ABSTRACT
When I fi rst began talking about regression, I said that one of the names for regular regression is linear regression. All the regression models we’ve examined so far have made a very large assumption: that the relationship between our variables is linear. For example, a relationship we’ve examined again and again has been that between education and income. Here is a model using education and hours worked as the independent variables and respondent’s income as the dependent variable (using 2012 GSS data):
If we were to hold hours worked constant at 40 hours, and calculate predicted incomes for various years of education, we would get the following graph:
■ Exhibit 15.1 : Linear Model Explaining Income
Dependent Variable: Respondent’s Income in 1000’s
Independent Variable Slope
Education (in years) 4.87*** Hours worked per week 0 .73*** Constant –54.63 R 2 0 .29
