“You don’t understand how to teach these kids.” This was something a principal said to me after I pointed out a pervasive issue of low expectations throughout the building. I have worked with many “low-performing” schools, and I often hear principals say there is a deficit mindset in the school.
He was not wrong that something in the building did not add up, but was wrong about where it lived. He had put the deficit in the students, and in me for not accepting it, when it was sitting in the expectations the building held for both. This is the pattern I have seen multiple times in low-performing schools. The leader who names a deficit mindset in their school is very often standing inside the thing they are describing. From where they sit, it does not feel like low expectations. It feels like knowing his kids
"These kids," the ones he was sure I did not understand, have been measured and missed the mark more times than anyone can count and keep showing up and learning anyway, which we treat as a surprise instead of a signal.
Doing the Impossible: The Bee Story
The “Bee Movie” starts with "According to all known laws of aviation, there is no way that a bee should be able to fly. Its wings are two small to get its fat little body off the ground. The bee, of course, flies anyway."
It is a great story that is often used as inspiration. It is also wrong, in a way that matters more than the story does.
There Was Never Any Proof
The tale most reliably traces back to 1934, when the French entomologist Antoine Magnan and his assistant, an engineer named André Sainte-Laguë, ran the numbers on insect flight (Magnan, 1934). They used the equations for fixed wings, the kind that keep an airplane aloft. Treated that way, a bee's wing does not generate enough lift. The conclusion was not that bees cannot fly. Magnan could see them flying. The conclusion was that the mathematics of the day could not explain how.
That is the whole story, and it is a better one. The bee was never breaking a law of physics. The model was using the wrong physics.
They Modeled the Wrong Animal
It took decades to build the right model. In 1996, Charles Ellington's group showed that insect wings generate a leading-edge vortex, a small tornado of air riding along the front edge of the wing, that produces enormous extra lift (Ellington et al., 1996). Then Michael Dickinson and his colleagues built a robotic wing the length of your forearm, dropped it into a tank of mineral oil to match the physics of a real bee's scale, and measured the forces at every point in the stroke (Dickinson, Lehmann, and Sane, 1999). They found three mechanisms working together, all of them depending on the fast, flapping, unsteady motion that the fixed-wing equations had thrown away.
The bee flies with physics, not in spite of it. The math simply modeled the wrong animal.
Every School Runs a Model
Every school runs on a model of what a child can do. We call it a lot of things: readiness, grade-level, ability, or data. Whatever the name, it is a set of equations that takes a student's inputs and predicts an output, and then we act on the prediction. Most of the time the model is useful. Sometimes it fails the same way Magnan's did. It measures a child against a template built for a different kind of learner, finds insufficient lift, and stops looking.
Now hold that next to your building. If your model is surprised when a student sitting in front of you is successful. That means it's time to fix the model.
A Miracle Is a Measurement Error
Instead of looking at a different model, we call the child an exception. We say she is beating the odds, defying expectations, overcoming. Listen to that language for a second. It sounds generous, but it is actually a confession that our prediction stands and the kid is the anomaly. We would rather call a success a miracle than admit our predictions were miscalibrated. But a miracle is just a measurement error wearing its best clothes.
The research on this is not subtle. Rosenthal and Jacobson gave teachers a fake list of students supposedly poised to bloom, and the children randomly chosen measurably outperformed their peers because the adults treated them differently (Rosenthal and Jacobson, 1968). The expectation did not just predict the outcome. It manufactured it. Christine Rubie-Davies has spent a career showing that high-expectation and low-expectation teachers produce different students out of similar ones, through the ordinary machinery of grouping, questioning, feedback, and pace (Rubie-Davies, 2015).
And this is not really an accuracy problem, which is part of why it hides so well. On where a student stands today, teachers are fairly accurate (Südkamp, Kaiser, and Möller, 2012). Judgment slips are about where a student could go, and they slip in patterned ways rather than random ones.
When two teachers recorded expectations for the same student, the predictions diverged along the teachers' own demographics, with non-black teachers tending to expect less of black students than black teachers did of those same children (Gershenson, Holt, and Papageorge, 2016). None of this requires anyone to think less of a child on purpose. It runs quietly, through the same habit that shortens a pause or simplifies a task. And it does not stay a prediction.
Tenth-grade teacher expectations measurably shift a student's odds of finishing college, so the forecast helps produce the outcome it claims to read (Papageorge, Gershenson, and Kang, 2020). The instrument is not broken. It is calibrated a few degrees off, in a consistent direction, and it acts on what it measures.
This is the fixed-wing trap in a classroom, and it has a name. Researchers call it the Golem Effect, the negative twin of Pygmalion: once the model says "cannot," we quit looking for lift, and the prediction makes itself true (Babad, Inbar, and Rosenthal, 1982).
So the bee gives us the wrong lesson if we take the poster version. "Believe and you can overcome" puts the whole burden on the child and lets the system off the hook for free. The systems-minded version is harder and truer. When a student outperforms your model of him, that is not a story about his grit. It is data about your model.
Hunt the Vortex, Not the Miracle
What does the honest response look like. It looks like Dickinson, not the motivational speaker. When the bee did the impossible thing, Dickinson did not marvel and move on. He got curious about the mechanism he could not yet see. He built the instrument that would let him see it. That is the move. When a kid succeeds against your prediction, the question is not "how inspiring," it is "what did we fail to measure." What was this student doing, or being given, or bringing, that our model does not have a variable for. Find that, and you have not celebrated one exception. You have found the vortex. You can build for it, and every child who needed it gets more lift.
The low-expectation reflex hides inside reasonable-sounding sentences.
- "He's just not there yet."
- "That group can't handle the complex text."
- "Given their background, this is about what you'd expect."
