This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: Econ 513, USC, Fall 2005 Discrete Response Models I:Binary Response Models In the next couple of lectures we consider models where the dependent variable is dis- crete. Initially, we look at the case where the outcome is binary: yes/no, participation/no- participation, employed/unemployed. After that we will look at more complicated cases where the outcome may take on a number of values, possibly ordered (highschool dropout / highschool /college), or categorical (employed/unemployed/out-of-the-labor-force). The example I will use in this lecture is the decision to go to college. I will use the white subsample of the NLS data, 815 observations. Out of these 419 go on to college, and the remainder left school before or after getting a high school degree. We will model this as a function of three covariates, iq, father’s education and mother’s education. The simplest thing to do is to use a linear probability model: y i = x i β + ε, with the minimal assumption on ε i being that it is uncorrelated with x i . Obviously the model has to have heteroskedasticity: if the conditional expectation is E [ y | x ] = Pr( y = 1 | x ) = x β , it must be that V ( y | x ) = x β (1- x β ). That is not to big a problem in practice. We can always adjust the standard errors to take account of this. A bigger problem with thecan always adjust the standard errors to take account of this....
View Full Document
This note was uploaded on 07/22/2008 for the course ECON 513 taught by Professor Rashidian during the Fall '07 term at USC.
- Fall '07