5/18 Monday - How to Think Critically About Things you See on the Internet - Part Deux - Revisited

Come to the gym and do wallballs or stay at home and do thrusters. What kind of choice is that?
Last Tuesday I wrote about the lovely predictability of probabilistic statistics to help us determine what's possible, or how to get by with a little help from mathematics. 
Sometimes, people who want to gaslight us into believing conspiracy theories use what looks like statistics to "prove" their point. They'll show graphs, charts, tables and other scientific looking things to show us the "real" truth. 
Statistics in and of themselves mean nothing without thorough analysis and exploration of confounding variables.  Two of the most common pitfalls of erroneous statistical analysis are correlative causation and confirmation bias. 
Correlation is not causation means that just because two things have a relationship it does not mean one thing causes the other. An example might be, every time you have a PR your Coach is wearing black pants. Therefore, your coach's black pants are the reason you PR. Sounds crazy right? What about when someone shows you the data?!? Here's one of my favorite examples of correlation is not causation. This graph shows a relationship between the U.S. per capita consumption of cheese and the number of people who died by becoming tangled in their sheets. 

Based on the data you could erroneously conclude that eating cheese increases your risk of death by becoming tangled in your sheets. But, you won't conclude that because there is no causal relationship. 
But, what if the data was something you thought could be "possible." What if it shows two things that you believe "could" be related. That's where we get into trouble with conformation bias. We are more likely to believe a relationship is causative if it reinforces our beliefs. 
You want to believe that a specific public health measure does or doesn't work? All someone needs to do is show you correlative data and you'll believe it because of your own personal confirmation biases. 
EVERYONE can fall prey to their confirmation biases, that's why when a scientist conducts a statistical analysis of the data they use something called a confidence interval, a mathematical projection of the probability of relationship. This confidence interval is symbolized by the letter R and it has a real value that helps guide statistical analysis. The stronger the R value the more confidence we can have in the analysis. 
There are other more robust and exciting ways to measure confidence and I don't want to get into the statistical weeds in my blog about CrossFit:) When all this craziness dies down and you want to talk about ANOVA and ANCOVA over whiskeys at Bozeman Spirits I will so happily meet you there. 
For now, I encourage everyone to think in terms of probability not possibility and to have confidence (or not) in scientific analysis rather than belief. Try to recognize what your personal confirmation biases (hint, they are deeply rooted in your beliefs) and see if you can look at things through an objective lens. Good luck! 

Click Here for The WOD