Why is regression analysis also statistical test?

Question

According to wikipedia, regression analysis is a statistical process for estimating the relationships among variables. Regression analysis is widely used for prediction and forecasting. So why is regression analysis also used as statistical test? For example, in this page, logistic regression and linear regression are listed among t-test, ANOVA, chi-square test etc.

Answer 1

On my viewpoint, regression analysis is a mathematical tool which can be used in various circonstances and for various kind of problems. Of course, in many statistical problems a regression technic is currently used. But they are also problems involving regressions where nothing is randomly specified and where no statistics are needed. A mathematical tool on one hand and mathematical or physical problem on the other hand, are two different thinks.

Answer 2

Regression analysis is surely an important tool for prediction and forecasting, but it is also commonly used as a statistical test for many purposes, e.g. to investigate whether a given variable is associated with another one independently of confounders, or whether a given multivariable model significantly predicts a given variable, and so on. In all these applications, regression provides a number of parameters, including measures of the strength of these associations and corresponding p values. In this view, it is one of the most important statistical tests used in many fields of sciences.