Question 3 a) Using a hypothetical example explain the difference between homoskedasticity and heteroskedasticity. How should we deal with heteroskedasticity in OLS estimation? (4 marks) b) A researcher is interested to study how computer use affects the test scores. Using data on school districts, she regresses district average test scores on the number of computers per student. Will the slope coefficient estimated in the regression be an unbiased estimator of the effect on test scores of increasing the number of computers per student? If you think the estimated slope is biased, then what is the direction of the bias-discuss why. (6 marks)