Behind The Curtain With Bayesian Statistics

What are statistics? Why do we need them?

- By Dani Glidewell

A prehistoric human crouches on the African savannah. Something moves in the corner of her eye. A lion? She freezes and readies her spear. But the breeze blows, the grass shifts, and she sees the “lion” was only a trick of the light.

Humans see patterns and connections everywhere. This makes sense when you realize that for most of our history, a person who thought they saw a lion got startled, while a person who didn’t see a lion g+ot eaten. Better to err on the side of caution xand interpret every unusual movement as a lion, even if most of the time no lion is really there. So how can we find out if a pattern or connection is meaningful or not? To figure that out, people invented statistics. Like stone spears and modern computers, statistics are tools. They let us see the mathematical patterns in systems too enormous and complex to understand without help, like Earth-spanning weather systems or cities full of people. We can use those patterns to help predict the future.

What Are Bayesian Statistics?

“The success of statistics is obvious, especially in scientific research,” says Wayne Stewart, a statistics professor at the University of Oklahoma. He uses numbers and computer code to teach people how to detect complex patterns in data, even when the data are messy or complicated.

There are two statistical frameworks, classical and Bayesian. According to Stewart, classical statistics focuses on whatever experiment you’re doing at the moment, so it avoids biases from previous work. Bayesian statistics deliberately uses information from previous work so scientists can get as much information as possible. Both can be useful.