Automated assessment in CS1

摘要

A system has been developed for providing automated assessment in CS1. During the academic year 2004- 2005 this system was evaluated empirically by examining a sample group of students using both the traditional assessment methods and also the automated techniques, four times during the year. A significant correlation was found between the performance in both tests, however the correlation was only strong for students who performed well during the year.To further this study, students were interviewed and asked their opinion on the generated questions. The students offered reasons for the variation in their performance and provided an insight into where the discrepancies lie. We discovered that weaker students were employing rote-learning and using it to score marks in the class exams.As this survey was conducted on paper, a large amount of student roughwork ("doodles") was collected, the analysis of this roughwork is also discussed.