FiveThirtyEight Proves That Bob Ross Painted Trees

by Corinna Kirsch on April 16, 2014 · 0 comments Blurb

Bob Ross at work, courtesy of Mentalfloss.

Bob Ross at work, courtesy of mental_floss.

This week at FiveThirtyEight writer Walt Hickey analyzes the late painter Bob Ross’s paintings through the lens of statistical analysis.

hickey-ross-tags-1

As we can see from this bar graph, 85 percent of Bob Ross’s paintings contained at least two trees.

Other stats we learn about from Walt Hickey include:

  • 97 percent of Ross’s ocean paintings show waves.
  • 66 percent of Ross’s mountains have snow caps.
  • 44 percent of Ross’s paintings show clouds.

Nothing’s wrong with this analysis per se, but it seems like FiveThirtyEight is applying statistical tools just for the sake of showing off the the tools—even when they aren’t particularly successful at providing consequential results. In the post, Hickey ends up dismissing some of his findings, saying his clustering analysis of how often trees and mountains showed up in the same painting was “Not supremely helpful, but still quite interesting.”

After Walt Hickey finished his “not supremely helpful, but still quite interesting” analysis, he called up Annette Kowalski of Bob Ross, Inc. to learn more about Ross’s work. Just by talking to her, he found out more about Ross’s work than was revealed by his data sets. Hickey writes:

She confirmed something I had discovered in my review of hundreds of Ross’s landscapes: His work isn’t defined by what is included in his paintings, but by what’s excluded.

Kowalski went on to explain that what’s excluded—what wasn’t quantified by Hickey’s analysis—is that Ross rarely ever painted people. Even his cabins rarely included chimneys; Ross just didn’t want to show a sign of humans being there. It’s strange, then, that Ross, a painter much loved by the public, didn’t seem to care much for people. He added some existential drama to his paintings, but you wouldn’t be able to find out about that from a spreadsheet.

Previous post:

Next post: