Lecture 4 Flashcards
What is a dummy variable and what values do they take up?
Dummy variables are categorical variables which can only take a value of 0 or 1 e.g X1=0 if female, X1=1 if non female
Consider a regression model Y=Bo+B1X1 + e, how would you interpret Bo and B1 based on X being able to tale a value of 0 or 1
Taking the expectation given X=0, you find out the awnser is Bo which shows that:
Bo- Represents the average of Y e.g wages given X=0 e.g non female
Taking the expectation given X=1, leaves you with Bo + B1, in order to get an interpretation of just B1 we minus E(Y|X=1)-E(Y|X=0) which gives us B1
B1- represents the differences in averages between Y when X=1 and Y when X=0
What makes dummy variables special?
In order to interpret dummy variables we are not able to use differentiation, as dummy variables are categorical and not continuous variables.
This is because categorical variables have a finite number of values e.g 0,1 therefore we cannot make X really small as we do with differentiation, so we cannot differentiate.
In an additive dummy variable such as Y= Bo + B1X1 +B2X2, with X1 being the dummy variable how would we interpret the coefficient B1
We would have to take the expectation of Y|X=1 - e(Y|X=0) and then we would have to fix X2 which is a continious variable to a certian number e.g 5 for both cases.
After this you will be left with 1. Bo+5B2 2. Bo+B1+16B2, doing 2-1, gives u B1.
B1 is the average difference between the two groups
Interpret the different B by taking the expectation Y=Bo+B1X1 + B2X2 + B3X1X2 + e, for when X= O,1 and Y=0,1
- E[Y|X1=0, X2=0]=Bo
- E[Y|X1=1, X2=0]=Bo + B1
- E[Y|X1=0, X2=1]=Bo
- E[Y|X1=1, X2=1]=Bo+B1+B2+B3
- Bo- Average wage for non female non hispanic workers
- B1- Differences in average wage between non hispanic female/ male
- B2- Difference in average wage between non female hispanic and non hispanic
- B3- Differences of differences, the first difference is between male and female for hispanic and non hispanic. Now we take the differences of differences B3+B1-B1(the difference)