The family doctor who shrank in two dimensions

The LA Times drew the decline of California’s family doctors as a doctor pictogram scaled in height and width at once. The share fell to 44% of its 1964 level; the doctor kept about 20% of his ink. Tufte scored it a Lie Factor of 2.8.

Los Angeles Times, Aug 5, 1979 (via Edward Tufte, “The Visual Display of Quantitative Information,” p. 69), 1979Added July 9, 2026

Pictogram titled “The Shrinking Family Doctor,” redrawn after the 1979 Los Angeles Times original: three doctor figures for 1964 (27%), 1975 (16.0%) and 1990 (12.0%), each scaled in both height and width so the 1990 doctor is a tiny fraction of the first, with the three years spaced evenly despite 11- and 15-year gaps — Exhibit A — as published

The claim

The California family doctor is vanishing before your eyes — by 1990 there will be barely a sliver left of the doctor there was in 1964.

The trick

The numbers say the share of California doctors devoted solely to family practice fell from 27% in 1964 to a projected 12% by 1990 — to about 44% of where it started. The picture says something much worse, because the doctor shrinks in height and width together, and your eye reads ink, not edge lengths: scale both dimensions by 0.44 and the little doctor keeps only 0.44², about a fifth, of the original’s area. Tufte measured the printed version at a Lie Factor of 2.8 — the graphic shrinks nearly three times as fast as the data — and that was before counting the extra shrink from the overlaid perspective. A second, quieter trick rides along on the x-axis: 1964, 1975 and 1990 sit at equal spacing, squashing an 11-year gap and a 15-year gap into the same width — and that last, steepest-looking step is a projection dressed up as a measurement.

The same three values on a plain line chart with a zero-based axis and years positioned correctly: 27% in 1964 falling to 16.0% in 1975, then a dashed segment to an open marker at 12.0% in 1990, labelled as a projection made in 1979 — Exhibit B — the honest redraw

The honest version

Drawn as a plain line on a zero-based axis with the years where they actually fall, the same three numbers show a real but ordinary decline: 27%, then 16%, then a projected 12% — a fall of about half over 26 years, not a near-extinction. The pictogram isn’t the crime; the scaling is. If a figure must stand in for a value, scale its area to the value (or repeat identical icons, one per unit), keep time to scale, and draw projections so they look like guesses — an open marker and a dashed line, not another solid data point.