FOR MANY YEARS, educators, designers, and policymakers have hoped that free and low-cost online technologies could bridge the chasm of opportunity that separates more and less affluent students. This dream has proven elusive.1
When I was a graduate student, I visited a classroom in rural New Hampshire where an ambitious young teacher was planning to have her students build collaborative presentations on wikis—websites that allow for multiple authors to publish together online. The technology in the lesson required a linked chain of resources: internet networks came into the building through cables to a wireless router broadcasting in her room; the students had laptops that had been successfully charged the night before and were receiving the wireless signal; the teacher had a computer plugged into a projector; the screen on the wall came down to receive the image (teachers reading this will shudder to remember those flimsy screens that wouldn’t stay down); and the teacher took the power cord from the projector and went to plug it into the wall to show students what to do from her computer. As she pushed the plug into the outlet, the outlet rocked out of position and fell back behind the drywall, unrecoverable. After a few moments of fishing behind the insulation, the teacher gave up on using the projector and shifted to plan B, rather miraculously making the lesson work without a demonstration.
Since that visit, I’ve often thought of the delicacy of that chain of resources—power, broadband, wireless signal, equipment, bulbs, chargers—that allows people to learn from technology in their classrooms, schools, dorms, and homes. Affluent communities—neighborhoods, families, schools, and institutions—have more resources devoted to building and maintaining this delicate chain to take advantage of free, online learning tools and apps. Sociologists call this kind of phenomenon a Matthew effect, named for a verse in the biblical Gospel of Matthew: “For whoever has will be given more, and they will have an abundance. Whoever does not have, even what they have will be taken away from them.”2 The edtech Matthew effect posits that this pattern is quite common in the field of education technology and learning at scale: new resources—even free, online resources—are more likely to benefit already affluent learners with access to networked technology and access to networks of people who know how to take advantage of free online resources.
As we shall see, despite robust evidence for the prevalence of the edtech Matthew effect, edtech funders, developers, and enthusiasts continue to be animated by three linked myths about technology’s potential to democratize learning. Working toward a more equitable future for education technology requires rejecting these myths and confronting the realities that could guide us toward more productive development, policy, and practice.
The first myth is that technology disrupts systems of inequality. For all the hope and hype that technology might enable major organizational changes in educational systems, the reality is that technology reproduces inequalities embedded in systems. New apps, software, and devices are put in the service of existing structures and systems, rather than rearranging them.
The second myth is that free and open technologies will democratize education. The rich data that can be collected from new digital platforms allow closer investigation than ever before about how learners from different life circumstances access and use new learning technologies. The research from these investigations makes the reality clear: new technologies, even free ones, disproportionately benefit already-advantaged students. In a sense, the digital divide is more of a digital fault line, and each new innovation opens chasms of opportunity between our most and least affluent students.
The third myth is that digital divides can be closed by expanding access to technology. Helping learners access functioning, modern computers with reliable broadband connections is only one step toward digital parity. Social and cultural forms of exclusion are as powerful, and often much harder to understand and address, than challenges of technology access. Turning the potential of education technology toward the benefit of the students who are furthest from opportunity will require reckoning with the social and cultural contexts in which marginalized students live and the very different contexts in which most new technology applications and services are developed.
The accumulated evidence disproves these myths and makes clear that education technology will never simplistically close digital divides. No matter how many transistors we squeeze into a square millimeter, no matter how many bits are passing wirelessly over our heads, the hard parts of reducing educational inequality will remain hard.
