Rosé, C. P., VanLehn, K. & NLT Group. (2003). Is human tutoring always more effective than reading?: Implications for tutorial dialogue systems. In Proceedings of AIED Workshop on Tutorial Dialogue Systems: With a View Towards the Classroom: Vol. VI.

Some previous studies of student learning have demonstrated a strong advantage in favor of human tutoring over a classroom control condition (Bloom, 1984; Cohen et al., 1982). These results have spawned an optimistic view towards building effective tutorial dialogue systems. Towards this end, many current tutorial dialogue systems have been evaluated successfully with students [21, 15, 1, 11, 8]. Nevertheless, so far none have demonstrated conclusively that tutorial dialogue systems provide a more effective or efficient means of instruction than on otherwise equivalent purely text based approach. In this paper we explore the question of whether it is even true that human tutoring is always superior to a reading control.

In recent years, much attention has been given to questions about tutorial dialogue, in particular about what makes it effective an in which contexts. The current study was motivated by the hypothesis that dialogue may be a more appropriate means of instruction for naive learners than for review learners In this study review learners are those who have been exposed to the material in a formal classroom setting but have not yet mastered the material. Our study focuses on learning conceptual physics. We used two different populations of students. In particular, the first population of students were those who were in the middle of taking or had already taken a semester of college level physics. Thus, for these students the topics covered in the study were a review of what they had already learned in class but did not yet master (as indicated by pretest scores). The second population of students were those who had never taken any college level physics. While students in both conditions demonstrated a significant difference in performance between the pre-test and post-test, there was no significant difference between conditions with the first population, whereas there was a significant difference to gain between the human tutoring condition and the reading control with the second population. This interaction supports the experimental hypothesis and highlights the benefit of adaptation to student knowledge state that dialogue affords. These results also have methodological implications for tutoring dialogue research.

 

Full PDF (112kB)