Course Evaluation using Fuzzy Linguistic Rules

: Course evaluation has become a main issue in higher education because it is viewed as an important overall performance indicator of a school or university. The notion among engineering faculty that end-of-semester course evaluation (CE) is subjective and potentially a retaliatory tool in the hands of disgruntled students creates much controversy about the validity of student-centric CE. This belief can be particularly worrying to engineering faculty in a tenure track position whose merit-based performance evaluations are strongly tied to his/her educational scholarship. Limitations in current evaluation processes raise questions, such as, how is teaching effectiveness inferred from the data, and by whom? This paper asserts that student-centric CE does not provide sufficient information at the level of detail needed to assess the success of a course. Colleges that use course evaluations apply averages of a five-point Likert-type response scale for an indictor on the performance of the instruction. Likert-type response scales have crisp boundaries and thus are considered limiting. In general, many tangible and intangible factors should weigh in on performance evaluations of both the instructor's teaching quality and the student's learning accomplishments. Under such conditions, it is more difficult from simple statistical analysis to gauge effectively teaching quality and student skills learning. An alternate approach is to use linguistic assessments instead of numerical values. The fuzzy logic method is widely employed when using data of low precision and high uncertainty, including sparse information. In recent years, fuzzy logic algorithms have also been applied in machine control, grading examinations, and analyzing sustainability and development. In this study, the fuzzy logic approach is applied to course evaluation. Our model includes a two-component input dataset. Student end-of-course evaluation is counter-balanced with instructor evaluation of each student's performance. Instructor end-of-semester learning evaluation includes test scores, quizzes, student behavior and attitude, grades on laboratory exercises, and final exam grades. Fuzzy linguistic rules are developed that determine teaching quality as well as student performance. Results are compared for an introductory surveying course offered in 2007 and 2010. The model accounts for pedagogical innovation and teaching technology. The model's results suggest that the instructor on average scored a 20 percent more favorable score than using traditional evaluation from simple statistical analysis. Course evaluations using a two-component input model to fuzzy linguistic rules and variables offer more meaningful results as well as being an effective counter balance mechanism compared to traditional student-only evaluation data.