Top Ed-Tech Trends of 2016

Education Technology and Inequality

This is part ten of my annual review of the year in ed-tech

The richest 1% now possess as much wealth as the rest of the world combined.

That was the conclusion of an Oxfam report issued in January. It was a slogan of the Occupy Movement too, of course, one reprised this year in the Presidential campaign of Senator Bernie Sanders, who would frequently repeat that “Now is the time to create a government which represents all Americans and not just the 1%.”

It doesn’t look as though we’ve done that, sadly. “Trump’s 17 cabinet-level picks have more money than a third of American households combined,” according to Quartz.

Income inequality continues to grow – both within nations and globally – and it poses a grave risk for democracy and for the environment.

The American Dream, a phrase invented during the Great Depression, feels more and more out of reach for more and more Americans, as increasingly people are making less money than their parents did.

One of the mantras of that dream – the idea that economic success is possible if not inevitable – involves the necessity of education. “Education can be the difference, that education can save lives, that education can put folks on a path to opportunity,” Secretary of Education John B. King Jr. told the students at Milwaukee Area Technical College’s graduation ceremony in May. But it’s a “false promise,” Jacobin’s David I. Backer contends. As economic inequality has grown, so has schooling: “United States citizens are more educated than they ever have been. More people have graduated from more kinds of schools than at any point in history.”

Indeed, rather than a “silver bullet,” education often serves to reinforce inequalities. Sixty-two years after Brown v Board of Education, segregation is worsening – in neighborhood schools, at elite schools, at charters. This comes as the majority of students in the US public school system are now students of color. (The majority of teachers are still white.)

Data released in June by the Department of Education’s Office for Civil Rights highlighting the ongoing disparities – between the experiences of white students and students of color, between the experiences of affluent students and low-income students, and between the experiences of students with disabilities and those without – serves to underscore the systematic failure to provide equitable education at the preschool and K–12 levels in the US.

Black preschool children, for example, are 3.6 times more likely to be suspended than white preschool children. Black K–12 students are 3.8 times more likely to be suspended than white students. Students with disabilities are more than twice as likely to be suspended than students without disabilities. Black students are 1.9 times more likely to be expelled than white students. Black students are 2.2 times more likely to be referred to law enforcement than white students. Charter schools, according to a study based on this OCR data, have an even higher rate of suspending Black students and students with disabilities. And some charters have been charged with purposefully refusing to enroll certain students – a violation of the law.

Black, Latino, and Native students are less likely to have access to high-level math and science courses. They are underrepresented in gifted and talented programs. They are underrepresented in AP courses. Black, Latino, and Native students are more likely to attend schools with high concentrations of inexperienced teachers. They’re more likely to attend schools where teachers have not met all state certification requirements. 87% of white students graduate on time; 76% of Latino students and 73% of Black students do. Native American students have the worst graduation rates in the country, particularly those attending schools run by the Bureau of Indian Affairs. And while white, Black, and Latino students enroll in college after graduation at roughly the same rate, students of color are much less likely to graduate with a Bachelor’s degree in six years or less. “That disparity hints at the large enduring difference in the quality of the K–12 preparation many minority students are receiving,” writes Ronald Brownstein in The Atlantic.

The inequalities of K–12 education extend into higher ed, exacerbated by high tuition, inadequate financial aid, and admissions policies that privilege white and affluent students. (Take, for example, ProPublica’s article on Donald Trump’s son-in-law and “consigliere”: “The Story Behind Jared Kushner’s Curious Acceptance into Harvard.”)

This fall, Georgetown University announced its plans to “atone for its slave past” – like many universities, the Jesuit-run institution had a long connection to the slave trade, selling 272 men, women, and children in 1838 to pay off its debts. The university said it would begin offering preferential admission status – like the children of alumna already receive – to the descendants of the slaves owned by the university. It was a gesture that The Atlantic’s Adrienne Green said “falls short,” and it certainly does not count as reparations according to sociologist Tressie McMillan Cottom, which she argues must contain three components: “acknowledgement, restitution, and closure.”

