Use a residual plot to assess a transformed model
Let x be the explanatory variable weekly training distance (km) and y the response variable fitness test total (points). Apply the square to the response variable only. Use the full fitted coefficients below. Report signed residuals to 1 decimal places; retain unrounded values until that final step
Use weekly training distance (km) to model fitness test total (points) for the displayed cases. Apply a square transformation to the response variable only
| Case | weekly training distance (km) | fitness test total (points) |
|---|---|---|
| Case 1 | 12 | 381 |
| Case 2 | 16 | 399 |
| Case 3 | 20 | 309 |
| Case 4 | 24 | 315 |
| Case 5 | 28 | 237 |
| Case 6 | 32 | 267 |
| Case 7 | 36 | 201 |
| Case 8 | 40 | 207 |
The fitted equation is written on the transformed coordinates. Residuals are calculated on the fitted line's own response scale: signed residual on the fitness test total² (points²) scale
A signed residual is observed transformed response minus predicted transformed response at the same transformed explanatory coordinate
The residual plot is shown below. A horizontal zero line is included