The R Book, 3rd Edition

10Regression

Regression models provide a flexible tool to understand the drivers behind an outcome, response, or phenomenon of interest. These models can be used to complete a variety of tasks, including estimating the collective relationship between variables, hypothesis testing, and prediction of future outcome values.

Perhaps the simplest form of a regression is the simple linear regression model, which quantifies the linear relationship between a covariate (otherwise known as a predictor), , and an outcome (otherwise known as response) of interest, . This model assumes a linear – or straight line – relationship between and , of the form

(10.1)

Box 10.1 Terminology

In other texts, the covariate may be referred to as an ‘independent variable’ and the outcome as a ‘dependent variable’, but this is misleading especially in terms of our covariate: the term ‘independent’ has a special meaning in statistics (see Section 8.1.3), and the covariate here is not necessarily ‘independent’ ...