The idea that preferred admission equals payment stems from the American ideology that opportunity, especially educational opportunity, is a “fair” form of recompense. Opportunity has a moral basis: It will only be valuable for those who deserve it and will not inconvenience or harm those who already have the opportunity (whether they deserve it or not). Our society likes opportunity because it does not demand redistribution of resources acquired through harm. As you can tell, I’m not a fan of this logic. But even if I were, preferred admission doesn’t equate to much of an opportunity.

Preferred admissions gives a narrowly defined group of black descendants a chance to compete for achievements that are defined by accumulated disadvantage. The chance to be preferred in admissions to Georgetown still relies on racial differences in college preparation, racial wealth, and income gaps that condition the ability to pay college tuition, and racial gaps in knowledge about competitive college admissions. Preferential admissions says if you somehow manage to navigate all those other legacies of slavery – wealth disparities, income disparities, information disparities – then we will give you additional consideration in admissions. That is generous when judged by how little other universities have done but it is not much of an opportunity and it isn’t a form of payment at all.

College campuses have become much more diverse over the past few decades, true, but these institutions remain insensitive, unwelcoming, and hostile to students and faculty of color, to students and faculty with disabilities, to queer students and faculty, and to women.

The Office for Civil Rights said it received a record number of complaints this year – a 61% increase from last year. The number of reports regarding sexual assault on college campuses increased 831%; complaints regarding web accessibility for persons with disabilities was up 511%; complaints involving the restraint or seclusion of students with disabilities increased 100%; and complaints involving harassment on the basis of race or national origin increased by 17%. Teaching Tolerance released a report on the increased harassment and bullying witnessed this year – something it tied directly to the Trump campaign: 90% of K–12 educators that the organization surveyed said that their school climate had been adversely affected by the racist, nationalist, sexist rhetoric of the Presidential campaign. Reports of hate crimes and racist graffiti spiked on school campuses across the country following Trump’s election.

Will President Trump make all these educational inequalities worse? Certainly there are serious concerns about his choice for Secretary of Education, Betsy DeVos, and the overwhelmingly negative impact that her political influence has had on Michigan schools, particularly for students in low-income urban schools. There are also fears that the Trump administration will be less likely to enforce Title IX and might scrap the Office for Civil Rights altogether. Furthermore, his promise to deport undocumented immigrants has schools scrambling to plan for how they will protect their students.

I also want to consider that, with or without a President Trump, education technology also might make things worse, might also contribute to these ongoing inequalities – and not simply because many in ed-tech seem quite eager to work with the new administration.

Don’t Believe “Don’t Be Evil”

My own concerns about the direction of education technology cannot be separated from my concerns with digital technologies more broadly. I’ve written repeatedly about the ideologies of Silicon Valley: neoliberalism, libertarianism, imperialism, late stage capitalism. These ideologies permeate education technology too, as often the same investors and same entrepreneurs and the same engineers are involved.

As I wrote in my article on “the ‘new’ economy,” automation, so we’re told, is poised to reshape “work.” It has reshaped work. None of this has played out equitably, as the benefits have accrued in management and not by labor. Indeed, the World Bank issued a report in January arguing that digital technologies – not just robots in factories – stand to widen inequalities as well, “and even hasten the hollowing out of middle-class employment.” While new technologies are spreading rapidly, the “digital dividends – growth, jobs and services – have lagged behind.” As venture capitalist Om Malik wrote on the eve of 2016, “In Silicon Valley Now, It’s Almost Always Winner Takes All.” Money and data – they’re intertwined for technology companies – are monopolized in a handful of corporate giants.

The technology industry – its products and its politics – furthers inequality, particularly in its own backyard in the Bay Area. Its high profile executives then have the audacity to claim that reality – human suffering – is merely a simulation. Or they say they’re prepared to leave Earth and colonize Mars. Or they back Donald Trump.

