Explain why a transformation can make a non-linear association suitable for linear modelling.
Key knowledgeExplain why a transformation can make a non-linear association suitable for linear modelling.
The study design publishes this in the combined Units 3–4 list; the stated content applies to Unit 3.
Build the idea
Residuals are observed values minus fitted values. A systematic curve in a residual plot suggests the straight-line form is unsuitable, even if correlation appears strong. Transforming one variable may straighten the relationship. Fit again on the transformed coordinates, inspect the new residuals and translate predictions back to the original variables and units.
Try the sample task
Data are x=1,2,3,4 and y=3,9,19,33. A least-squares fit in x is y-hat=10x−9. Calculate its residuals and explain their pattern.
The full worksheet includes the worked solution and remaining task set.
What this activity covers
This focused activity develops the stated idea through one context. It is practice, not evidence that every application or subclause of a broad study-design bullet has been mastered.