This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Problem Set #2 Answer Key 1. a. (Score (Freedonia, 2009) – Score (Sylvania,2009)) – (Score(Freedonia,2007) – Score(Sylvania,2007) = (0.2 – 0.3) – (0 – 0.2) = 0.1; Or, you could do (Score (Freedonia, 2009) – Score (Freedonia,2007)) – (Score(Sylvania,2009) – Score(Sylvania,2007) = (0.2 – 0) – (0.3 – 0.2) = 0.1 b. Difference in difference does give you a better estimate. If you just compare the results from 2009, you may incorrectly conclude that the laptop program had no effect, or actually made kids worse off. If you just compare Freedonia to itself, you may exaggerate the effect of the laptop program (reporting double the actual increase). Taking the difference in difference explains for the improvement made in Freedonia due to the laptops while also accounting for the general trend of score improvement that could be occurring in that region. c. Score = α + β 1 *FD + β 2 *A + β 3 *FD*A + β 4 *Par_Educ + β 5 *Par_Inc + ε where FD is a binary for if they lived in Freedonia, A is a binary for 2009 (after), Par_Educ is parental education level, and Par_Inc is parent’s income level. If the laptop program works, I would expect β 3 to be positive. I expect β 2 to be positive because there seems to be a growth trend in the region. I expect β 1 to be negative because Freedonia students scored lower in both years than Sylvanian students. I expect β 4 and β 5 to both be positive, as parents who are educated and have more income tend to have more time and ability to teach their children and/or pay for a tutor to help them in school. d. We would no longer have an isolated effect where we could clearly compare children who are similar to observe what will happen to those who receive laptops. Thus, children with laptops could share with those who do not have laptops, allowing the benefit to be felt somewhat on both groups. Also, increased trade between the parents would allow for the possibility of parents trading who do not need the free laptop trading with parents who do, thus spreading the laptops around and further blurring what really is determining the observed outcomes. A variable that you could include to possibly account for this would be distance from the border. The children who live closer to the border of the two cities would be more likely to play with children from the other city, as would parent’s interaction also increase. Another possible variable you could include would be play time neighboring kids (from other city), or parental involvement in the other city. These would help account for which parents and children spend the most time with the neighboring city and therefore are the most likely to receive/spread the effects of the laptop program. e. Another situation that would invalidate our results would be if a landslide occurred and hit just e....
View
Full Document
 Winter '08
 Mcintyre,F
 Regression Analysis, Gini coefficient, Freedonia, Aunt Eppie

Click to edit the document details