How Not to Be Wrong: The Power of Mathematical Thinking

by Jordan Ellenberg

That’s a weakness as well as a strength. You can do linear regression without thinking about whether the phenomenon you’re modeling is actually close to linear.