This study applied nonparametric bootstrapping to test null hypotheses for selected statistics (KR-20, difficulty, and discrimination) derived from a student-made test. The test, administered to 21 students enrolled in a graduate-level educational assessment class, contained 42 items, 33 of which were analyzed. Random permutations of the data yielded a bootstrapped mean KR-20 equal to 0.733 (p=0.012), a bootstrapped mean level of difficulty equal to 0.769 (p=0.212), and a bootstrapped mean point-biserial correlation equal to 0.302 (p=0.273). The bootstrapped KR-20 was unusual given random permutations of the data. Results failed to reject the other two null hypotheses, suggesting each model was not unusual given random permutations of the data. (Contains 3 figures and 14 references.) (Author/SLD)