9-12 · 60 min · Math · Science · Social Studies · ELA · CS/Technology
The plugged companion to Systems & Feedback: students collect or load real data, analyze it in a spreadsheet, audit it for bias, and write an evidence-based recommendation — meeting data science and the discipline of trusting data only after you've checked it.
Materials last updated Aug 3, 2026.
60 min in class~15 min prepDevices required
The hookContextualize6m
Fix the misconceptionReframe5m
Do the activityAssemble27m
Check the machineFortify14m
Wrap up + connect forwardTransfer + review8m
Before class~15 min
Load one small, clean dataset per team on the devices (a few dozen rows, tied to a real question).
Print the Audit checklist (audit-checklist.pdf): sampling, missing data, display, confounds.
Have a real published claim or poll ready for the Transfer audit. Teaching one subject? Also print its page: Math stats-summary, Science CER-uncertainty, Social Studies representation-memo, ELA argument-hedging, or CS bias-at-scale.
Data to Decision
Make it yours
One lesson, woven into your subject
No co-teacher needed. Open your subject for a single card with the core
content you teach and the specifics for weaving this lesson into
your room — nothing to look up elsewhere.
Topic refresher
New to a concept? Tap a topic for a printable cheat sheet — plain-language definitions and classroom examples.
Teacher cheat sheet · refresher
Data to Decision
9-12 · 60 min · Math · Science · Social Studies · ELA · CS/Technology
Woven together — one idea, every subject
One idea runs through every subject here: data only becomes evidence after you audit how it was made — sampling, missing data, display, confounds — and a sound recommendation states its claim, its evidence, and its limits. Statistics, scientific argument, civic data, argument writing, and algorithmic accountability are all that same verify-before-you-act discipline.
Math — Summarize center and spread; correlation isn't cause.
Science — A result isn't a finding until you state its uncertainty.
Social Studies — Who's in the data — and who's missing — shapes the policy.
ELA — 'Suggests' vs. 'proves' — hedging is the whole lesson.
CS/Technology — Bias in the training data becomes bias at scale.
Concepts in this lesson
Data collection & analysis
This is the cycle of gathering data, organizing it, picturing it (a chart or map), and drawing an evidence-based conclusion — using what the data actually shows rather than a hunch.
ExampleSurveying the class, tallying responses, building a bar graph together, and stating one claim the graph supports and one it does not.
Key wordsdata · tally · chart · claim / evidence
Watch forA claim must be supported by the data on the page. Watch for conclusions that go beyond what was actually measured.
Bias & data ethics
Data can be skewed by who and what gets counted — and using it has consequences for real people. Bias is a systematic tilt in the data; data ethics is asking whether collecting and acting on it is fair.
ExampleA "favorite recess game" survey taken only of the soccer team — then asking whose preferences were missed and who that would affect.
Watch forBias usually hides in who was left out, not in the math. Always ask "whose data is missing, and would they change the answer?"
Run it in your subject — core content + how it weaves in
Math
Summarize center and spread; correlation isn't cause.
Core contentS-ID and S-IC.B.6 are statistics: summarize data with center (mean/median) and spread (range, IQR, standard deviation), choose a display that fits the data type, and evaluate reports based on sampling. In plain terms: describe the data with a typical value and how spread out it is, remember a small sample carries more uncertainty, and never claim one thing caused another just because they move together — name a possible confounder first.
StandardsS-ID · S-IC.B.6 · 7.SP
Weave it inThis is your statistics-and-inference unit made real (S-ID, S-IC.B.6, 7.SP). Run it as: (1) Students compute center (mean/median) and spread (range, IQR or SD) for their variable and choose the chart that fits the data type — bar for categories, histogram/box plot for continuous, scatter for two variables. (2) They state the sample size and discuss variability: would a different sample change the result? (3) Introduce margin of error informally — 'a result from 30 people carries more uncertainty than one from 3,000.' Watch the classic error: treating correlation as causation; require students to name a possible confounding variable before claiming a cause. Deliverable: the chart + a one-sentence claim the statistics actually support.
Science
A result isn't a finding until you state its uncertainty.
