Module 08 Summary

What this module established

Correlation measures straight-line association only. A strong U-shaped relationship correlated at 0.113, and adding one distant point to forty unrelated ones took the correlation from 0.039 to 0.85.

Carry forward

• Pearson correlates values and Spearman ranks. A large gap between them is evidence that the relationship is monotonic but curved.
• A regression slope is an association within the observed range unless the predictor was randomised. The intercept is often outside the data entirely.
• R-squared is the least informative number in the output. The residual SD is in the outcome's units and a residual pattern reveals the wrong shape.
• A prediction interval for a new observation was seven times wider than the confidence interval for the mean response at the same x.

Before moving on

Move on when you can say whether your model is for prediction or explanation, and what that choice forgives.

Mark complete