This paper investigates students' behavioral patterns within web-based multi-trial error-correction-tasks. By analyzing logfiles and considering students' initial perception of competence, we contribute to micro- as well as macro-adaption. We describe and visualize task processing data of 159 students, considering performance as well as attempts to solve a task. Taking preceeding behavior into account, it was possible to identify (maladaptive) behavioral patterns. Furthermore we compare the behavior of students with low vs. high perceptions of competence subsequent to a failure. In line with research regarding the influence of self-concept on performance and motivation, our findings suggest that students with a low perception of competence perform poorer and tend to skip trials more often after a failure, indicating motivational losses. Further research should build upon this study to enhance technology-based learning by designing learning environments and adaption strategies that take student behavior and prerequisites into account.