Statistics Module 15 V2 - Multiple Linear Regression, Problem 15-1Abcd

Module 15 covers the ordinary least squares method of a multiple linear regression. The relationships between the different values in a regression output are the same as simple linear regression, so the exercises in module 14 will provide practice for similar types of problem using multiple regression analysis. For example, the relationship between an estimated coefficient, its standard error and the t-statistics are the same in a multiple regression as in a simple linear regression. Calculating these values from a dataset requires matrix algebra, which exceeds the prerequisites of an undergraduate course in regression. In order to avoid the use of matrix algebra, this module focuses on the interpretation of estimated regression results, using estimated regression equations for prediction and hypothesis testing. We also introduce the use of dummy variables to allow the inclusion of categorical data into the model. Dummy variables are also used to illustrate the similarities between regression analysis and ANOVA analysis covered in module 13. Also included in this module is a discussion on the problem of multicollinearity.