The Analytics Power Hour #277: ANOVA? I Hardly Know Ya'! with Chelsea Parlett-Pelleriti
Aug 5, 2025
In this discussion, Chelsea Parlett-Pelleriti, a consulting statistician with a PhD in computational data science, breaks down the intricacies of ANOVA. She emphasizes its connection to linear models and clarifies common misconceptions. Topics like the importance of covariates in regression and the challenges of multiple comparisons are explored. With humor, she also touches on automating data integration, while sharing relatable parenting tales and the amusing limitations of AI. A truly enlightening and entertaining dive into statistics!
AI Snips
Chapters
Books
Transcript
Episode notes
ANOVA Is Just Linear Models
- ANOVA is fundamentally a linear model analyzing variance rather than a distinct test.
- Teaching ANOVA separately from regression causes misunderstanding and missed connections.
How ANOVA Partitions Variance
- ANOVA partitions observed variance into explained (by groups) and unexplained (random) variance.
- It tests if differences in group means are statistically significant beyond random variation.
Correct For Multiple Comparisons
- Be mindful of multiple comparisons when running many pairwise tests post-ANOVA.
- Use correction methods like Bonferroni or Tukey HSD to control family-wise error rate.
