Pedagogical methods for graduate-level statistics courses have rarely focused on the pursuit of conditional knowledge or the ability to choose which concepts/procedures are relevant given a specific research situation. However, utilization of an innovative approach called decision-based learning (DBL) not only provides students with the conceptual, declarative, and procedural knowledge of traditional statistics courses, it also demystifies the process of gaining conditional knowledge; thus decreasing "statistics anxiety." This study examined the impact of a DBL course on students' ability to select appropriate statistical methods based on the wording of story problems, and specifically looked at pre-post differences. Participants were graduate students enrolled in an introductory statistics course who completed a combination of a pre, and post, and follow-up interviews. Interviews were coded and scored based on students' ability to correctly identify statistical methods, run and interpret statistical output. Results indicated that students' conditional knowledge increased significantly from pre- to post- to follow-up (effect sizes of 0.63 to 0.64). This compares favorably with the range of effect size increase from published studies of other innovative approaches (0.21 to 0.52). Results also showed nominal conditional knowledge decay, suggesting that DBL can be an effective and efficient means of teaching introductory graduate-level statistics. Implications for other disciplines are noted.