Question 3 [20 points] This question requires the use of gretl. The related file is Violentcrime gdt. You are not required to show script (although you can feel free to). But you are required to show the regression output that helped you to reach your conclusions. A study provided data on variables that may be related to the violent crime rate for 50 states in the United States. The dependent variable in the study (Violent) is defined as the number of murders, rapes, robberies, and aggravated assaults per 100,000 people in the population. The following independent variables were used in the study: Poverty: percent of the population with income under the poverty level Single: percent of families headed by a single parent Metro: percent in metropolitan areas . White: percent white Highschool: percent high-school graduates The data are available in the file named Violentcrime.gdt Note: The variable "State" is a count variable. Please do not include 'State' as a regressor in any of your analyses. For the five explanatory variables (Poverty, Single, Metro, White, Highschool): a. [10 points] Use the forward elimination procedure to develop the best estimated equation. Use a=0.10 in tests b. [5 points] Develop the best one-variable estimated regression equation using the R2 criterion c. [5 points] Discuss two limitations of using automatic variable screening methods (as in part a)