SAS regression analysis -- 2

PROBLEM1. There are three variables d,x and y in the data [login to view URL] note d only takes values 0 [login to view URL] to find a good regression model relating y to x and d. Please note, x and d do not interact with each other. (Actually those numbers are real data.) (1). Fit the simplest linear regression model y = β0 + β1x + β2d + ε to the given data. Write down the least squares regression equation. Is this model appropriate to the given data? What is the main issue? (2). Use the Box-Tidwell procedure to find a power transformation on the regressor x. You only need compute α1 and α2 (in most cases) to find a reasonable transformation on x. (3). Apply the power transformation you found to the data, and write a SAS program to find the coefficients of the model. Write down the model. Does the new model fit the data better? Give some reasons. (4). Initially, you may want to add a quadratic term (x2) to fix the issue. Fit a quadratic model y=β0 +β1x+β2x2 +β3d+ε to the data. Compare the model with the one in the previous question. Which one do you prefer? Give your reason