A computer-based, problem-solving exercise was developed, using dynamic graphics to assist beginning programming students in devising and testing solutions to novel problems (problems requiring solution methods with which the students were not already familiar). For evaluation, 103 beginning computer science students at Southeast Missouri State University, were each randomly assigned one of three forms of the exercise containing four array-manipulation problems. The control group worked all four problems without the aid of dynamic graphics; whereas, on Problems 3 and 4, the first experimental group was allowed to test possible solutions by observing graphical execution of attempted code, and the second experimental group was allowed to interactively manipulate the elements of the array by means of arrow keys in order to devise a solution, in addition to observing the graphical execution of their code. Six measures of improvement in performance on Problem 4 over Problem 2 were made for each subject. The one-tailed Wilcoxon two-sample test was applied, using significance level .05, to compare mean scores between groups. The second experimental group showed significantly more improvement than the control group as measured by four of the six scores, and the first experimental group showed significantly more improvement than the control group as measured by two of the six scores. Although the second experimental group scored slightly higher than the first on all but one measure, no significant difference was observable between the two experimental groups. On a follow-up quiz, each experimental group scored slightly, but not significantly, higher than the control group. Additionally, it was noted that only students who had made a score of 13 or more on the KSW Computer Science Placement Exam were able to work either of the two problems used for evaluation.