From the earliest days of signals technologies, inventors and evangelists have promised that new technologies would provide more equitable learning experiences for young people. In his 1984 book Teachers and Machines, Larry Cuban shows an image from the 1930s of children huddled around a giant radio transceiver with the caption, “With Radio, the Underprivileged School Becomes the Privileged One.” Cuban chronicles developments in film, radio, television, and early efforts to place personal computers in classrooms, and he observes how each generation of technology advocates promised a radical reconfiguration of teaching and learning. Cuban found a handful of bold, interesting experiments, but for the most part, they primarily reproduce the patterns, behaviors, and inequalities that already exist within schools.3
One of the most unfortunate and inequitable practices in schools is the separation of students into learning experiences of very different quality. In her classic study Keeping Track, Jeannie Oakes documents how students in honors and college preparatory courses engage with rich content, solve complex problems, and communicate their understanding in diverse ways. In the same schools, students in basic and remedial courses encounter a simplified curriculum, solve less interesting problems, and find fewer opportunities for creative and intellectual expression. Assignment into these tracks correlates with race and class; affluent majority students are most likely to find themselves in the most challenging and meaningful learning environments, while students from low-income families or racial minorities are more likely to be placed in classrooms that limit their growth and potential.4
For the past thirty years, education technology researchers have collected evidence showing that similar patterns of diverging educational quality emerge in the implementation of digital learning technologies. At the turn of the twenty-first century, sociologist Paul Attewell proposed that educators think of these patterns as two digital divides. The first digital divide is the divide of access: students from low-income or marginalized backgrounds typically have less access to new technologies than more affluent students. But even more important is the second digital divide of usage: students from low-income or marginalized backgrounds are more likely to use technology for routine drill and practice with limited adult support, while more affluent students use technology for more creative activities with greater mentorship from teachers, parents, and adults.5
In the 1990s, Harold Wenglinsky at the Educational Testing Service analyzed test scores and survey data from the 1996 National Assessment of Educational Progress (NAEP). Students were asked about patterns of technology usage in their math classrooms, and Wenglinsky found that low-income, non-white children used technology primarily in math class for drill and practice, while affluent white children were more likely to use technology for graphing, problem solving, and other higher-order exercises. Wenglinsky argued that “poor, urban, and rural students were less likely to be exposed to higher-order uses than non-poor and suburban students.”6 The same survey questions were used again in the 2009 and 2011 NAEP tests with basically identical results: the most promising uses of educational technology were mostly available to already-advantaged students. Qualitative researchers have found similar patterns through close observations of schools and families. For his dissertation, Matthew Rafalow conducted an ethnographic study of three high schools in southern California that all had the same levels of technology access but served different populations of students. One of his study sites was a majority white school in an affluent neighborhood, where teachers and educators described creative and playful uses of technology—even playing games like Minecraft—as a valuable part of student development. By contrast, in the two schools Rafalow visited serving middle- and lower-income students, educators described these uses of technology as irrelevant or disruptive, and their technology use focused on more basic skills. In affluent schools, kids who played around with computers were hackers, and teachers saw their play as preparing for careers in technology; in schools serving low-income families, kids who played around with computers were treated as slackers.7
For the most part, new technologies don’t rearrange practices in schools. They reinforce them.
In recent years, technology advocates have used the phrase “democratizing education” to describe the potential for new technologies to reduce inequality. Democratize is a slippery word, and it’s usually used to describe making something more fair, more equitable, or more just, without describing exactly how these goals might be achieved. One theory of change is that because affluent learners and families already can afford high-quality learning experiences, if new technologies are made free or more easily accessible, then less affluent learners will get access to what the affluent already have.
