A mismatched chart forces readers to compute the answer mentally

What it is

The hardest test of whether a chart fits: how many steps stand between the reader and the answer. A matched chart makes the answer directly visible ("B exceeds A" read at a glance); a mismatched chart buries it inside differences demanding mental arithmetic — per-series trends wanted but a stacked chart supplied, so the reader subtracts segments in their head; magnitudes of two lines wanted but dual axes supplied, so the reader rescales first; shares of two groups wanted but a pie supplied, so the reader takes the difference of two angles. Every extra mental step loses readers and adds error: mismatch is not a stylistic blemish but a structural fault that bills the reader for computation.

Why it happens

A chart's purpose is to externalize reasoning into perception; a mismatched chart confiscates that gain — the data is on the page, but the quantity the question needs is not in the encoding, so readers must rebuild the encoding in their heads (subtract, convert, estimate area differences). Multi-step mental arithmetic in working memory has limited precision and shorter patience: after two or three steps errors accumulate, and readers either fall back to vague impressions or misread outright. Mismatch often adds illusion amplification: in stacked charts the top series appears to wobble with the series beneath it (visually moving while numerically still), and by the time mental checking begins the surface impression has already misled.

Where it stops holding

"Forces mental arithmetic" targets the question the chart claims to serve: a chart may legitimately decline to answer certain questions (a total-trend chart need not serve point lookups) — that is division of labour; mismatch means the chart's own stated question is not directly encoded. Some arithmetic is unavoidable (complex relations in real data), and the chart's duty is to minimize the inevitable part. Compound questions (totals and composition) may take composite charts or chart sets, but each sub-question still needs its own direct encoding.

Applying it

• For every question, write the "answer sentence" (comparison direction and magnitude) and check that every quantity in it is individually encoded; a quantity that appears only after subtraction signals mismatch.
• Keep a mismatch table: per-series trends ↔ stacked areas (switch to facets); cross-axis comparison ↔ dual axes (switch to shared axes or indices); share differences ↔ pies (switch to bars).
• Verification: have readers answer the target question from the chart while thinking aloud; any "let me subtract" or "convert that" step means mismatch — rework until the answer is visible.

Related

• Same group: U2.17.2 One question, several viable charts — they differ in reading cost · U2.17.3 Automatic recommendation reads data types, not analytical intent · U2.17.4 Choosing the chart first manufactures questions to fit it · U2.17.5 As data magnitude shifts, a once-fitting chart stops working
• Nearby: U2.01.1 Comparison, trend, distribution, composition, and relation each map to different charts · U2.04.3 Stacking suits totals, not per-series trends
• Search terms: chart-task mismatch · mental arithmetic · external cognition · reading cost