Trump, for his part, indicated on the campaign trail he might be interested in creating a registry to track Muslims’ whereabouts in the country; and while some technology companies and tech workers have sworn they would never participate in building a database to do this, no doubt, the metadata to identify us and track us – by our religion, by our sexual identity, by our race, by our political preferences –already exists in these companies’ and in the government’s hands. Trump will soon have vast surveillance powers – thanks in part to technology companies like Palantir, thanks in part to expanded NSA surveillance, authorized by President Obama – under his control.

Meanwhile, schools and education companies have also expanded their surveillance of students and faculty, with little concern, it seems, to how politically regressive all this data-mining and algorithmic decision-making might actually be.

Inequality and the “Top Ed-Tech Trends”

The inequalities that I’ve chronicled above – income inequality, wealth inequality, information inequality – have been part of our education system for generations, and these are now being hard-coded into our education technologies. This is apparent in every topic in every article I’ve written in this years’ year-end series: for-profit higher education, surveillance in the classroom, and so on.

These inequalities are apparent in the longstanding biases that are found in standardized testing, for example, often proxies for “are you rich?” and “are you white?” and “are you male?” Despite all the Common Core-aligned revisions and all the headlines to the contrary, “The New SAT Won’t Close the Achievement Gap.” (Shocking, I know.) In fact, according to Reuters, the College Board has redesigned the SAT in ways “that may hurt neediest students.”

Ed-tech’s inequalities are evident too in the results, in many cases, of moving standardized testing from pencil-and-paper to computer. Scores for some students who took their PARCC exams on computers were lower – lower in Rhode Island and lower in Maryland, for example.

There were also significant gaps on a new NAEP exam administered this year, one measuring “technology and engineering literacy”: “Students whose families are so poor that they qualify for free or reduced-price lunch scored 28 points lower, on average, than students from more affluent families. The gap between black and white students was even more pronounced, with 56 percent of white students scoring at or above ‘proficient’ and just 18 percent of black students meeting that bar,” Chalkbeat reported in May. (Girls, for what it’s worth, out-performed boys.) Another study conducted by the Department of Education found that using computers widens the “achievement gap” between high-performing and low-performing students. The latter group, which is more likely to be comprised of Black, Latino, and low-income students, performed better on writing assessments when writing with pencil and paper.

This “gap” seems to extend to online courses too. A study from Northwestern University, for example, found that “high-achieving North Carolina 8th graders who took Algebra 1 online performed worse than similar students who took the course in a traditional classroom.” A study from the American Institutes of Research found that “students working online were 10 percentage points less likely to pass than the students randomly assigned to take the course face-to-face – 66 percent compared with 76 percent.” A report issued by the National Education Policy Center confirmed what we’ve known for some time now – that students at virtual schools fare very poorly – but added that students at blended schools (those that combine face-to-face and online instruction) are struggling as well, with 77% of the blended schools the NEPC reviewed performing below state averages. And this problem exists at the college level too. Research from California’s Public Policy Institute found that students in the state’s community college system are 10 to 14% less likely to pass a class when they take it online. – but there’s an “online paradox,” according to The Chronicle of Higher Education, because students who successfully complete at least one online course are 25% more likely to graduate than those who only take classes face-to-face.

Despite the serious flaws in online and blended learning, many education technology advocates continue to push for more and more education technology, and Silicon Valley investors in turn continue to fund the expansion of use of these products, particularly in low-income schools, in the US as well as in the developing world.