One of the virtues of new large-scale learning technologies is that they generate data sources that can be used to delve deeper into how technologies are used differently by students from different backgrounds. In the last ten years, I’ve worked on two major studies about the second digital divide of usage. In the late 2000s, I studied how teachers and students used social media and peer production tools such as blogs and wikis. At the time, there was a real optimism that these tools would be as transformative in education as they had been in journalism, business, social relationships, and information management. Wikis and blogs could create new opportunities for students to collaborate with peers across the classroom or around the world, and they could provide new means for students’ digital expression. Before the arrival of these web-based collaboration tools, collaborative computing projects would require a major investment in computers, networks, and specialized software (like Hypercard). The hope was that with computers more accessible and the software free or low cost, these opportunities for rich digital learning would be much more widely available, especially for low-income students.8
Starting in 2009, I looked at data from hundreds of thousands of wikis used in K–12 education settings. Each wiki recorded every change made by every user, and in many wikis, we could examine the content created on the site, identify the school where the wiki was created, and use data from the National Center for Education Statistics to learn more about the demographics of the schools creating the wikis. We found that wikis were more likely to be created in schools serving affluent neighborhoods, and the wikis created in those affluent schools were used for longer periods of time with greater opportunities for student involvement. Wikis created in schools serving low-income families were more likely to be used for teacher-centered content delivery, and they fell into disuse more quickly.9
To try to understand these patterns, I started visiting wiki-using teachers around the country, which is how I found myself in the New Hampshire classroom with the dodgy power outlets. I saw a phenomenon similar to what Attewell observed in the 1990s: in more affluent schools, more resources were devoted to the maintenance of technology networks, which could then be used more easily and more reliably. Teachers had more access to planning time and professional development, and because they had fewer pressures from standardized testing, they could take more pedagogical risks and try new things like wikis. Teachers in affluent neighborhoods could count on more access to technology resources in homes, which made them more comfortable assigning complex projects that might require online homework. While the wiki cloud software was free for teachers to use, the maintenance of technological systems and expertise was expensive.
After the passing surge of interest in integrating social media and peer production in education, MOOCs became the next participant in the edtech hype cycle, with advocates again declaring that free online courses would democratize education. As I discussed in Chapter 1, when my colleague John Hansen and I connected MOOC registration and participation data from edX with demographic data from the US census, we found that people living in more affluent neighborhoods were more likely to sign up for edX MOOCs. Moreover, markers of socioeconomic status—like having a parent with a college degree or living in a more affluent neighborhood—were positively correlated with course completion. MOOCs opened a door of opportunity, and although there are remarkable stories of learners from very difficult circumstances who took advantages of these new opportunities, the majority of people who walk through that door of opportunity are already educated and already affluent.10
Learners in affluent neighborhoods are better able to take advantage of new education technology, even when those technologies are free. The research record becomes increasingly clear with each passing year: we should expect that most education technology initiatives—including those made available for free online—will disproportionately benefit the affluent.
While financial and technical barriers are real and important, social and cultural exclusions are often the thornier obstacles to educational equity. These social and cultural barriers are harder to discern, require more nuance to address, and vary substantially across different communities and contexts.
The “digital divide” is too simple a metaphor to characterize educational inequality. I prefer the image of a digital fault line; inequalities emerge and disappear as tectonic shifts from new technologies change the social landscape. As new technologies arrive and as older technologies become obsolete, the shape of the digital fault line constantly changes. In the 1980s and 1990s, the digital divide in education referred primarily to the differences in student-computer ratios between more and less affluent schools or to gaps in home desktop computer ownership. As more schools began to acquire computers, the most pronounced digital gaps have shifted toward broadband access rather than device ownership. For a short time, when mobile internet first became accessible, youth of color actually led adoption, since it was a better value for many families than home broadband subscriptions. As mobile phones have become increasingly sophisticated, however, gaps in mobile broadband adoption have emerged. In their study of the broadband subsidy program Connect2Compete, Vikki Katz and her colleagues found that policy decisions often showed a profound misunderstanding of the populations that they were trying to serve. The subsidy program assumed that families were completely disconnected and would welcome even a single ethernet connection. In reality, most families were already online and needed to reliably connect multiple devices via Wi-Fi in order to meet the needs of the entire family for school, work, community engagement, and leisure. The subsidy program was trying to close a gap in the digital fault line that had already mostly closed while ignoring the one that had opened right next to it.11
The easiest gaps to measure along the digital fault line are gaps of access—who has access to how many computers or mobile devices with what kinds of broadband speeds. But the more challenging obstacles are those that are social and cultural in nature. Katz and colleagues documented that in the Connect2Compete program, uptake across low-income communities differed by levels of community trust. In one case study where community trust in schools and municipalities was high, the program was more effective in helping families get access to new resources. In their case study site in Arizona, however, uptake was much less, because many immigrant families were concerned about using school-issued laptops out of fear of state surveillance. Matt Rafalow’s three-school case study, discussed above, reveals how even when technology is held constant across different schools, teachers can celebrate technology usage by privileged students while questioning it among other students.
