Question

I am interested in the effect of majoring in economics on the probability that a Binghamton student will work in the financial sector during the first year after college. Define a treatment variable and an outcome variable. What do you expect is the sign of the difference in means, positive or negative? What do you think is the sign of the selection bias? Explain.

Answer 1

Here we are trying to look at the regression of whether or not a student works in the financial sector on whether or not a student majors in economics. The ouitcome variable here will be a binary variable with "1" being if the students works in the financial sector and 0 that she does not work there (F). This is the outcome variable. The treatment variable will be a dummy variable in effect that shows that whether students major in economics or not (D). Thus the model will be.

F = a + bD, where a and b are coefficients.

The sign of the difference in means will be postive as as students studying economics will generally have increased chances of working in the financial sector as compared to other students. But this will be a binary regression and so "means" will not be something we will ideally look at. The selection bias sign will also be positive as we are only focused on students studying economics. Thus this would be cherry picking.The sign of the selection bias would be positive.