Question: Using the “test taking” data set:

Use logistic regression to determine whether you can predict success or failure on a test (“testrslt,” where 1 = success and 0 = failure) from the test-taker’s gender (“gender”), the time of day when the test is taken (“time”), the person’s level of life stress (“stress”), the person’s level of social support (“support”), and the person’s anxiety about the test (“anxiety”). Give beta coefficients for each predictor, and indicate (using all of the predictors together) what percentage of cases was correctly classified.

How many predictors would you include in an efficient model to predict test-taking success (i.e., if you were including only those predictors that provide significant independent contributions to the predictive equation)? Using only these predictors, what percentage of cases is correctly classified?