At some point this year, a board member is going to bring a book about screens to a superintendent and ask what the district plans to do about it. The book will be serious, the board member will have read it carefully, and the question will be fair. The superintendent will be sitting in a district that spent the better part of a decade putting a device in front of every student, and the case for doing that was serious too. Devices were going to personalize learning, increase engagement, expand access, and prepare students for a changing world. Districts invested heavily, one-to-one access became part of the ordinary infrastructure of school, and the evidence that followed was less clear than the promise. Greater access alone did not reliably improve learning, though some specific uses did.
Now there is serious research supporting a move in the other direction. What the research did not do the first time, and cannot do this time, is tell anyone how much evidence should be enough. That question mostly went unasked when we were adding devices, and it is going unasked again now that we are talking about taking them away.
A strong case still requires a decision
Jared Cooney Horvath makes a serious case in The Digital Delusion for substantially reducing student-facing technology. He does more than point to falling test scores and blame laptops. He brings together international assessment data, research on reading and note-taking, evidence about distraction, large reviews of educational technology, and cognitive research about how people learn, and he acknowledges along the way that some narrower uses of technology can help. His conclusion goes considerably further than asking schools to be more thoughtful about screens. He argues that educational technology has contributed substantially to declining learning and cognitive performance and that schools should reverse course.
I am less interested in settling whether Horvath is right about technology than in something his argument shows about how these decisions get made.
At one point he reviews the large body of research on educational technology and finds that the effects are, on average, positive. He points out that nearly every educational intervention shows some positive effect when it is studied, so positive by itself is not the same as meaningful. He sets a line for what counts as a meaningful gain, chooses the more generous end of the range researchers use, says so, and then shows that most educational technology falls below it. That is a fair way to work. It is also a judgment. The studies estimated the effects. Horvath decided how large an effect had to be before it mattered, and a district reading his book will have to decide whether it agrees, because nothing in the studies makes that call.
That is the part that transfers. Every consequential recommendation contains the same judgment somewhere, usually unstated. The research can tell us that a benefit is likely, or that a harm is plausible, and roughly how large either one might be. Someone still has to decide whether that is worth the money, the time, the disruption, and whatever else the district gives up to pursue it.
How much is enough
We already know how to do this when the numbers are ours. If reading scores rise after a school introduces a new intervention, we ask what else changed, who was tested, whether staffing turned over, whether the improvement was widespread or driven by one group, and whether the same pattern shows up anywhere else. Nobody thinks those questions mean the improvement is fake. We ask them because we want to know how much weight the result can carry before we build something on it.
The same judgment applies to research from somewhere else, and the answer depends on what we are about to do with it. Trying a low-cost tool with one class for a semester is a different decision from requiring it across a district. A practice that can be changed next month is different from a multiyear contract, years of professional learning, and a change in what thousands of students do every day. As the cost, the disruption, and the difficulty of reversing a decision go up, it is reasonable to want more confidence before making it, and no study sets that bar for us.
That does not mean waiting for certainty. In education we would be waiting a long time. The students in a study may not look like ours, the conditions rarely match, and with technology the product usually changes before a mature research base ever develops. Limited evidence may be enough to justify trying something carefully. Evidence of possible harm may justify a restriction before every causal question is settled. A real benefit may still be too small to justify what the district would have to spend or give up to get it. The research helps with each of those judgments. It does not make any of them for us.
Strong enough to act
We are very good at saying something is research-based. The harder work is explaining why the research is strong enough to justify what we are proposing, and for a consequential recommendation we should be able to finish one sentence:
We believe the evidence is strong enough to act because...
That sounds simple until you try to write what comes next. If a district is preparing to make a major multiyear investment in an instructional platform, citing studies that show the product can work is only the beginning of the case. The recommendation still has to say why those findings apply to the students we serve, why the likely benefit justifies the cost and disruption, what the alternatives were, and what uncertainty we are accepting by going ahead. The same is true when the recommendation is to pull something back. Evidence that a practice may be harming learning matters, and it may matter enough to act before every question is answered, but someone still has to connect that evidence to this action, in this district, and explain why this response fits these conditions. That reasoning belongs in the recommendation itself, where the next person can find it, and "research-based" should not have to carry the whole argument.
Write down why it was enough
There is a second reason to put the reasoning on paper, and it has to do with what happens after the vote.
A district should be able to come back to a decision later, and that gets harder than it sounds. People change jobs, budgets move, the original presentation disappears into a shared drive, and the conversation shifts from why we bought something to how many people are using it. By the time a contract comes back for renewal, the fact that we already have it can become the strongest argument for keeping it.
Knowing what we expected to see is half of what a later group needs. The other half is knowing what we thought was enough. A recommendation should leave behind what we expected to happen, why the evidence made that expectation reasonable, how much uncertainty we accepted, and what result would count against the decision. If a district restricts a student-facing technology because leaders expect fewer distractions and stronger learning, the recommendation can name what it will watch over the next year and what would force a second look. If a district adopts an instructional tool because it expects stronger writing or mathematics outcomes, it should say how long that should take to show up and what would tell us the investment is not delivering. None of this requires predicting the future, only being clear enough about the reason for the decision that a different group of leaders can evaluate it without having been in the room.
The reasoning should outlast the decision
The technology debate makes this unusually easy to see because the direction has changed so completely. Districts once had research supporting more technology and now have serious research supporting less of it, and I do not think the lesson is that the research failed us either time. Evidence can point a district in a direction while leaving its leaders to decide how much benefit is enough, how much uncertainty they can live with, and what happens if the assumptions behind the recommendation turn out to be wrong. That was true when the devices were going in, and it is true now.
Three or five years from now, the people making the next decision may not have been in the room for this one. They will inherit the contract, the policy, the training materials, and a folder of old presentations, and they will have no way of knowing which assumptions mattered most when the choice was made or how confident anyone actually was. They should inherit the reasoning too.