As I wrote in the first article in this series, one of the latter companies, Bridge International Academies, was poised to take over Liberia’s public school system. Bridge International – funded by the Chan Zuckerberg Initiative, the Gates Foundation, the Omidyar Network, and others – is a private school startup that hires teachers to read scripted lessons from a tablet that in turn tracks students’ assessments and attendance – as well as teachers’ own attendance. Expansion of Bridge International Academies has been controversial, and the Ugandan government ordered all BIA schools there to close their doors. Other companies with similar models: Spark Schools, which raised $9 million this year from the Omidyar Network and Pearson, and APEC, also funded by Pearson. In April, journalist Anya Kamenetz looked closely at “Pearson’s Quest to Cover the Planet in Company-Run Schools”: "Pearson would like to become education’s first major conglomerate, serving as the largest private provider of standardized tests, software, materials, and now the schools themselves.

Whether it’s selling schools or MOOCs or access to the Internet itself, technology companies and education companies are, as Edsurge put it, “Building Effective Edtech Business Models to Reach the Global Poor.” Whether or not the education itself is “effective,” let alone equitable, is another question altogether.

Data about who’s funding the expansion of private schools in the developing world can be found on funding.hackeducation.com.

From the “Digital Divide” to “Digital Redlining”

Discussions about education technology (and new digital technologies more generally) were, for many years, framed in terms of the “digital divide” – that is, the gap between those who have access to computers and to the Internet and those who do not. It’s a gap resulting from a variety of factors, including socioeconomic status, race, age, and geographic location.

Community college professors Chris Gilliard and Hugh Culik contend that there’s a “growing sense that digital justice isn’t only about who has access but also about what kind of access they have, how it’s regulated, and how good it is.”

We need to understand how the shape of information access controls the intellectual (and, ultimately, financial) opportunities of some college students. If we emphasize the consequences of differential access, we see one facet of the digital divide; if we ask about how these consequences are produced, we are asking about digital redlining. The comfortable elision in “edtech” is dangerous; it needs to be undone by emphasizing the contexts, origins, aims, and ideologies of technologies.

Sociologist Tressie McMiillan Cottom, briefly banned from Facebook for not using her real name on the site, argues that,

This kind of stratified access to information and participation in digitally-mediated social interactions isn’t just about who can post cat memes and who is denied.

As Facebook itself had to admit this week, its platform has become a central means for distributing access to favorable information about jobs, housing, banking, and financial resources.

Being othered on Facebook increasingly means being relegated to unfavorable information schemes that shape the quality of your life.

How do digital redlining and these “unfavorable information schemes” permeate education technology – in its implementation and in its very design?

Discrimination by Design

Discriminatory practices can be “hard-coded” into education technologies through the data they collect and how they label and model that data. Information systems that offer only two choices for sex or gender, for example, fail to accommodate transgender students – and violate Title IX, according to the Department of Education. This year, the Department of Education also encouraged schools to stop asking applicants about their criminal histories, and while some researchers have sought the collection of data about students’ sexual orientation – ostensibly to identify discrimination – there are concerns about how this information might easily be used against LGBTQ students.

Harassment is pervasive online, but harassment and cyberstalking are not experienced equally by everyone. A report by Data & Society issued this fall found that 47% of American Internet users say they’ve personally experienced online harassment or abuse. 72% say they’ve witnessed online harassment or abuse. “Internet users ages 15–29, Black internet users, and those who identify as lesbian, gay, or bisexual are all more likely to witness online harassment,” and LGB Internet users are more than twice as likely to experience harassment online than their straight peers. Black and LBG Internet users were more likely to say that people online are “mostly unkind.”

“Mostly unkind” – and yet education technology (and digital technologies more generally) demands students and faculty be online.

Discriminatory practices online are certainly a reflection of discriminatory practices offline, but it’s important to recognize how these become part of the technological infrastructure, part of the code, in ways that are both subtle and overt. Harassment in virtual reality. Harassment using annotation tools.

These new technologies are designed (predominantly) by white, able-bodied, English-speaking heterosexual men from the global north – designed by men for men.

“Just use your initials online instead of your name,” was one venture capitalist’s advice to women this year.

