A research study counted how often ChatGPT made up citations for three different categories of mental disorders (binge eating, body dysmorphic, and major depressive). They used a chi-square to determine if rates of made up citations differed by disorder (they do).  If ever there was an article that belonged on this blog, this is it. You can use it in your stats class as an example of chi-square and/or as a warning to students if you ask them to perform literature reviews for your class. The original paper, Influence of topic familiarity and prompt specificity on citation fabrication in mental health research using large language models: Experimental Study was published in December 2025, and summarized by PsyPost  shortly after publishing.  What the researchers did: What the researchers found: How to use in class: 1. This is a good chi-square results section. They shared the test value and the p value, of course, but I like how they shared the varying rates of inaccuracy...

I've been writing this blog, statistics pedagogy articles, chapters, and a whole statistics textbook for over ten years. I'm at the point where I see silly stuff on the internet, and it automatically translates to a statistics example. Like this recent Tweet from Sheetz about the Pirates/Philly series this weekend. https://x.com/sheetz/status/1923397811778785489 This is an unapologetically Western PA tweet. I will be using it as a chi-square goodness-of-fit example with my Western PA students at Gannon University this Fall. I even created a data file that mimics the findings (Methods:  n  = 380, Results: p < .001. Conclusion: Sheetz followers on Twitter love some curly fry). If you are a poor, unfortunate soul who has never enjoyed treatz from Sheetz, I feel bad for you. Look up your favorite regional brands on Twitter and translate one of their polls into a chi-square example. Or travel to your nearest Sheetz to experience some damn joy.