• University of Maryland

• BMGT 431

 Problem Description The SAT has become an important component of college admission in the United States. The data set is given in “Kaplandata”. Use logistic regression to test whether Kaplan Coaching is associated with an increased probability of improved SAT scores after controlling for other explanatory variables. The attached “Data Dictionary” explains how variables in your dataset are defined. 1. What percentage of students did not improve their SAT score on their second exam? According to the data, approximately 26% of students did not improve their SAT score on the second exam. 2. Write the “theoretical” or population-level model that should be estimated. You should try to use all possible variables. P(Improvement = 1 | (X1,….Xn)) = o + B1(coaching_kaplan) + B2(coaching_other) + B3(hs_prep) + B4(hs_voc) + B5(hs_other) + B6(Male) + B7(Income) + B8(HS.Type) + B9(Rank) + B10(Math1) +B11(Verb1) (Coaching_no) and (hs_general) were removed 3. Estimate the logistic regression model and write it below. Log ( P( x ) 1−p ( x) ) = 1.203 + 1.826(coaching_kaplan) + .2004(coaching_other) + . 2836(hs_prep) - .2239(hs_voc) + 12.88(hs_other) + .05977(Male) - . 01597(Income) -