Suppose you are investigating the impact of a school lunch program for low incom
ID: 3224450 • Letter: S
Question
Suppose you are investigating the impact of a school lunch program for low income students on student performance on a standardized test, using aggregate data from 500 schools. You regress school average test score on the average pupil per teacher ratio in the school, and the average spending per student (by the school district). a) What coefficient signs do you expect on the explanatory variables? Why? b) Suppose you suspect that the average parental income in the school district is an omitted variable. Is parental income likely to be correlated with the average spending per student variable? What would be the effect of omitting parental income? c) While you do not observe parental income in your data, you do have data on the percentage of students in the school participating in a subsidized school lunch program for low income students. Is the lunch variable a valid instrument for average spending per student? How would you recommend using the lunch variable in the analysis?Explanation / Answer
In case of the first variable i.e. the average number of pupil per teacher ratio should have a negative sign on the coefficient because it is expected that smaller this ratio or in other words the larger smaller the number of students assigned to each teacher the better or more will be the score because that way the teacher could concentrate better on the students performance. However we could expect a positive coefficient on the second variable viz. average spending on a pupil because it is expected that more the resources the school agrees to spend behind a student the better should be his score.
Yes the average parental income could be correlated with the average spending per student variable because the higher the parental income the more the school can seek from them and in turn spend more at the same time, however if the parental income is less the school fund would also fall. Omitting this variable should devoid the model of an useful covariate that can alter the results of the outcome variable.
The lunch variable is a valid instrument in deciding over the spending of the school because it is expected that students joining the program belong to a lower economic ranks of the society and thus give the school an idea of exactly howmuch extra it should spend. However, the lunch variable should be used as an interaction effect to the spending covariate, that way we get to know of its individual as well as secondary effects on the other variable.
Related Questions
drjack9650@gmail.com
Navigate
Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.