Despite the study findings, the larger question (stated above) remains essentially unanswered in the literature. The authors propose a call to action by various professional groups and organizations to use rigorous and complex research efforts to seek answers to this very important question.

Medical educators should hold limited expectations of the SDLRS to measure medical students' readiness to engage in self-directed learning. The definitions and theoretical assumptions that inform readiness for self-directed learning should be reconsidered. Alternative approaches to studying self-directed learning should be explored.

Entering the 1960s, more than half of the medical doctors in the United States were family physicians, pediatricians, or general internists. Today, about one-third of all U.S. physicians are primary care practitioners. Although opinions vary on the optimal ratio of primary care to specialty physicians, in the mid-1990s, the consensus among leaders in medicine was that more primary care physicians would be needed to ensure access to quality care. The target output of graduates was set for a minimum of 50% primary care physicians, and medical school admission committees responded. The present study examines research related to career decision making in primary care medicine. We address career decision making in primary care with the expectation that selection of a medical residency is based on multiple factors, and multivariate statistical techniques comprise the most appropriate statistical procedure for developing predictive models of medical student residency choice. Additional multivariate studies for simultaneous analysis of multiple dependent and multiple independent variables are needed to determine whether health policy planners and medical schools should continue to address the distribution of primary care physicians through medical school admissions. Further to enabling prediction, researchers must renew efforts to base investigations on theoretical models, summarizing and organizing previous research, and providing one another with means to focus future studies while building on previous work.

The Bland-Meurer Model may be used for variable selection in multivariate predictive models of family medicine, internal medicine and pediatrics residency choice. The authors' results may help guide admissions decisions, while providing medical educators and health care policymakers with a clearer picture of residency choice and a better understanding of how to meet the demands of selected patient populations.

Students who fail USMLE Step 1 have lower final clerkship grades due in part to lower NBME subject exam scores. The majority of these students, however, successfully pass USMLE Step 1 prior to graduation, go on to graduate medical training, and become board certified in their specialty.