Core contentThe HS science practices of analyzing/interpreting data and arguing from evidence mean separating signal from noise and supporting a claim with reasoning. In plain terms: plot the data, name the trend and the scatter around it, state the result as a value with uncertainty (X ± Y), and say what evidence would prove the claim wrong — a claim-evidence-reasoning argument.
StandardsHS SEP: data & argument · CER
Weave it inFrame it as analyzing experimental or environmental data and arguing from evidence (HS SEP: analyzing & interpreting data, engaging in argument). (1) Use real data — class experiment results, local air/water readings, or a public dataset (e.g., NOAA, USGS). (2) Separate signal from noise: plot it, identify the trend and the scatter around it, and state the uncertainty explicitly ('the trend is X ± Y'). (3) Students write a claim-evidence-reasoning (CER) paragraph and must state what would falsify their claim. Tie-in: a result isn't a finding until you've quantified how confident you are and ruled out alternative explanations — the scientific version of Check the Machine.
Social Studies
Who's in the data — and who's missing — shapes the policy.
Core contentC3 D3.1 and D4.1 are evaluating sources and using evidence to develop and communicate a claim, including informed civic action. In plain terms: civic data (turnout, budgets, surveys) carries the choices of who got counted, so a responsible recommendation names an excluded group, says how including them might change the conclusion, and states whose interests the data serves.
StandardsC3 D3.1 · C3 D4.1
Weave it inUse civic, economic, or historical data to argue a policy recommendation (C3 D3.1, D4.1). (1) Pick a question with stakes — school start times, a local budget choice, voter turnout by group. (2) The central move is representation: who is in the data and who is missing, and how does that shape the conclusion? Have students name an excluded group and how including them might change the recommendation. (3) Students write a recommendation memo with claim, evidence, and an acknowledgment of the data's limits and whose interests it serves. This builds the civic-data literacy to read polls and government statistics critically.
ELA
'Suggests' vs. 'proves' — hedging is the whole lesson.
Core contentW.11-12.1 and RST.11-12.7 are writing arguments with precise claims, evidence, and reasoning and integrating quantitative information. In plain terms: a strong argument runs claim → evidence → warrant (why the evidence supports the claim) → counterclaim → rebuttal, and the verbs matter — 'the data suggests' and 'the data proves' are different claims, and each must cite evidence the data actually contains.
StandardsW.11-12.1 · RST.11-12.7
Weave it inThis is argument writing with real evidence (W.11-12.1, RST.11-12.7). (1) Students structure the recommendation as claim → evidence (their data + chart) → warrant (why the evidence supports the claim) → counterclaim (the strongest objection) → rebuttal. (2) Require precise hedging language: 'the data suggests' vs. 'the data proves' — the difference is the whole lesson. (3) Peer review targets one thing: does each claim cite evidence the data actually contains? Deliverable: a one-page evidence-based argument. Reading tie-in: analyze a published op-ed or report for the same structure and for where it overreaches its evidence.
CS/Technology
Bias in the training data becomes bias at scale.
Core contentCSTA 3A-DA and 3A-IC-24 cover data analysis at scale and the social impacts of computing. In plain terms: a recommendation engine or model does the same pattern-finding students did by hand, but over millions of rows — so if the training data underrepresents a group, the system's outputs will too, invisibly and at scale, which is why auditing the data and testing outputs across groups matters.
StandardsCSTA 3A-DA · CSTA 3A-IC-24
Weave it inConnect the hand-analysis to how algorithms and AI scale it (CSTA 3A-DA, 3A-IC-24). (1) Students do the analysis manually, then discuss: a recommendation engine or AI model does this same pattern-finding over millions of rows. (2) The critical idea: bias in the data becomes bias in the system — if the training data underrepresents a group, the model's outputs will too, at scale and invisibly. (3) Optional: have students articulate one safeguard (audit the training data, test outputs across groups, keep a human in the loop). This is the through-line from a spreadsheet to algorithmic accountability.
Loom & Lesson — printable teacher refresher · Data to Decision
Overview
The screen-side twin of Systems & Feedback. Students take a real question,
work a dataset in a spreadsheet, audit it for bias and validity, and produce
an evidence-based recommendation — a one-pager with a claim, the data behind
it, and its limits. They practice data science and, crucially, the habit of
trusting data only after checking how it was made. Point the question at
whatever you teach.
