Return to question 5 A researcher used stepwise regression to create regression models to predict CarTheft (thefts per 1,000) using four predictors: Income (per capita income), Unem (unemployment percent), Pupil/Tea (pupil-to-teacher ratio), and Divorce (divorces per 1,000 population) for the 50 U.S. states. Regression Analysis-Stepwise Selection (best model of each size) 50 observations CarTheft is the dependent variable p-values for the coefficients Nvar Income Unem Pupil/Tea Divorce Standard error Adj R2 1 0. 0004 167 . 482 0. 218 0. 234 0. 0018 0.0000 152. 362 0. 353 0. 379 AWN 0. 0013 0. 0157 0. 0001 144. 451 0. 418 0. 454 0. 0007 0. 0323 0. 0003 0. 1987 143. 362 0. 427 0. 474 (a) Which model (Nvar 1, 2, 3, or 4) best balances fit and parsimony? The variable model best balances fit and parsimony. w