Naïve Statisticians

Most people without statistical education nevertheless have an intrinsic instinct for visual explanations. It works on at least two levels: They can usually tell whether a graph is well made, i.e. aesthetically pleasant, and they have a rough impression of what it wants to say. In reference to a great psychologist (Fritz Haider) with an absoluy uneventful life, I like to call them “Naïve Statisticians”.

Heider’s attribution theory proposed that the real workings of human mind and personality are of little interest. In their place, the crude models that ordinary people form about their likes offer powerful explanations of human behavior.

What if we take this perspective into statistics?

Cern 250

Through the eyes of a naïve statistician, the most complex, the most dense types of data display suddenly look the same, be they a Delaunay, or maybe flying paths in a supercollider (fig.1). The first and most important decision in any graphing, plotting and charting must therefore be to decide on the audience. Skilled thinkers need margins of error, small multiples (probably one of the next visual explanation of the week candidates), confidence intervals, many dimensions, absolute numbers. Naïve statisticians need few dimensions, zero-ed scales, comparable measurements and legends.

The need that unites them, the least common denominator, is what made Edward Tufte so famous and his books so good: Aesthetics. Keep an eye on the blank graph for reviews of tools that help.

Recommended reading:

“The Life of a Psychologist: An Autobiography” (Fritz Heider)

“The Chicago Guide to Writing about Numbers (Chicago Guides to Writing, Editing, and Publishing)” (Jane E. Miller)

“Calculated Risks: How to Know When Numbers Deceive You” (Gerd Gigerenzer)

“The Probabilistic Revolution, Volume 2 : Ideas in the Sciences (Bradford Books)” (The MIT Press)


About this entry