I don’t want to overlook two of those descriptors above: English-speaking and able-bodied. 53% of the World Wide Web is in English. The majority of programming languages are in English. (English-language learning software has long had a large market globally, and venture capitalists seem keen to fund companies that offer these products to K–12 schools as the number of ELL students grows.) What sorts of biases are built into digital technologies because of this?

What sorts of discriminatory practices are we reinstating and reinforcing online?

Despite the requirements of the Americans with Disabilities Act, much education technology remains in accessible. This includes software, digital content, and websites. There were several lawsuits this year demanding schools and their technology vendors comply with the law.

UC Berkeley, on the other hand, announced in September that “may eliminate free online content rather than comply with a U.S. Justice Department order that it make the content accessible to those with disabilities.” The material involved MOOCs that it had produced with edX as well as videos posted to iTunes and YouTube. MOOCs. “Free and open.” “In many cases,” the university said, “the requirements proposed by the department would require the university to implement extremely expensive measures to continue to make these resources available to the public for free.”

So instead, it opted to pull them offline altogether.

Predictive Analytics and Algorithmic Discrimination

In January, the student newspaper at Mount St. Mary’s University in Maryland reported that the school’s president had a plan to push out students at risk of dropping out in the first few weeks of class. Doing so early in the semester would mean these students would not count against the university’s retention rate. The paper recounted a conversation the president reportedly had with faculty, encouraging them to rethink their approach to struggling students: “This is hard for you because you think of the students as cuddly bunnies, but you can’t. You just have to drown the bunnies ... put a Glock to their heads.” President Simon Newman, a former private equity CEO, said it was “immoral” to keep struggling students enrolled.

Education technology companies now promise that they can help schools identify these struggling students, through an algorithmic assessment of who’s at risk. These systems weigh a variety of data: standardized test scores, grades, attendance, gender, marital status, age, military service, learning management system log-ins, and “digital footprint.” “Digital footprint” – that is, all manner of students’ online behaviors might be tracked by this software, purportedly “for their own good.”

Predictive analytics like this are supposed help to guide schools so they can offer support services – ideally, better and more responsive services – to struggling students, keeping them enrolled and on a path to graduation. Or, no doubt, predictive analytics can help identify those “drowning bunnies” that must be eliminated.

In a report released this fall titled “The Promise and Peril of Predictive Analytics in Higher Education,” New America’s Manuela Ekowo and Iris Palmer cautioned that,

Predictive models can discriminate against historically underserved groups because demographic data, such as age, race, gender, and socioeconomic status are often central to their analyses. Predictive tools can also produce discriminatory results because they include demographic data that can mirror past discrimination included in historical data. For example, it is possible that the algorithms used in enrollment management always favor recruiting wealthier students over their less affluent peers simply because those are the students the college has always enrolled?

Discrimination, labeling, and stigma can manifest in different ways depending on how colleges use these algorithms. For instance, colleges that use predictive analytics in the enrollment management process run a serious risk of disfavoring low-income and minority students, no matter how qualified these individuals are for enrollment. Predictive models that rely on demographic data like race, class, and gender or do not take into account disparate outcomes based on demographics may entrench disparities in college access among these groups.

Furthermore, predictive analytics, recommendation engines, and other analytics software might keep some students enrolled – and that’s a boon to schools’ bottom lines – but it might also steer them into courses that are less intellectually challenging.

Data about who’s funding predictive analytics in education can be found on funding.hackeducation.com.

“We will literally predict their life outcomes,” claims one scientist. Another group of researchers says they can predict “which children will grow up to be drain on society – when they are just three years old.” Others say they’re working on the nascent field of “educational genomics,” to predict students strengths and weakness and, of course, “personalize” their education.

While much of this sounds like (dystopian) futuristic science fiction, predictive analytics are currently being used to identify students who might be suicidal and those who might develop “extremist” political beliefs or are at risk for “radicalization.”

