Business and Economic Statistics lecture note 6.pdf

ECMT 1020 Lecture 6 Multiple Linear Regression Analysis II 2 ? Topics covered 1. Indicator/Dummy variables 2. A variable selection method 3. Hypothesis testing for a subset of variables ? References Black 15.2, 15.3 3 ? Perform multiple regression analysis with indicator/dummy variables ? Interpret the results for indicator variables ? Build a regression model from a subset of important explanatory/predictor variables using KaddStat ? Perform a hypothesis test for a subset of variables manually Learning Objectives 4 Indicator (or Dummy) Variables 5 ? In many situations, some independent variables are qualitative (or categorical). Examples: ? X = Gender ? X takes two possible values: Male (coded as 1) and Female (coded as 0) ? X = A question in a student feedback survey ? X takes five possible values: Strongly disagree (1), Disagree (2), Neutral (3), Agree (4) and Strongly agree (5) Indicator Variables 6 ? An indicator variable (or a dummy variable) is a variable that takes only two possible values, coded as 0 and 1. ? For dichotomous variables, such as gender, only ONE dummy variable is needed. ? For polychotomous variables, such as a student feedback survey question with c possible choices, c-1 dummy variables are needed. Indicator Variables 7 ? Let a categorical variable X that represents the location of a department store or a head office of an insurance company in a capital city. For example, NSW (1), WA (2), SA (3), TAS (4), NT (5), Vic (6), QLD (7), ACT (8) ? You can either use ONE independent variable X (not recommended) or SEVEN indicator variables (recommended) d1, d2, ..., d7 in the regression model. ? Here d1=1 and other di = 0 (i = 2,…,7) ? NSW ? d2 = 1 and other di = 0 (i = 1,3,…,7) ? WA ? ... ? All di = 0 ? ACT Indicator Variable