Roeder, writing for fivethirtyeight.com, has come up with a new way to illustrate the Central Limit Theorem. And it uses Plinko, the beloved The Price is Right game! http://www.businessinsider.com/price-is-right-contestant-plinko-record-2017-5 Well, a variation upon Plinko, featured on the NBC game show The Wall. Their Plinko is much larger and more dramatic and the slots at the bottom go up to $1 million. See below. http://selenahughes.com/1553-2/ How does CLT come into play? Well, the ball is randomly thrown down The Wall. And people jump around and hope for certain outcomes. But what outcome is most likely over time? For the pattern of ending positions to conform to the normal curve. Which it did. See below. https://fivethirtyeight.com/features/what-if-god-were-a-giant-game-of-plinko/ The article itself gets pretty spiritual as the author starts talking about randomness, of the show and of life. You can steal, but cite, the game show as example of CLT as he ...

Bad Math with Drawings has graced us with statistical funnies before (scroll down for the causality coefficient ). Here is another one, a quick guide pointing out how easy it is to lie with descriptive statistics. Here are two of the examples, there are plenty more at Math With Bad Drawings. https://mathwithbaddrawings.com/2016/07/13/why-not-to-trust-statistics/ https://mathwithbaddrawings.com/2016/07/13/why-not-to-trust-statistics/

In the past, I've used fake data based on real research to create stats class examples. Baby names, NICUs, and paired t-test . Pain, surgical recovery, and ANOVA . Today, I've decided to use fake data and fake research to create a real example for teaching one-sample t-test. It uses  this research report from The Onion: https://www.theonion.com/toddler-scientists-finally-determine-number-of-peas-tha-1820347088 In this press release, the baby scientists claim that the belief that a baby could only smash four peas into their ear canal were false. Based upon new research recommendations, that number has been revised to six. Which sure sounds like a one-sample t-test to me. Four is the mu assumed true based upon previous baby ear research. And the sample data had a mean of 6, and this was statistically significant. Here is some dummy data that I created that replicates these findings, when mu/test value is set to 4. : 5.00 6.00 7.00 6.00 5.00 6.00 6.00 5.0...

The sum of the parts is greater than the whole? Nathan Yau's article on creating readable, useful graphs   is a perfectly reasonable list of how to create a proper graph. The content is sound. So, that's good. However, the accompanying images and captions are hilarious. They will show your students how to make not-awful charts.