This preview shows pages 1–4. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: 5/24/2010 1 Binary Dependent Variables  I So far, our dependent variables of interest have always been continuous variables, e.g. average test scores, sales, wages, etc. What happens when the dependent variable we are interested in, Y i , is a binary (dummy) variable (i.e. takes on only values 0 or 1)? Binary Dependent Variables  II Examples: 1) Studying what determines the decision of whether or not to attend college. Y i = 1 if individual i attends college Y i = 0 if individual i doesnt attend college 2) Studying what determines whether a website visitor purchases anything. Y i = 1 if individual i purchases something Y i = 0 if individual i doesnt purchase anything 3) Studying what determines if a mortgage loan application is denied. Y i = 1 if individual i s application is denied Y i = 0 if individual i s application is accepted 5/24/2010 2 Binary Dependent Variables  III First, lets think about the expectation of a binary variable. By definition of an expectation: So the expectation of a binary variable is equal to the probability that the variable equals 1. e.g. Y i =1 if the website visitor purchases the product; Y i =0 if the visitor does not purchase the product. Then the expected value of Y i is equal to the probability that the visitor purchases the product. The above result also holds conditional on any value of X i , , i.e. given that X i equals some particular value: [ ] ( 29 ( 29 ( 29 = Pr 0 *0 Pr 1 *1 = Pr 1 i i i i E Y Y Y Y = + = = [ ] ( 29  = Pr 1 i i i i E Y X Y X = Binary Dependent Variables  IV We will study 3 different regression models that can be used when Y i is a binary variable : 1) Linear Probability Model 2) Probit Model 3) Logit Model 5/24/2010 3 Linear Probability Model  I The Linear Probability Model (LPM) is the simplest of the above models. To preview, it essentially involves doing exactly what we did when Y i was not discrete (i.e. using the linear regression techniques that we have already learned.) Model: Assume A1b, A2, A3, A4. Note: In models with a binary dependent variable, we will always assume A1b rather than A1. This will allow us to make more concrete statements about the estimated model....
View Full
Document
 Spring '07
 SandraBlack
 Econometrics

Click to edit the document details