Large-scale quantitative research also sheds some light on these challenges at a global scale. One robust finding from MOOC research is that a person’s country of residence can be a strong predictor of course completion. Using the Human Development Index (HDI), a measure of a country’s affluence, education, and general well-being developed by the United Nations Development Program, research shows that MOOC learners from countries with low HDI measures are significantly less likely to complete courses than learners from countries with high HDI scores, a kind of global achievement gap in MOOCs. Obstacles that online learners in developing countries face that contribute to this gap might include unstable internet and electricity, structural poverty, and many other challenges.12
My colleague René Kizilcec theorized that learners from less developed countries might face an obstacle called social identity threat. Social identity threat occurs when learners use cognitive resources attending to concerns about stereotypes or feelings of exclusion rather than to learning. We cannot say definitively what causes these feelings in online courses, but triggers could include the elite branding of universities offering MOOCs, the predominantly white American and European faculty who offer these courses, the English language usernames in the forums, and other markers that feel excluding to minority participants. Feelings of social identity threat can lead to negative recursive cycles: when people start a class, they may feel like an outsider. This feeling of exclusion might lead to their not doing as well in the first week as they might have, which makes them more attuned to things that would make them feel like outsiders, which makes them do less well in the course, and so on. In other words, Kizilcec hypothesized that some of the challenges that learners from developing countries might face would be psychological in nature, in addition to the structural and economic barriers that learners from developing countries might face in trying to participate in online courses.13
To test this hypothesis, Kizilcec, myself, and a team of colleagues from Harvard, MIT, and Stanford randomly included an intervention that addressed social identity threat in the pre-course surveys of more than 250 MOOCs. In this simple intervention, learners were asked to read a list of values and identify two or three that were most important to them. They then wrote a short note about why taking an online class aligned with those values. The intervention was designed to deepen the connection between a learner’s sense of purpose and his or her participation in a class, and in theory, it served as a kind of “inoculation” to the negative feedback cycle induced by social identity threat. The writing exercise addressed feelings of exclusion by having learners themselves write new narratives about inclusion. Learners in the control group received the standard survey without the intervention. We found that the intervention worked in courses where students from more affluent countries earned certificates at much higher rates than students from developing countries. In these courses, students from medium- and low-HDI countries who were randomly assigned to the intervention were more likely to complete courses than students in the control condition. In other words, where a “global achievement gap” existed, the intervention closed a portion of that global achievement gap. While MOOC students from less developed countries may still face challenges from infrastructure quality and access, our study suggests that at least some of the differences seen in educational attainment in MOOCs might result from feelings of exclusion.14
This idea that social and cultural exclusions could be as powerful as economic and technological obstacles can seem very counterintuitive. When Kizilcec and I first submitted an earlier paper about our interventions to Science, the paper was rejected without explanation and without having been sent out for review (a process called desk rejection, not uncommon for a prominent journal). One anonymous editor had written in his comments, “Not clear to me that this ‘fear that they could be judged in light of negative stereotypes about their nation’ is more of a hurdle than, say, irregular power supply, intermittent internet, gender, lack of tailoring of materials to local contexts / cultures, etc.” The central premise of our paper was that our experiment only affected learners’ psychological state; students from low-HDI countries in the treatment condition faced identical challenges with technology and access as students from low-HDI countries in the control condition. The experiences of students in the experiments differed only in terms of access to one short exercise designed to provide psychological support. Even in the face of experimental evidence that cultural exclusions matter, this one reviewer remained convinced of the importance of technical dimensions. The story does have a happy ending. After we reframed our writing to emphasize that the only factor we changed in learner experiences was a psychological one, the editors chose to accept the paper.15
One challenge in addressing cultural exclusions is the chasm of social status between edtech developers and advocates and the communities of learners that they are trying to serve. Sociologist Tressie McMillan Cottom argues that technologists often imagine their students as “roaming autodidacts,” which she describes as “a self-motivated, able learner that is simultaneously embedded in technocratic futures and disembedded from place, culture, history, and markets.”16 If this description fits any group aptly, it might be the developers of education technology themselves—primarily white and Asian men coming from a handful of top-tier universities, at ease whether in Cambridge, Massachusetts, or Palo Alto, California, and part of the global elite that uses technology and affluence to transplant their culture and comforts wherever they go. Most people, however, live in a particular place, grounded in cultures and families that shape their interactions with learning experiences. It is astoundingly complex to create learning experiences that effectively scale to millions while responding productively to local differences, cultures, and contexts. It’s even harder to do when the developers of technologies share a homogenous set of formative cultural and educational experiences, and the students they serve come from places both less privileged and more diverse.
