normative

Use conventional pre-specified graphics to avoid cheating

The most common flaw in data presentation is multiplicity—trying many graphics and cherry-picking the best; the remedy is to use utterly conventional graphics that are the best in your field and to specify those graphics in advance, which is why randomized control trials now require pre-specifying graphics before seeing the data.

normativepending

Speaker

Edward Tufte

Evidence Quote

“They try five times, five different graphics, or maybe 10. They have a programmer it does is something special and they cherry-pick the results.”

Source

224 | Edward Tufte on Data, Design, and Truth— Sean Carroll's Mindscape
Created: 6/13/2026, 12:26:56 AM

Your notes

Loading notes...