Data 4 — The Map Is a Chart
Rung 4 · where it is · one to two periods · the lead-in to every GIS rung
Uses library/investigations/_protocol/evaluating-map-data.pptx, 11 slides, and supplies the prompt that record says is missing. That record's own instruction is the reason this lesson exists:
"A student who has not been taught that a shaded state can mislead will read those maps credulously. Run this deck first."
The opener
Slide 4: "Danger Presented by Ohio."
Same job as the planet slide in Data 1, and running both in the same strand makes the method audible: this room opens with an absurd chart on purpose, and you should expect it. A student who says "this is the Ohio slide" in year 11 has learned the move.
Then the claim the lesson rests on:
A choropleth is a bar chart with the bars laid down on geography.
It inherits every failure from Data 1 and Data 3 — proxy, baseline, denominator, bin, order, encoding — and then adds four of its own, because geography brings its own lies.
The four map-specific failures
Every one of them is on a slide in the deck. Teach the failure, then find it.
1. Plurality read as majority. Most Common Self-Identified Ancestry. A state is coloured for the largest group. The largest group can be twelve percent of the state. The map says "most common" and the reader hears "most."
2. Counts where rates belong. Major Power Disruptions by State, 2017–2022. Raw counts across states with wildly different populations, land areas, and grid sizes. Per capita, or per mile of line, is a different map. Ask: which state changes rank the most if you divide?
3. The unit you drew hides the variation inside it. Alcohol Consumption, Credit Card Delinquency. State-level shading averages away everything that varies within a state — and within a state is where the variation actually is.
The boundary you chose decides the answer. Same data, drawn by county instead of by state, can reverse the pattern. This has a name in the field, and seniors should have it: the modifiable areal unit problem. Nobody chose those boundaries for your question. They were drawn for governing, and you inherited them.
4. A category with no magnitude. UN Vote: The right to food. Yes, no, abstain. There is no "how much." A categorical map carries no size at all, and readers supply one anyway because the colours look like a scale.
And the one that is not on any slide but is on every map:
Area is not population. Big empty places shout and dense small places whisper. Every national map in your life over-represents the places with the fewest people in them.
The exercise — redraw one
Each table takes one map from the deck and answers:
1. What does this map appear to say? 2. Which of the four failures is present? Name it. 3. What would you change — the denominator, the unit, the bins, the encoding? 4. What would your redrawn map stop being able to say?
Question 4 is the one that matters and it is the one students skip. A fix is a trade. Going from state to county recovers the internal variation and loses the ability to make a state-level claim at all — which may have been the point of the map.
Then point it at the year's rung
This lesson is the instrument. The investigation is where it gets pointed. Name the rung out loud, and name last year's, because the ladder only works if the student hears it:
| Year | Lead-in | Rung | What the student is looking at |
|---|---|---|---|
| 9 · ECS | this lesson | SF — human waste | what's there — a thing you can photograph |
| 10 · IS1 | MAPS1 = Data 2 + this lesson | NYC — rats | what's reported — a proxy for a thing |
| 11 · IS2 | MAPS2 = Data 5 + this lesson | Los Angeles | what's related — the pattern, not the point |
| 12 · IS3 | Data 6 + this lesson | site selection | what isn't there — a thing that does not exist |
The sentence _planning/the-gis-stack.md asks for, said at the top of year 10:
"Last year you mapped something you could photograph. This year you cannot photograph a rat report."
This answers an open question in the stack
_planning/the-gis-stack.md records that the investigations are preceded by MAPS blocks — MAPS1 into rats, MAPS2 into Los Angeles — and then asks:
"Open: does ECS have a lead-in before SF MAP, and does IS3 have one before the sushi placement? If the pair structure holds across all four years, two more blocks exist that this inventory has not captured."
Proposed answer: it is this lesson, run in every year, and the deck already existed for it. The pair structure holds because the instrument is the same instrument at every rung — what changes is what it is aimed at and what the student already knows about proxies by then. That does not require two uncaptured blocks; it requires this one being run four times.
This is a recommendation, not a recovered fact. If ECS and IS3 have their own lead-in material in Classroom, it is the one that should run — six claimed gaps in this corpus have dissolved on a fuller dump, and this would be the seventh.
Assessment
Critical Analysis rubric.
- SKEPTICAL — a shaded map is the most trusted and least interrogated object in the data world
- WIDE ANGLE — question 4 is explicitly the larger picture: what the fix costs
- STRUCTURED — the four named failures are the organisation
- CONTRASTING — the redraw is a comparison against the original
Open
- "I" in the deck title implies a part II. Not supplied. Worth finding; it may be where the correlation work lives.
- If this runs four times, three of the four runs need different maps or students will recognise the deck rather than the failure. Sourcing a second and third map set is outstanding work —
_planning/source-list.mdhas the sources. - Whether ECS should get all four failures or only the first two is a live question. The modifiable areal unit problem is a senior idea; plurality-vs-majority is not.