Addressing the challenges of education technology and equity requires rejecting the myth that free online educational opportunities are in and of themselves democratizing. Instead, educators, developers, and policymakers must grapple with the extensive data that show us the reality: educational technology is implicated in perpetuating inequality. We now know enough about these dynamics that when the next wave of educational hype comes along, whether that is with virtual reality or artificial intelligence, we can step back and question who will be best served by new innovations.
For all of the incredible opportunities for learning that may be generated by education technology, the hard parts of creating a more equitable future through education will remain hard. And for all these challenges, there are many educators, developers, and researchers who are experimenting with a variety of approaches that do have some traction in closing digital divides. The solution space in this domain is not nearly as well understood as the problem space: we can confidently describe how new technologies typically reproduce or expand educational inequalities, but as a field, we don’t have a good handle on what kinds of approaches or strategies can reliably address new gaps along the digital fault line. There are a variety of intriguing cases, experiments, and initiatives but not enough by way of unifying theory or principles.
What the field needs is a set of “design principles for digital equity”: guidelines that can be used by funders, venture capitalists, philanthropists, developers, educators, and policymakers to guide investment decisions, development strategies, policy, and practice around the equitable use of learning technologies. My colleague Mizuko Ito and I have taken a first stab at this set of design principles, and we’ve come up with four main themes: unite around shared purpose; align home, school, and community; connect to the interests and the identities of culturally diverse children and youth; and measure and target the needs of subgroups.17
First, designers should unite around shared purpose with learners and their communities. Equity-oriented efforts can bring developers, educators, and learners together with common purpose. When initiatives are codeveloped and cofacilitated with stakeholders, they are more likely to be attuned to important elements of social and cultural contexts, and teachers and learners are more likely to take ownership of these initiatives.
Second, designers should align home, school, and community. While affluent students often have tech-savvy parents and the latest technology at home, less well-resourced students cannot count on these supports. This disconnect can be exacerbated when developers and reformers focus all of their efforts at building technology literacy and capacity in schools. One fruitful strategy for reducing this gap is building the capacity of parents and mentors alongside that of young people. Intergenerational learning experiences can strengthen family ties while giving parents and children skills to explore new domains.
Third, educators should connect to the interest and identities of minority children and youth. Peer learning communities are exclusionary when they reflect a dominant culture in ways that create a hostile environment for outsiders, but they can also be harnessed to create safe affinity spaces for minority children and youth. Powerful learning experiences result when students have the opportunity to connect their interests from outside of school to learning opportunities in more academic contexts.
Finally, designers and researchers should measure and target the needs of subgroups. When developers and reformers understand the specific needs of the communities they serve, they can deploy targeted programs that give the greatest advantage to the neediest groups. These strategies might include addressing psychological threats, addressing specific costs that matter more to low-income groups, and targeting high-risk moments in students’ learning trajectories.
