Statistics / regression · BEGINNER LESSON
Regression coefficient
A regression coefficient is the estimated number in a regression equation that shows how the dependent variable changes, on average, when one explanatory variable changes while the others are held constant. It is a sample-based estimate of a population parameter, so it depends on the data and the model used.
What it is
A regression coefficient is one of the numbers in a regression model that describes the relationship between an outcome and one or more predictors. The outcome is the dependent variable, and the predictors are independent variables, also called regressors or explanatory variables. In multiple regression, each coefficient summarizes the average contribution of one predictor after adjusting for the others.
[1] 6. The conundrum of regression, correlation, association, and ... - PMC [2] Reference Guide on Multiple Regression and Advanced ...
How to read it
If a coefficient is positive, the outcome tends to increase as that predictor increases; if it is negative, the outcome tends to decrease. In a simple linear model, the coefficient is the slope of the line. In multiple regression, it is the adjusted slope: the change in the outcome for a one-unit increase in one predictor, while the other predictors are kept fixed. The intercept is different: it is the predicted average value of the outcome when the predictors equal zero.
What affects its size
The numerical value of a regression coefficient depends on the measurement units. If age is measured in months instead of years, its coefficient changes because one year equals 12 months. For comparing predictors with different units, standardized coefficients are used; these are computed after variables are rescaled to mean 0 and standard deviation 1, making them comparable across variables.
[1] 6. The conundrum of regression, correlation, association, and ... - PMC
Mechanism and interpretation limits
A coefficient is estimated from a sample using regression methods such as least squares. The estimate comes with uncertainty, and statistical tests such as the t-test are used for individual coefficients, while an F-test can assess the overall regression. Coefficients describe average associations in the fitted model, not individual outcomes. Their precision depends on sample size, how well the model explains the outcome, and how independently predictors vary. Perfect collinearity means the predictors cannot be separated, and strong multicollinearity can still make estimates unstable.
[2] Reference Guide on Multiple Regression and Advanced ... [3] Quantile Regression in Epidemiology: Capturing ... - PMC - NIH
When ordinary regression may miss patterns
Ordinary regression focuses on the mean of the outcome. That can hide differences across the outcome distribution when data are skewed or relationships differ at low, middle, and high values. Quantile regression is an alternative that estimates effects at specific percentiles, showing whether a predictor has a similar effect everywhere or a different effect in different parts of the outcome distribution.
[3] Quantile Regression in Epidemiology: Capturing ... - PMC - NIH
SEE THE IDEA
How a regression coefficient is used
An explanatory variable such as age or body mass index.
Estimated slope showing average change in the outcome for one unit of the predictor.
The dependent variable being modeled, such as systolic blood pressure.
Adjusted for other variables, units, and model assumptions.
Predictor changes are translated into an estimated adjusted slope, which is then interpreted for the outcome.
A regression coefficient connects a predictor to the average modeled change in the outcome, but its meaning depends on units and the rest of the model.
TRY IT YOURSELF
Test your understanding
Choose an answer, then check the explanation.
PUT IT TOGETHER
Match each concept to its meaning
Drag a label onto its description, or select a label and then select a description. This also works with a keyboard.
4 pairs to match
Sources and further reading
This general lesson was generated from public source material. It explains a concept; it does not interpret a private paper or provide medical advice.