// module overview
Most data training starts by handing you a spreadsheet. This one starts by asking
what a number actually is — because before you can read school data well, it helps
to know that every data point is really three decisions somebody made: what to measure,
how to measure it, and what to call it.
"Absent" isn't a fact of nature. It's a coding choice. A grade is a teacher's judgment passed through a rubric passed through a gradebook rounding rule. None of that makes data untrustworthy — it makes data human, and knowing that is the first analytical skill.
No math background needed. No code required on the practitioner track. This is the module for anyone who has ever been handed a dashboard and wondered what they were actually looking at.
"Absent" isn't a fact of nature. It's a coding choice. A grade is a teacher's judgment passed through a rubric passed through a gradebook rounding rule. None of that makes data untrustworthy — it makes data human, and knowing that is the first analytical skill.
No math background needed. No code required on the practitioner track. This is the module for anyone who has ever been handed a dashboard and wondered what they were actually looking at.
// key insight
The first bias worth understanding isn't in the data. It's in where you're standing when you read it.
// the bias field guide
Seven ways data misleads well-meaning people
Each one gets a plain definition, an example from a real school building, and a pointer to exactly where you'll catch it red-handed later in the curriculum.
| Bias | What it means | You'll catch it in |
|---|---|---|
| Front-porch bias | Generalizing from your own vantage point without noticing you have one — like building a literacy program around what your own neighborhood loves. | Community Intelligence |
| Measurement bias | The recording rules distort what's recorded. Absences during Diwali, Eid, and Yom Kippur coded "unexcused" — devotion recorded as truancy. | Module 04 |
| Aggregation bias | Summaries erase the people inside them. A 92.4% average attendance rate hiding one in six students missing a day every week. | Module 03 |
| Selection bias | Who's in the data isn't who you think. Comparing classrooms by referral rate when rosters were never randomly assigned. | Staff Analytics |
| Survivorship bias | You only hear from the ones who made it through. Climate surveys given in class — the most disengaged students are, by definition, absent that day. | Module 05 |
| Confirmation bias | The label changes how you read everything after it. Once a student is "at-risk," every tardy confirms it; the same tardy from an honors student is "a rough morning." | Module 04 · 07 |
| Automation bias | "The dashboard said so" ends the conversation. Acting on a flag without investigating it — the opposite of everything this curriculum teaches. | Every module |
// the three habits
What you'll carry into every module after this one
01
Ask who decided this counts. Every number had a collector, a rule, and a reason. Find out what the rule excluded.
02
Ask to see the shape. When someone shows you an average, ask for the distribution. The average is the least interesting number in the dataset.
03
Tell three stories before acting on one. Any pattern has rival explanations. Name them out loud, then go find out which is true.
// what you'll learn
What Educators Will Learn
- What a data point actually is — the three human decisions inside every number
- Why the average is the least interesting number, and what to ask for instead
- The difference between "these move together" and "this causes that" — and why it decides which intervention gets funded
- A seven-bias field guide with examples you'll recognize from your own building
- Three habits that make every later module easier — no math, no code, no prerequisites
Python Walkthrough (short & optional)
- Loading one table and computing a mean in two lines — then plotting the distribution that contradicts it
- Recreating the 92.4% illusion: same data, average vs. histogram, side by side
- Watching a pattern reverse when you split by subgroup — Simpson's paradox with
groupby() - Writing your first "three stories" comment block: documenting rival explanations before you analyze
- No joins, no pipelines, no flags — those start in Module 02