People from different backgrounds and life circumstances can experience life very differently, and these experiences can lead to divergent ways of understanding the world. An affluent white computer science major from an elite college who goes to work in education technology may see the world quite differently from black and Latinx children who grow up in poverty-impacted neighborhoods. Uniting around shared purpose involves including diverse stakeholders in education technology initiatives. For her doctoral thesis, Betsy DiSalvo wanted to develop new pathways for African American boys into computer science, and she worked with a cohort of young men to create a program called the Glitch Game Testers. Glitch Game Testers became an after-school program where high school students worked on quality-assurance projects for computer game companies and were involved in shaping the trajectory of the initiative—choosing the name and logo, developing practices, and instigating more formal computer science education learning experiences. Including the young men in the design ensured that participants were engaged and that educational opportunities met the real needs of learners rather than the needs and circumstances that program leaders might have imagined.18
School-based student tech teams are another mechanism for uniting around shared purpose. The Verizon Innovative Learning Schools program is in the midst of a ten-year initiative to provide tablet computers and three years of free 4G wireless access to more than five hundred Title I middle schools across the United States. One of the requirements of the program is that schools create a student tech team to codesign technology policy, help lead the program roll-out, and serve as advisors and troubleshooters throughout the program. Giving students co-ownership of the program increases engagement and reduces disciplinary issues.19
In addition to including student voices as key stakeholders in educational programs, venture capitalists and philanthropists can include more diversity in their funding and program teams and encourage developers and entrepreneurs to include people on their teams who share the backgrounds with those they are trying to serve. Edtech companies can do more development and testing in a variety of real-world contexts and include more teachers and families as paid advisors and consultants. Uniting around shared purpose empowers new voices for leadership in education and increases the chances that designers of new edtech products and programs will meet the needs of their target audiences.
If it takes a village to raise a child, then edtech developers should consider how programs can build capacity among not just young people but their family members and caretakers as well. One promising approach for supporting educational equity for young people is creating new kinds of learning experiences for the adults around them. As we discussed in the chapter on peer-driven learning environments, Scratch is a block-based programming language and social community where young people can code up games, animations, and other programs. Qualitative interviews with Scratchers, especially from before the widespread use of Scratch in schools, revealed that many are from families in which one parent is a computer programmer or engineer or has some other connection to STEM and programming. When Scratchers grow up in families with computing expertise, their learning is supported and accelerated through that expertise. These kinds of advantages inspired Ricarose Roque, now at the University of Colorado Boulder, to do her doctoral work at the MIT Media Lab on a project called Family Creative Learning. Roque engaged parents and children in creative-technology workshops held at community-based organizations like Boys and Girls Clubs. Unlike in more traditional crafts such as knitting, parents and children have fewer intergenerational touch points when it comes to new technology, particularly among less tech-savvy parents. By hosting meals and conducting activities using Scratch and the Makey Makey invention kit within safe and welcoming spaces, the project builds capacity within families to support children’s becoming digital creators. Some activities are done separately by youth and parents or guardians, and some activities are shared and done together. Kids and families build new skills, parents have new strategies for supporting young people’s learning with creative technologies, and families have an experience that bonds and connects.20
Tech Goes Home is a similar project in the Boston Public Schools in which families can get access to a $50 laptop and discounted internet options. Parents of Boston Public School children take a fifteen-hour computer literacy course taught by a teacher in the child’s school, and at the end of the course, they can buy their computer. The course isn’t only about using technology for learning; it is also about helping parents use computers to find work and community opportunities and stay connected through social media, among other applications. Parents go home with a computer that everyone in the family can benefit from, and parents have new skills for guiding technology usage at home and a new relationship with a teacher inside their child’s school. The program simultaneously improves computer access, parent and child technology literacy, home-school connections, and teacher relationships with communities, all by situating a program about technology access in a broader social context.21
