Statistics Foundation · Lesson 5.4
Multiple regression and confounding.
Multiple regression extends simple regression to several predictors. This lesson explains adjusted coefficients, confounding, crude versus adjusted associations, overadjustment, collider bias, multicollinearity and responsible model choice.
Lesson route
Move from one predictor to adjusted comparisons.
0–20 min
Why simple regression may mislead
Understand why a single predictor relationship may be distorted by other variables.
20–50 min
Multiple regression equation
Learn the model with more than one explanatory variable and interpret adjusted coefficients.
50–80 min
Confounding
Study how a third variable can create or distort an association between exposure and outcome.
80–110 min
Adjustment
Understand what it means to compare observations at the same value of a covariate.
110–140 min
Model choice
Explore why variables should be chosen using research design and subject knowledge, not only automatic statistics.
140–170 min
Interpretation limits
Recognise overadjustment, collider bias, multicollinearity and causal overclaiming.
Mastery checklist
Students should know what adjustment means.
Write the multiple regression model.
Interpret adjusted coefficients.
Explain confounding.
Compare crude and adjusted coefficients.
Explain overadjustment.
Recognise collider bias.
Understand multicollinearity.
Distinguish prediction from causal explanation.