ANOVA is widely used in experimental situations in which one can control potential confounding variables by fixing their values at a constant for the entire experiment or by randomizing the confounding variables across subjects.
Problem Statement
It requires the same basic assumptions and uses the same estimation method, but linear regression focuses on the estimated coefficients of the model rather than the variation explained. The ANOVA results include a table that decomposes the variation in the dependent variable into the systematic variation explained by each X variable and the random variation left unexplained by any independent variables.