A base integrated lesson (plugged) — evidence-based reasoning across math,
science, and civics that students steer toward your subject — run it in your
room as-is.
Pre / Post assessment
Pre: “A dataset says X. What do you check before you act on it?”
Post: “What’s the strongest objection to your recommendation, and how does your data answer it?”
Objectives
Students will (1) analyze a dataset to address a question, (2) audit it for
sampling/validity issues, and (3) write a recommendation that states evidence
and limits.
CONTEXTUALIZE — why it matters
The decisions that shape a community — where a clinic opens, how a budget is
split, which policy advances — increasingly run on data, and bad data quietly
produces bad decisions that fall hardest on the people who were never in the
dataset. This is the work of scientists, economists, planners, and the engineers
who design the systems that crunch it. The students who learn to interrogate
evidence before acting on it are the ones equipped to steer those decisions
responsibly — and to be the experts their communities trust with the question.
Spreadsheets and AI tools are just instruments here; the judgment is the point.
REFRAME — surface the wrong model, install the right one
Students treat data as objective truth. Reframe: data is shaped by who was
measured, how, and what’s missing — analysis without an audit can launder bias
into a confident-looking chart.
ASSEMBLE — I do / we do / you do
I do: Load a small dataset, compute a summary, and make one chart; state a tentative claim.
We do: Run the audit checklist on that dataset together; revise the claim.
You do: Teams analyze their dataset, audit it, and draft a recommendation one-pager.
FORTIFY — Check the Machine
The audit is the Check the Machine step, applied to data: (1) Sampling —
who/what is included, and does it represent the population the claim is about?
(2) Missing data — what’s absent, and does it skew the result? (3) Display
— does the chart’s scale or framing mislead? (4) Confounds — could something
else explain the pattern? Only after the audit do teams commit to a
recommendation, explicitly stating its limits. Tie directly to evaluating
AI-generated analysis and statistics.
TRANSFER — forward
Connect back to Systems & Feedback: data reveals how the system actually behaves.
Forward: examine a real published claim or poll and run the same audit on its methodology — then name a decision in your own community that data is steering, and who is positioned to ask whether the evidence holds. The students who can audit it are the ones who could one day design or govern the systems that produce it.
What to listen for
Use the Post prompt — “What’s the strongest objection to your recommendation, and how does your data answer it?” — as your read on mastery.
Proficient: states the strongest objection and answers it with the audit. “Strongest objection: we only surveyed one class, so it may not represent the grade. We flag that as a limit and don’t generalize past it.”
Getting there: analyzes the data and recommends, but skips the audit. Walk the four checks before they commit.
Not yet: treats the dataset as objective truth. Reframe: data is shaped by who was measured, how, and what’s missing.
Proficient when a recommendation names its evidence and at least one limit (sampling, missing data, display, or a confound) the data can’t rule out.
Differentiation
Support: provide a clean dataset and the audit checklist; focus on one chart + one claim.
Extension: introduce correlation vs. causation, sample-size effects, and a basic confidence statement.
3-2-1 Review
3 findings from your data · 2 audit issues you caught · 1 limit your
recommendation plainly states.
Family / community connection
“Find a statistic in the news. Run the audit — who was sampled, what’s missing,
does the chart mislead — then decide whether you believe it.”
The national computer-science learning standards from the Computer Science Teachers Association.
3A-DA-11+
Data & Analysis strand, grades 9–10
3A-DA-12+
Data & Analysis strand, grades 9–10
3A-IC-24+
Impacts of Computing strand, grades 9–10
CS/Technology:CSTA 3A-DA and 3A-IC-24 cover data analysis at scale and the social impacts of computing. In plain terms: a recommendation engine or model does the same pattern-finding students did by hand, but over millions of rows — so if the training data underrepresents a group, the system's outputs will too, invisibly and at scale, which is why auditing the data and testing outputs across groups matters.
You just read a free Math lesson. There are 31 more, going deeper.
The Educator plan ($9/mo) unlocks the standards-aligned Math library
for your grade band — extended lessons, slides, answer keys, and standards maps. Advanced adds PD contact-hour certificates for license renewal.