Add discussion of relationship of common statistical tests to linear models

[zaneveld]

Use cartoons to show high and low R2, multiple predictor variables, and interaction effects.

Use equations similar to https://lindeloev.github.io/tests-as-linear/#:~:text=Most%20of%20the%20common%20statistical,most%20students%20know%20from%20highschool. to show mathematical relationships

[zaneveld]

This discussion of studentized residuals, the difference between the true error model and the observations in the residuals is likely to be useful: https://en.wikipedia.org/wiki/Studentized_residual#:~:text=In%20statistics%2C%20a%20studentized%20residual,in%20the%20detection%20of%20outliers.