Each one is a fixed-wing calculation, run on a napkin, at a dinner party, decades ago. Each one predicts insufficient lift and then quietly stops the search. The work is to catch the prediction before it hardens into a fact. Below is what that looks like on a Monday.
Remediation Looks Back. Previewing Looks Forward.
Nowhere does the fixed-wing calculation get built into a building more completely than in what we do with students who are behind. The standard model is remediation and intervention. Pull the student out of grade-level content, send them back to the skills they missed, hold the harder work until they have caught up. It sounds like help. It is deficit thinking turned into a master schedule.
Watch what it does. A student judged not ready gets less demanding work, so they see less grade-level content than their peers, and the gap that justified pulling them out grows wider. Remediation is the Golem Effect with a bell on it. The prediction writes the schedule, and the schedule confirms the prediction.
Previewing runs the other way. Instead of looking back at what a student lacks, you look forward at what is coming and frontload it, the vocabulary and background knowledge they need to walk into next week's grade-level lesson ready to work. Same student, opposite theory of who they are. Remediation treats the deficit as the truth about the child. Previewing treats the grade-level task as within reach and builds the lift before takeoff instead of studying the fall.
That is the whole argument made operational. Fixing the model is not a posture you adopt in a meeting. It is a decision about where a struggling student spends third period.
What This Looks Like for Leaders
Interrogate the language of prediction. In data meetings and PLCs, ceilings hide inside sentences that sound like analysis. Install one question as a standing norm: is that an observation or a prediction? An observation describes what a child did this week. A prediction describes what you have already decided they will do. The first is evidence. The second is a hypothesis nobody is allowed to act on until it has been tested.
Run a bright-spot review. Once a cycle, pick a student who is outperforming what your model predicted, and refuse to just celebrate it. Spend fifteen minutes reconstructing the mechanism. What is this teacher doing, or this child bringing, that our data has no column for? You are not admiring an exception. You are hunting the variable you failed to measure, so you can give it to the next twenty kids who need it.
Put expectation into the walkthrough you already do. Stop scoring only the lesson. Track distribution. Who gets the higher-order questions and who gets recall. Who gets five seconds of wait time and who gets rescued in one. Who gets specific, next-step feedback and who gets a nod. The gap between your strongest and weakest students on those three lines is your expectation problem, made visible on a form you already carry.
Audit the groups that have not moved since September. A group that has been "the low group" for six months is the fixed-wing equation set in amber. Rubie-Davies found that high-expectation teachers group flexibly and give every group access to genuinely demanding work, while low-expectation teachers sort early and hold the sort (Rubie-Davies, 2015). Ask who has been locked in place, and what it would take to move them this month.
Rebuild the intervention block around what is coming, not what is past. Most support time is spent reteaching last year's gaps while the student misses this year's core, which is the remediation trap running on a schedule. Flip its purpose. Use that block to preview the vocabulary and background knowledge for the grade-level lessons landing next week, and protect core instruction from pull-outs. The question shifts from what did this student miss to what will this student need on Tuesday.
What This Looks Like in the Classroom
The four behaviors a student actually sees and feels, the level of challenge you assign, how long you wait after a question, the feedback you give instead of a grade, and who you call on to do the thinking, are their own discipline. I walked through all four, with the research and the coaching questions behind each, in 4 Teacher Behaviors That Communicate High Expectations to Students. Those are the how. The two moves below come before behavior. They decide whether you ever reach for it, because they happen inside your model of the child.
Turn "he can't" into a test. When the thought arrives, and it will, convert it from a conclusion into a hypothesis. What would I have to try, and how many times, before I am actually allowed to believe it? Then go try that. Most ceilings do not survive the third serious attempt.
Do not reserve the rich task. The complex text, the open problem, the discussion you were saving for the top group. Give it to the group you were about to hand a worksheet. Access is the first expectation, and every other one follows it.
Trust Your Eyes
None of this means every model is wrong or every high expectation is met. Bees fly. Not everything does. The point is narrower and sharper than blind optimism. It is this. When your data and your eyes disagree about a child, trust your eyes and go fix your data. The bee was flying the whole time. The failure was never in the bee.
References
Babad, E. Y., Inbar, J., and Rosenthal, R. (1982). Pygmalion, Galatea, and the Golem: Investigations of biased and unbiased teachers. Journal of Educational Psychology, 74(4), 459 to 474.
Dickinson, M. H., Lehmann, F.-O., and Sane, S. P. (1999). Wing rotation and the aerodynamic basis of insect flight. Science, 284(5422), 1954 to 1960.
Ellington, C. P., van den Berg, C., Willmott, A. P., and Thomas, A. L. R. (1996). Leading-edge vortices in insect flight. Nature, 384, 626 to 630.
Gershenson, S., Holt, S. B., and Papageorge, N. W. (2016). Who believes in me? The effect of student-teacher demographic match on teacher expectations. Economics of Education Review, 52, 209 to 224.
Magnan, A. (1934). Le Vol des Insectes. Paris: Hermann et Cie.
Papageorge, N. W., Gershenson, S., and Kang, K. M. (2020). Teacher expectations matter. Review of Economics and Statistics, 102(2), 234 to 251.
Rosenthal, R., and Jacobson, L. (1968). Pygmalion in the Classroom: Teacher Expectation and Pupils' Intellectual Development. New York: Holt, Rinehart and Winston.
Rubie-Davies, C. M. (2015). Becoming a High Expectation Teacher: Raising the Bar. London: Routledge.
Südkamp, A., Kaiser, J., and Möller, J. (2012). Accuracy of teachers' judgments of students' academic achievement: A meta-analysis. Journal of Educational Psychology, 104(3), 743 to 762.
