We describe recent progress developing and evaluating step-based tutoring software for introductory linear circuit analysis courses. The system is based on a novel concept of automated problem generation, where the entire circuit or other problem to be solved is generated randomly (both topology or structure and numerical values), providing an unlimited supply of problems of gradually graded difficulty with corresponding isomorphic, fully worked examples. Further, students are never penalized when they give up on a problem, but are instead presented with a fully worked and explained solution, followed by a completely different problem of the same type. This approach boosts student confidence and adapts to the differing needs of individual students. Recent progress includes the release of several important new tutorial modules, including waveform sketching and Laplace transforms (for a total of 17 released modules). The former involves the sketching of waveforms as a function of time in an interactive web-based interface, where students are required to find the voltage across an inductor when given its current (by differentiating), and other similar problems. The latter involves computation of direct and inverse Laplace transforms from randomly generated functions of various types, where students use an interactive template-based interface to enter their equations for checking. Other important capabilities have been added this year, such as voltage and current division equations, generation of circuit solutions using superposition, and generation of transient circuit problems involving switches. The waveform sketcher is further being adapted to permit sketching of Bode plots from system response functions, and vice versa.
A previously reported laboratory evaluation showed a large, statistically significant (1.21 standard deviation) advantage for an earlier version of our system over traditional paper homework. In Fall 2015, we conducted a large in-class evaluation with ~70 students to compare our Circuit Tutor system to a widely used commercial publisher-based system. Students were randomly assigned to either use one system for node analysis homework, and the other system for mesh analysis homework, or vice versa. An in-class quiz then compared student performance. We further surveyed students on their preferences. We found a large, statistically significant [t(64) = 3.09, p < 0.05] advantage for Circuit Tutor on node analysis of 0.72 standard deviations (average score of 72% for students who used Circuit Tutor, compared to 49% for those who used the publisher system). For mesh analysis, the Circuit Tutor average was 71% vs. 65% for the publisher system, but the difference was not statistically significant [t(64) = 0.88, p = 0.38]. The larger advantage of our system for node analysis may be due to the fundamentally easier nature of mesh analysis. In the survey, 86% preferred Circuit Tutor and 9% preferred the publisher system, and 94% felt that Circuit Tutor more effectively taught them the topic for which they used it, and 3% felt that the publisher system was more effective. The Circuit Tutor system has now been used by over 2300 students in 54 class sections at eight different colleges and universities, with generally very favorable ratings.
Are you a researcher? Would you like to cite this paper? Visit the ASEE document repository at peer.asee.org for more tools and easy citations.