In a 2025 survey of 350 teachers, 99 percent said their school gave students devices for classroom use, but just 4 percent said their school set any limits on screen time. That gap between deployment and evaluation is not a technology problem but rather a behavioral one. The same herd mentality that drove schools to adopt unproven ed tech without evaluation is now driving the coast-to-coast ed tech backlash.
The bias is the bandwagon effect, which refers to our tendency to adopt what others adopt, regardless of whether it works. When COVID forced schools online overnight, panicked districts bought what other districts had already bought. My co-authors and I have published two randomized experiments this year that directly measure the bandwagon effect in ed tech adoption. We surveyed 1,104 teachers about their likelihood of recommending math learning apps. Telling teachers an app had strong research evidence increased their likelihood of recommending it by 0.24 standard deviations. Telling them the same app was popular among other teachers increased it by 0.21. These two effects are each substantial, but they are statistically identical. However, in a second experiment, showing 154 school principals a two-minute informational video presenting research findings supporting the use of a digital math game produced no significant change in their likelihood of recommending it or their willingness to pay for it. In this case, the evidence presented directly moved nothing. The market told vendors that they didn’t need rigorous research to make a buck.
Is The Ed Tech Backlash A Mistake?
The current ed tech backlash is the bandwagon bias running in reverse. Los Angeles voted 6-0 to restrict devices, banning them for first graders. Legislators in 16 states have proposed similar restrictions. Each move signals the next. The problem? This approach treats a phonics game and a YouTube rabbit hole as identical problems because they both involve a screen.
The evidence doesn’t support either extreme. A 2020 review of 19 rigorous studies of the effectiveness of technology-based approaches to teaching and learning found that 12 showed substantial positive effects on test scores, ranging from 0.14 to 0.56 standard deviations, with an average of 0.18. One of those studies tested math achievement in 2,850 seventh-grade students in Maine, finding that an online homework tool improved math scores by 0.18 standard deviations even though students used it for less than 10 minutes per night. A 2024 meta-analysis of 119 digital literacy interventions found positive effects across decoding, comprehension, and writing. These tools share three features that most deployed tools lack: deliberate design, independent evaluation, and a specific instructional purpose. Throwing out the tools that work along with the ones that don’t is a reaction, not a policy.
The bandwagon effect thrives because evaluation is hard, research is technical, vendor claims are opaque, and independent reviews are scarce. Nobel laureate Richard Thaler and Cass Sunstein, in their foundational work on choice architecture, showed that one doesn’t fight a cognitive shortcut by demanding people think harder. One wins that battle by making the right information as easy to access as the wrong information.
Popularity Doesn’t Equal Evidence
The bandwagon effect thrives because evaluation is hard, research is technical, vendor claims are opaque, and independent reviews are scarce. Nobel laureate Richard Thaler and Cass Sunstein, in their foundational work on choice architecture, showed that one doesn’t fight a cognitive shortcut by demanding people think harder. One wins that battle by making the right information as easy to access as the wrong information. Our experiments found that research evidence and peer popularity moved teachers by the same amount, and that when both signals were equally visible, the outcome was a coin flip. Educators are not wilfully ignoring evidence. The problem is that the procurement environment puts evidence and popularity on equal footing.
One might argue that peer adoption is a reasonable proxy for quality and that if most teachers recommend an app, it probably works. But national test scores have been falling steadily for over a decade, during the exact same period schools were racing to embrace technology. The herd was wrong. The fix is to restructure the decision environment so that evidence is present at the key moments of the purchase pipeline, as legible as a vendor’s claim of popularity. Think of it as a nutrition label for ed tech, like a mandatory independent evidence rating that appears alongside every product a district considers buying. The bandwagon signal doesn’t disappear, but it loses its information advantage.
Our experiments found that research evidence and peer popularity moved teachers by the same amount, and that when both signals were equally visible, the outcome was a coin flip. Educators are not wilfully ignoring evidence. The problem is that the procurement environment puts evidence and popularity on equal footing.
Evaluating Ed Tech With Data Is A Must
Randomized trials can take years, longer than the product cycles of a fast-moving market. Demanding an RCT before every purchase would paralyze procurement. But the choice isn’t between a five-year trial and nothing. Vendors already collect granular usage and outcome data at scale. What’s missing is the will to share it and the infrastructure to analyze it independently. A/B testing, such as randomizing features or dosage across classrooms, can generate causal evidence within a single school year. Rapid-cycle evaluations designed and analyzed by independent researchers can distinguish signal from noise in months rather than years. All of this requires researchers and developers working together from the design stage, not after the fact.
For school boards: Before banning devices, ask whether the specific tools in your classrooms have been independently evaluated.
For foundations: Redirect a portion of ed tech adoption funding toward rapid-cycle independent evaluation with open data and findings published regardless of outcome.
For state policymakers and ed tech developers: Build the research partnership before the product launches, not after the backlash arrives. The same cognitive shortcut that filled classrooms with unproven technology is now clearing them without asking which is which. That’s the bandwagon, not progress.
For parents: Before deciding whether to support your school board’s device ban, don’t default to arguing about whether screens are good or bad. Ask whether the environment requires anyone to find out.
1 thought on “How Behavioral Bias Broke Ed Tech — and How to Fix It”
Hi, i want more information about your services thak foe everything.