H2 Mathematics Correlation and Linear Regression is a canonical JC topic guide for Singapore-Cambridge H2 Mathematics 9758, aligned to the 2027 SEAB syllabus. It is designed to sit between the stable mathematical object in BTT’s Knowledge Warehouse and the actual H2 examination demand.
Correlation and regression are modelling tools, not automatic evidence of causation. H2 students must decide whether a linear relationship is plausible, interpret the product-moment correlation coefficient, select the appropriate regression line and judge whether a prediction is sensible.
The 2027 syllabus includes scatter diagrams, correlation coefficient, least-squares regression, interpolation/extrapolation, prediction and square/reciprocal/logarithmic transformations to achieve linearity. Hypothesis tests on correlation are excluded.
Correlation measures linear association
A value of r near +1 or −1 indicates strong linear fit; near 0 indicates weak linear association. A nonlinear relationship can still be strong while r is near zero. Always inspect the scatter plot.
Correlation does not establish causation
A third variable, selection bias or reverse causality can create correlation. In H2 contexts, interpret r as evidence about a linear model, not as proof that changing one variable causes the other to change.
Which regression line?
Use the regression of y on x to estimate y from x. Use x on y to estimate x from y. These are not generally algebraic inverses. Choose by the prediction direction stated in the problem.
Interpolation versus extrapolation
Interpolation predicts inside the observed x-range and is generally more defensible. Extrapolation extends beyond observed data and can be unreliable because the relationship may change. State this limitation when the context demands evaluation.
Transformations to linearity
H2 can use square, reciprocal or logarithmic transformations so a nonlinear model becomes linear in transformed variables. The student must interpret the transformed regression and then return to the original variables correctly.
- Failure: using r as evidence of causation.
- Failure: selecting the wrong regression line for prediction.
- Failure: extrapolating far outside the data without comment.
- Failure: transforming only one side incorrectly.
- Failure: reporting calculator coefficients with no interpretation.
H2 examination control
- State the mathematical model or condition before calculating.
- Keep exact values until the question requires approximation.
- Show enough working to expose the method; unsupported incorrect answers earn no marks.
- Use the approved graphing calculator as a tool, not as a substitute for interpretation.
- Re-read the answer in the context of the original question.
Return to JC Mathematics. Stable conceptual owner: Mathematics Knowledge Warehouse. Examination execution: Mathematics Examination Craft.
Syllabus check: SEAB H2 Mathematics 9758 for examination in 2027; checked 26 September 2026.
World Mathematics route: return to the World Mathematics Atlas to connect this JC topic with its prerequisites, international equivalents, examination routes and university Mathematics.
H2 Mathematics 9758 correlation and linear regression: use models without overstating them
Start with the scatter diagram
A plot reveals direction, form, outliers and whether a linear model is plausible. A correlation coefficient without the graph can hide important structure.
PMCC measures linear association
The coefficient describes strength and direction of linear association under the data model. Values near zero do not rule out a strong non-linear relationship.
Correlation does not establish causation
Confounding variables, reverse direction and common causes can produce association. State this limitation when interpreting real contexts.
Least-squares regression has direction
Regressing y on x is designed to predict y from x under the model. Reversing variables produces a different line. Choose the equation that matches the prediction task.
Interpolation is safer than extrapolation
Predictions inside the observed x-range rely less on untested behaviour. Extrapolation assumes the relationship continues beyond the data and should be treated cautiously.
Transformations can linearise relationships
Where the syllabus uses transformations, interpret the transformed variables and translate fitted parameters back to the original model.
Outliers deserve investigation
An unusual point can strongly affect correlation and regression. Do not remove it merely because the fit improves; consider measurement, context and the question’s instructions.
Technology computes; students interpret
Enter lists correctly, select the right regression and report coefficients with appropriate precision. Then explain what the model says.
Common errors
Using correlation as proof of causation, predicting x from a y-on-x line, extrapolating wildly and quoting r without discussing the scatter.
Transfer
Use unfamiliar datasets and ask whether linear regression is appropriate before calculation. Model selection is part of the mathematics.