LaunchCode is a program for adults based in St. Louis that originally began as an opportunity for community members to take an introductory programming MOOC together. Jim McKelvey, who cofounded the payment processing company Square, wanted to create more opportunities in his hometown of St. Louis, so he sponsored a community group to take HarvardX’s CS50x course together. The initial demand was much greater than expected, and the program quickly moved from the local library to the Peabody Opera House to accommodate demand. CS50x is an extremely demanding course, and participants are very unlikely to finish without external support—Harvard’s on-campus version of CS50 has a veritable army of teaching assistants available almost around the clock during the semester to support Harvard undergrads. LaunchCode provided some of these same supports to help participants successfully complete the MOOC, and then moved on to include additional support for employment and interview preparation, internships and apprenticeships, and other structures to help CS50x grads (and later, grads of their own Introduction to Computer Science course) move into careers in computer programming. The HarvardX MOOC provided a valuable learning resource for bootstrapping LaunchCode, and LaunchCode built a set of human supports around the experience. These wrap-around services mean that LaunchCode looks more like a community college than a MOOC, and scaling supportive communities to create pathways of opportunity is far more difficult that scaling and spreading a new MOOC learning management system. But for those concerned with issues of equity, blended approaches to addressing inequalities are far more likely to be successful than online-only efforts.22
A third strategy for addressing inequalities with technology is to create entry points into learning experiences that connect to student interests. As discussed in Chapter 3, connected learning is a model advanced by Mizuko Ito and others that puts student interests at the center of learning design and builds connections between learning institutions so that interests that young people pursue at home and in after-school activities can have academic connections with school-based learning opportunities. Interest-based learning pathways start with student passions and showcase connections between their interests and academic subjects.23
In the Coding for All project, Scratch developers and designers are building new entry points into the Scratch community that target common interests of girls and students of color. The projects that have been historically popular and featured on the Scratch website include many with traditionally geeky themes around video games, anime, and related topics. These are enticing entry points for some young learners but discouraging for others. To create new entry points, the Scratch team is creating Microworlds that provide different ways into Scratch programming—including the Microworld mentioned in Chapter 5 that emphasizes hip-hop and others that feature such topics as fashion, art, and comedy—to help different young people connect their interests to coding.24
In Chicago, the Digital Youth Network’s Digital Divas project creates culturally relevant STEM learning activities for middle school girls offered in after-school programs. Girls learn programming and engineering through e-textiles and other design projects. In these cases, designers are not only creating engaging new programs that build on student interests, but also working with community partners to ensure that they serve lower-income and minority students.25
There is no assurance that these kinds of approaches will work. There is perhaps no group of parents that is more successful at surrounding their children with interest-driven learning experiences than upper-class white families. An edtech culture that simply focuses on aligning technologies with student interests (“all student interests matter”) will likely contribute to educational inequality, but a focus on connecting with the interests of learners alienated from schools or unable to access rich educational experiences could be a powerful bridge across digital fault lines.
Finally, serving subgroups well requires actively studying them, addressing specific needs in particular communities, and deploying targeted programs that give the greatest advantage to the neediest groups. The research that I have conducted with René Kizilcec and colleagues addressing social identity threat in MOOCs followed this model; we studied variation in how students from different backgrounds performed differently in MOOCs, identified ways we might be able to support particular groups of learners who might need extra scaffolding, and experimentally tested different interventions, first in pilots and then in larger scale replications. Only by understanding how different subgroups experience learning at scale differently can we try to address some of the gaps in opportunity that emerge.
The issue of pricing reveals how similar product features can operate very differently in different social contexts. From 2013 to 2018 or so, most MOOCs were available for free (during this period, Udacity, Coursera, and eventually edX added a variety of paywalls to their products), and the evidence suggests that the bulk of benefits of MOOCs accrued to more affluent learners. But there are other online learning experiences that set prices at zero that do appear to disproportionately benefit the families furthest from opportunity. In particular, when the cost of an educational product is substantial for low-income families but trivial for affluent families, free goods may be particularly effective at closing gaps.
