Correlation and regression
Pearson r in [−1, 1]; least-squares regression line y on x.
Y12 Stats. Calculator: scatter mode → r and regression coefficients. Use regression to predict; test correlation against tables (Y13 hypothesis test).
Worked examples
r = 1: perfect positive linear; −1 perfect negative; 0 no linear.
Regression line y on x: y = a + bx where b = Sxy/Sxx.
Predict only within the data range (interpolation).
Frequently asked questions
y on x or x on y?
y on x predicts y from x. Don't use it the other way; the slopes differ.
Correlation vs causation?
Same warning as GCSE. Correlation doesn't prove causation.