Charts Make Claims

A chart is an argument, whether you meant it or not. Two lines on one pair of axes implies a relationship. A truncated axis makes a small change look enormous. A chosen date range can reverse the apparent direction of a trend.

None of these require dishonesty. They happen by default, and the defense is checking your own chart the way a skeptical reader would.

Four checks before publishing a chart.

Does the axis start at zero, and if not, is that clearly visible and justified?

Is the date range the full available range, and if not, why this one?

Are the group sizes visible somewhere?

If someone read only the chart and its title, what would they conclude, and is that what the data supports?

That last question catches more than the other three combined.