IB Maths: Correlation & Linear Regression (AI) — FAQ

It's tested via calculator-active questions asking you to find the Pearson's correlation coefficient (r), write the regression line y on x, and interpret both in context. According to the IB Mathematics: Analysis and Approaches guide (first exams 2021, current for 2025 assessment), this sits in the SL/HL Statistics and Probability topic and appears in Paper 1 and Paper 2 for AI.

Both — it's core content in AI SL and AI HL, not an HL-only extension. The mechanics (finding r, the regression line, interpolation) are identical at both levels; HL students just meet it alongside more statistical machinery like chi-squared tests later in the same topic.

Yes, for HL students. Paper 3 is a 60-mark extended problem-solving paper with two structured questions, and regression frequently forms part of a longer real-world modelling scenario alongside other statistics content. SL students don't sit Paper 3, so they only meet it in Papers 1 and 2.

Correlation just tells you two variables move together (measured by r, between -1 and 1); causation means one variable actually causes the change in the other. IB examiners specifically reward students who flag that a strong r value alone never proves causation — a classic follow-up mark on "comment on" questions.

You almost never calculate r by hand in the exam — you enter the bivariate data into your GDC's statistics mode and read r off the regression screen. Examiners expect the correct value to at least 3 significant figures, plus a one-line comment on strength and direction.

Enter your data into the GDC's linear regression function (y = a + bx or y = mx + c) and read off the coefficients directly — you don't derive the formula from scratch in the exam. Always state the equation using the actual variable names from the question, not generic x and y.

Interpolation means predicting a y-value for an x inside your original data range — this is reliable and what the regression line is built to do. Extrapolation means predicting outside that range, which examiners flag as unreliable unless the question explicitly asks you to comment on it.

Because IB AI only requires the regression line of y on x, which minimises vertical distances between points and the line — it's built to predict y from a given x, not the reverse. Using it backwards to predict x from y gives a subtly wrong answer, since the x on y line is a different line entirely.

The three recurring ones I see marking mocks: rounding r or the regression coefficients too early and losing accuracy marks, forgetting to comment on strength/direction in words (not just stating the number), and using the regression line to extrapolate without flagging the reliability issue.

It's one of the more accessible topics in AI Statistics — the GDC does the heavy calculation, so most lost marks come from missing the written interpretation, not the maths itself. Students who practise the exact wording examiners want (strength, direction, context) tend to pick up these marks consistently.

Both AA and AI cover correlation and regression at a similar level, so this topic alone shouldn't decide the choice. The bigger difference is elsewhere: AI leans more into statistical modelling, technology-driven problem-solving and real-world applications overall, while AA is more algebraic and proof-focused.

Look for topic-specific worked examples with real GDC screenshots, past-paper-style questions on interpolation versus extrapolation, and mark schemes that show exactly how examiners phrase interpretation marks. Revision Notes and Topical Worksheets on revisionprep.com cover this exact Statistics sub-topic with worked GDC steps.