OpenStax develops free, peer-reviewed, openly licensed textbooks for introductory college courses. A substantial portion of all college enrollments in any given semester are in a relatively small number of introductory courses: algebra, biology, calculus, economics, psychology, and government, among a few others. The textbooks for these survey courses can cost more than $100, a substantial burden on students in community colleges and other settings who may be paying only a few hundred dollars per credit hour. OpenStax claims that by providing free alternatives, in the 2016–2018 school years, they saved students about $177 million in textbook costs. While students from all backgrounds may be benefiting from these resources, the families and students benefiting most are those for whom a $175 textbook represents a major financial hurdle.26
Similarly, Desmos has developed a free browser-based graphing calculator as a direct competitor to the Texas Instruments line of calculators, such as the TI-84+ that retails for over $100. The Desmos graphing calculator has substantially more functionality than a handheld calculator and represents a major improvement in terms of accessibility through integration with screen readers and other accessibility software. As with the cost of a textbook in the case of OpenStax, the $100 price tag for a TI calculator is modest for affluent families but a substantial burden for low-income schools and families—especially when many families already sacrifice to make more fully functional laptops, phones, and tablets available to their children.
Targeting subgroups can also mean understanding the barriers to access and progress that are unique to particular groups of people. In 2014, Ben Castleman and Lindsay Page defined a phenomenon that they called “summer melt.” Castleman worked at the MET School in Providence, Rhode Island, an urban charter school that was unusually effective at supporting students in making it through high school and toward graduation with acceptances to colleges and universities. As MET staff tracked their graduates past high school, they discovered a shocking phenomenon: a surprising number—between 10 and 40 percent—of high school graduates who were accepted into college did not register for their first semester. This is a problem almost unimaginable to the parents of elite students, who can assume that college matriculation will naturally lead to attendance and graduation. In studies of three large urban districts, they discovered that schools that had successfully improved graduation rates and college acceptance rates were losing students in the transition to college.27
The bureaucratic hurdles that high school graduates faced throughout the summer in registering for classes and applying for financial aid were particularly mystifying for first-generation college students. Castleman and Page joined with foundations, schools, and other researchers in try to tackle this particular challenge, though their initial efforts have had mixed results. They launched a series of text-message-based interventions that reminded students of key dates and actions related to enrollment, registration, financial aid, and orientation. These interventions didn’t address every challenge for every student, but they successfully raised college entry by several percentage points at very low cost. Unfortunately, these initial improvements in registration did not lead to improvements in graduation rates among students receiving the “summer melt” treatment; students who got a boost over the bureaucratic hurdles into college didn’t necessarily persist all the way through. The ultimate success of Castleman and Page’s research program will depend upon deeply understanding the needs and challenges of an important group of students and then identifying widely accessible technologies that could be used to support students’ learning trajectories.28
As a field, we understand the challenges of digital fault lines much better than we understand potential solutions. We have a substantial body of research that characterizes myths and realities about education technology and equity, but our understanding of solutions is a spottier collection of case studies. The field needs a surge of research into strategies for digital equity to guide new education technology efforts.
One way to approach this research effort would be to consider the full life cycle of an education technology product through a set of stages—bringing together a team, funding an idea, developing the technology, selling and marketing to schools and learners, implementing in schools or other environments, collecting feedback and data, and evaluating programs. Issues of equity can be addressed at each of these stages: Does the founding team have a diverse leadership that can bridge social gaps between technology developers and student communities? Will the funders hold grantees or entrepreneurs accountable for addressing equity goals? Are the learners who need the most support included as codesigners in the development process? Do data collection and evaluation practices investigate the needs, strengths, and opportunities of different subgroups?
To provide guidance at each of these stages, we need much more research about effective, equity-focused practice. The case studies we have of Tech Goes Home, Family Creative Learning, OpenStax, and other initiatives are a terrific start, but we need to develop a much richer understanding of the programs and strategies that are effectively addressing the needs of the learners farthest from opportunities. From additional case studies, we could identify practices worthy of more rigorous research and develop a set of design principles for digital equity that could guide the work of all the different stakeholders who have a hand in supporting learning through new technologies. The principles in this chapter offer a starting point for that work.