block social media and video-sharing sites, and give students a school Google account to eliminate any connection to personal accounts. Yet the EFF remains concerned that Google released its free Apps for Education platform for reasons that are less than philanthropic.
Along with 200 other companies, Google signed the Student Privacy Pledge in 2014, which promised to “not collect, maintain, use or share student personal information” and “not to build a personal profile of a student” Jose Ferreira, CEO of Knewton, also signed the pledge, but prompted many raised eyebrows during a 2012 Education Datapalooza talk when he called education “the world’s most data-mineable industry,” claiming that his company gets 5–10 million actionable data points per student per day. Ferreira promised that data was not for sale, but his boastful proclamation that “we literally know everything about what you know and how you learn best” continues to make privacy advocates nervous.
Google and Knewton are still in the education business, but InBloom, the data analysis corporation used by New York state public schools and which was backed by $100 million in seed money from the Bill and Melinda Gates Foundation, had to shut down because of privacy concerns. The closure wasn’t due to a scandal or a break in the data shield, but a simple issue of mistrust. Parents are wary of corporations mining their children for data.
Constant surveillance seems like a high price to pay for the hope of better educational outcomes. In a world of WikiLeaks, cyberattacks and security breaches at major financial institutions, this wariness is not unfounded. The federal Education Department has been accused of doing a poor job of safeguarding data. Much is still unknown about privacy and security, and perhaps the inner workings of a third-grader’s mind is something that shouldn’t be transparent until the benefits are clear.
While data has the great potential to personalize instruction, students may still be more likely to achieve inside a traditional classroom rather than on a data-enabled online platform. Malia Burns, the executive director at Summit Sierra, points out that a school’s most critical component is not data or technology, but its teachers.
“Technology takes the teacher’s role and makes it more impactful,” she says, “but the teacher is still the one making all the decisions about the personalization.” This is a lesson that education reformers have been quick to learn. Data does not personalize education, teachers do.
But data helps. And even though the playing field is mucked up by questions about ethics, profiteering, privacy and testing, the promise of big data is too big to ignore. If education reformers and app developers can address concerns, personalized learning will put adaptive assessment tools at teachers’ fingertips, provide students with individualized online tutors, and help schools keep every student engaged and accountable for their own education.Google+