9. Misspecification of functional can lead to some problems when doing regression analysis. Identify which of the following statements below are true: (i) Misspecification of functional is a threat to internal validity. (ii) It could result in a type of omitted variable bias. (iii) Probit regression could be a solution to functional form misspecification. (iv) It is overcome by adding the squares of all explanatory variables. A) (i), (ii). B) (ii), (iii). C) (iii), (iv). D) (i), (ii), (iii). 10. When the outcome variable Y is binary A) You can still use OLS estimation if you are willing to use a linear probability model for Y. B) You can use logit of probit if you are worried about predicted probabilities going above one or below zero. C) When using logit or probit, instead of OLS one must use maximum likelihood. D) All of the above. 11. When the outcome variable Y is binary and we use a probit model for Y as a function of X1 and X2, suppose we are interested in how a change in X1 affects Pr(Y = 1) A) Themaximumlikelihood estimate of 1 which is the coefficient associated with X1 directly tells us how Pr(Y = 1) changes when X1 changes by 1 unit. B) if X1 itself is a binary dummy variable, you can compute the marginal effect of X1 at the mean of X2 or the mean (of the marginal effect of X1) across values of X2. C) The marginal effect X is not defined because the probability Pr(Y = 1) could be above 1. D) The marginal effect of X1 on Pr(Y = 1) cannot be