Measured and perceived effects of computerized scientist mentors on student learning and motivation in science

Description

Unease about declining U.S. science literacy and inquiry skills drives much innovation in science education, including the quest for authentic science experiences for students. One response is student-scientist partnerships (SSP), involving small numbers of students in scientific investigations with scientist mentors. Alternatively, science inquiry programs provide large numbers of students with opportunities to pursue their own investigations but without extensive access to experts, potentially limiting the possible cognitive and affective gains. This mixed methods study investigates whether it is possible to replicate some of SSPs' benefits on a larger scale through use of a computerized agent designed as a "virtual" scientist mentor. Middle school students ( N =532) were randomly assigned to two versions of an agent (or to a control group) providing either content-only or content and interpersonal mentoring while they participated in a three-week curriculum. Results indicate that, on average, students gained in content knowledge but there was no statistically significant difference between the three conditions. In terms of motivation, students exhibited no change, on average, with no statistically significant difference between the three conditions. These data indicate that the treatment conditions neither facilitate nor inhibit student learning and motivation. Interviews with a subsample ( n =70), however, suggest that students believe the agents facilitated their learning, eased the workload, provided a trusted source of information, and were enjoyable to use. Teachers reported that the agents provided alternative views of scientists and science, generated class discussion, and met the needs of high and low-achieving students. This difference between measured and perceived benefits may result from measures that were not sufficiently sensitive to capture differences. Alternatively, a more sophisticated agent might better replicate mentoring functions known to produce cognitive and affective gains. Even without established learning or motivational gains, practitioners may want to employ agents for their ability to provide reliable information, expanded perspectives on science and scientists, and a non-intimidating setting for students to ask questions. For computerized agent researchers, this study provides a first step in exploring the affordances and challenges of sustained use of agents in real school settings with the goal of improving science education.

Subjects