11 years ago
Using Statistics and Visualization as Complementary Validation

 I don’t often find myself thinking or saying common maxims (e.g. Better to be safe than sorry). But there is an uncommon maxim I learned from Jeffrey Luftig, who taught one of my favorite statistics courses. He had a number of colorful sayings but the one that has stuck with me in my professional life is Never show a statistic without a picture; never show a picture without a statistic. I think about this often while validating analytical work. Anscombe’s quartet has become a well-known example of where statistics alone can be misleading. All four data sets below