Q4

Please help me with this question, thank you

4. Suppose you're interested in knowing whether girls who attend all girls' schools are better in math than girls who attend coed {mixed gender] schools. You have a random sample of girls who graduated from a Canadian high school with information on their score on a standardized math test {score} and an indicator of whether they attended an all girlsI school {girihs}. {a} If you used an GL3 estimator to identify the effect ofgirihs on sham-e1 what other control vari- ables would you want to include in your regremion {assuming that you had an unlimited research budget to collect data}? 1|Ii'qi'l'ite down the regression equation that you would estimate. \"Why might you be 1worried that the girihs is correlated with unobserved factors that have an independent ef- feet on girls1 math ability?l Explain clearly with reference to a real-world example of this correlation? ECGNSES 4Df5 {b} If you instead decided to use an IV estimator and could collect any data, what do you think might be a good instrumental variable to identiiij.r the true causal effect of girihs in your equation in part {a}? Explain clearly why you believe your instrument will satisfy both of the two key iden- tifying assumptions of the IV estimator. {c} If you ran the regression with girihs as the dependent variable and your instrument and control variables as independent variables. What sign {negative or positive} would you expect the coef- cient on girihs to have?II Suppose that you got the opposite sign and the coefcient was statistically significant at the 1% level. Would you be more comfortable believing your I'v' estimate from part {b} or your {)LS estimate from part {a}. Justify your