Law enforcement increasingly uses predictive analytics to identify future criminals, and courts are using predictive analytics to determine sentencing. These have a demonstrable bias against African-Americans, and some of these systems admitting they use facial features to identify criminality. Phrenology 2.0. Schools work with these companies, handing over student data in the process. This relationship between schools and law enforcement cannot be understated, and a study released this fall found that “campuses with larger populations of students of color are more likely to use harsh surveillance techniques.”

In China, credit scores will be determined using people’s Web browsing history, and again, we shouldn’t just dismiss this as something from another time or another place. In the US, loan companies are already starting to use analytics to determine loan eligibility, and there’s talk of expanding the type of data that’s used to determine student loan eligibility as well.

Predictive analytics are being utilized in hiring decisions, testing job candidates for “culture fit.” (Code for “white guy.”) MIT professors Erik Brynjolfsson and John Silbert have called for “moneyball for professors,” using analytics to determine tenure. One education technology startup claims it’s devised a proprietary screening tool that can “accurately predict whether a prospective hire will be an effective teacher, and more specifically whether they will be able to boost students’ test scores.”

There’s no research to back up these claims. And when the software is proprietary, there’s little chance one can examine the algorithms in play.

That’s a problem with all these algorithms – we can’t see them, we can’t evaluate them, and we can’t verify their “accuracy.” In April, high school students in France demanded to know what powers the algorithm that’s used to dictate their post-baccalaureate education options. Everyone should know how these sorts of decisions are being made for them. ProPublica, for its part, has published a series of stories this year “breaking the black box” and investigating algorithmic decision-making, noting how these often function as “discrimination by design.”

There remains very little insight and very little accountability in these algorithms, particularly in education. And, based on what we know about institutional and corporate biases, there is every reason to believe that these algorithms are exacerbating educational inequalities.

Education Technology and Digital Polarization

We trust algorithms to make more and more decisions for us, often quite uncritically. Whose values and interests are actually reflected in these algorithms?

Algorithms dictate much of what we see (and what we don’t see and who sees what) online, the news and media we consume – whether on Facebook, or Google, or Amazon, or Twitter, or Netflix. How algorithms shape new information technologies will have profound effects on education, on knowledge – and on democracy.

We saw hints of this, no doubt, in this year’s US Presidential election, although the malaise is much deeper and broader than one electoral event. “Fake news.” Red feeds versus blue feeds. “Post-truth.” Information warfare. The fragmentation of knowledge. Distraction. Expertise, trumped. Digital polarization. It’s “personalization,” we’re told, and we’re supposed to like it.

I’ll close here with words from Maciej Cegłowski, who runs the bookmarking site Pinboard, speaking at the SASE conference in June on “The Moral Economy of Tech”:

The first step towards a better tech economy is humility and recognition of limits. It’s time to hold technology politically accountable for its promises. I am very suspicious of attempts to change the world that can’t first work on a local scale. If after decades we can’t improve quality of life in places where the tech élite actually lives, why would we possibly make life better anywhere else?

We should not listen to people who promise to make Mars safe for human habitation, until we have seen them make Oakland safe for human habitation. We should be skeptical of promises to revolutionize transportation from people who can’t fix BART, or have never taken BART. And if Google offers to make us immortal, we should check first to make sure we’ll have someplace to live.

Techies will complain that trivial problems of life in the Bay Area are hard because they involve politics. But they should involve politics. Politics is the thing we do to keep ourselves from murdering each other. In a world where everyone uses computers and software, we need to exercise democratic control over that software.

I recognize that many people are committed to the belief that the adoption of education technology means “progress.” But it isn’t necessarily politically progressive. At all. We must understand how education technology, in its current manifestation, might actually to reinforce education’s longstanding inequalities.

We must consider too, as we move into a new year with a new President, that it might also be – algorithmically, financially, culturally – profoundly anti-democratic.

This post first appeared on Hack Education on December 21, 2016. Financial data on the major corporations and investors involved in this and all the trends I cover in this series can be found on funding.hackeducation.com. Icon credits: The Noun Project