A diagnostic plot evaluates whether a fitted statistical model satisfies important assumptions. For a one-way ANOVA, Residuals vs. Fitted and Normal QQ plots provide focused checks for equal variances and residual normality. section 14.5.2
Read the plots as evidence about model fit:
- Residuals represent each team’s difference from the mean standardized wins for its scoring category; positive values exceed that category mean, while negative values fall below it. section 14.5.2
- Residuals should be similarly dispersed across fitted-value levels when the equal-variances assumption is reasonable. section 14.5.2
- The Normal QQ plot supplies a visual clue about whether residuals follow a normal pattern. section 14.5.2
NBA ANOVA check
After fitting an ANOVA for standardized wins by offensive-points category, use plot(fit1, which = c(2, 1)) to display only the Normal QQ and Residuals vs. Fitted plots. Unequal residual spread at the highest fitted values suggests that equal variances may be violated, though not significantly. section 14.5.2
