This research applies the roles of variables to the novice programmers in the C language programming. The results are evaluated using the Structure of Observed Learning Outcomes (SOLO) taxonomy. The participants were divided into an experimental group and a control group. The students from the control group learned programming in the traditional case-based teaching method. The students in the experimental group learned programming by applying the role-based approach to the case-based teaching method. For the purposes of determining the effects of the role-based teaching, the SOLO taxonomy was used to evaluate the results of the participants' final paper-pencil test. The SOLO level of the code reading and code writing was graded according to the SOLO categories for program comprehension and construction. Data analyses show that (a) there was a higher level of performance and cognitive ability in the experimental group, (b) there were some positive and significant correlations between student's final examination scores and their SOLO level of the code reading and code writing, and (c) the SOLO level of the code writing and code reading was significantly related. These results suggest that the role of variable is effective on improving novice programming.