TL:DR: Sugary drinks might up your risk for oral cavity cancer (so says relative risk) but probably won't be the thing that kills you (so says absolute risk).  In depth: I love teaching applied statistics, including showing my students how to identify and properly attention-grabbing examples of relative risk ( 1 , 2 ). HOWEVER, relative and absolute risk aren't lying. But they can scare people, so I think it is important to share both calmly.  This example from JAMA Otolaryngology is a good example of how to responsibly share relative and absolute risk. It has a very calm, non-click bait article title:  https://jamanetwork.com/journals/jamaotolaryngology/fullarticle/2831121 Cool. Also, thanks for using female research participants. Next, the results are described in a non-salacious manner, with the absolute risk in red and relative risk in blue. How could you use this in class? Sharing relative risk isn't in and of itself unethical. Using it to scare people is questi...

I came across this article on social media: https://www.pimlicojournal.co.uk/p/mps-are-almost-certainly-using-chatgpt This got my attention, because I'm sick of people ragging on college students using AI. EVERYONE is using AI. That doesn't mean it is always OK or evil, but let's stop ragging on the kids. Anyway, the author used data to make their claims via z scores: https://www.pimlicojournal.co.uk/p/mps-are-almost-certainly-using-chatgpt Ways to use in class: 1. I like to talk to students about data as evidence. In science, it can be evidence to reject or not reject a hypothesis. In real life, it can track trends, both innocuous and suspicious. 2. This is another way of talking about z scores, a crucial but less exciting aspect of basics statistics. As best as I can tell, this was the z score formula used:  frequency z score = (number of times phrase was used in a year - mean times the phrase was used in all years)/standard deviation of number of times the phrase was...

No, that is not a dig at her romantic relationships. It is instead a question about the impact of her numerous variants on descriptive data in the music industry. And if you found your way to this blog, you know that I love a good, relevant, pop-culture-driven example for explaining statistical concepts. Especially somewhat dry concepts, like the assumption of independence when collecting data. Anyway. Per Microsoft Copilot: Across all formats, there are 34 different versions of the album: 8 vinyl 18 CD 1 cassette 7 digital Microsoft Copilot. (2025, November 8). Response to query about Taylor Swift’s album variants [AI-generated response]. Microsoft Copilot. https://www.reddit.com/r/TrueSwifties/comments/1n04kdu/the_life_of_a_showgirl_vinyl_variants_announced/ When overall album sales are counted by major industry players (Billboard, Luminate), every variant sale counts as one sale.  So, she is  selling many albums, but it raises the question of whether her sal...

Goodhart's Law is a truism in assessment circles, which are always statistics-adjacent. And that is why I'm sharing this fine embodiment of Goodhart's Law on my blog. Always pair the important stuff with something ridiculous, I swear, it makes it easier to remember the important stuff.

 There are so many ways to use this tool: Nathan Yau's Flowing Data is one of those websites I check every few days for statistical inspiration. He shares  the work of others and his own, including this  interactive bee swarm plot that illustrates salaries  for various  jobs. The bee plot, with the cursor of Psychology Teachers. https://flowingdata.com/2025/09/09/salary-and-occupation-2024/ There are many ways to use this in stats class: 1. At some point, you should talk about career exploration with your students.   2. Statistics students should be learning about modern data visualizations like this jitter plot, aka bee swarm plot.  3. If you cursor over any dot, you can see the 25th and 75th percentile scores and n size for that occupation's salary. 4. The size of each circle corresponds to the n size. Which I love because jitter plots do a great job of illustrating variability in a data set. However, each data point here represents an average...

This blog post contains one small analogy for explaining family-wise error to your students. I was making French toast for dinner the other night.  While I was measuring out cinnamon, I realized using one tablespoon instead of three teaspoons to avoid measuring errors is sort of like using a one-way ANOVA with three levels instead of doing three  t  tests to avoid Type I error.   Stick with me here. If I were to use three teaspoons to measure out an ingredient, there is a chance I could make a mistake three times. Three opportunities for air pockets. Three opportunities to not perfectly level out my ingredient. Meanwhile, if I just use one tablespoon, I will only risk the error associated with using a measuring spoon once.  Similarly, every time we use NHST, we accept 5% Type I error (well, if you are a psychologist and using the 5% gold standard, but I digress). Using three tests ( t tests) when we could use one (ANOVA) will increase the risk of a false positi...

Beth Morling is my friend and fellow author at W.W. Norton & Co. Recently, we thought of each other when a news story came out about the repatriation of the remains of 19 Black New Orleanians whose skulls were used for racist phrenology research in Germany in the 1880s. It made us think about the various forms that self-correction can take in science, as well as the importance of adhering to the scientific method.  Check it out here .