ABSTRACT
This chapter discusses the simple regression model by adding additional predictor variables to a regression model. It also discusses how to use multiple regression to discount spuriousness association as an explanation for a relationship between two variables. Confidence intervals provide more information, and it doesn't hurt to report confidence intervals whenever readers have randomly sampled from some population with the goal of making a population inference. The National Election Study regularly includes a question asking respondents how many days in the last week prior to the interview that they talked about politics with family or friends. Theories can be validated or refuted on the grounds that a relationship that the theory predicts should be nonzero is in fact nonzero. So null hypothesis testing does have an important role to play in research. The partial association measures in model 2 for political interest and demographics control only for each other, but also all of the communication variables in the model.
