ASSESSING LEARNING IN INTRODUCTORY GEOSCIENCE COURSES USING NORMALIZED GAIN SCORES AND THE GEOSCIENCE CONCEPT INVENTORY

Libarkin and Anderson (2005) created the Geoscience Concept Inventory (GCI), a valid and reliable instrument for measuring student understanding of critical geological concepts. The GCI guides instructors to select a range of questions of varying difficulty that match the learning goals of a specific class. Standard concept inventories are used in many science disciplines and a normalized gain score (g, where g = post-pre/1-pre) is typically reported. Some disciplines (e.g., physics) have identified specific normalized gains as indicative of different degrees of improvement in student learning. Such values are not currently available for the geosciences.

We reanalyzed Libarkin and Anderson's original GCI data in combination with data from recent earth science and physical geology courses in an effort to better constrain the range of gains that would be representative of introductory geoscience classes. Paired data from 27 introductory classes showed normalized gain scores ranging from -0.017 to 0.33 and exhibited no clear trend when compared to environmental factors such as class size. However, when we considered only classes showing a statistically significant difference between pre and post-scores (p<0.01), the range of normalized gain scores diminished (0.1-0.25) and a trend of decreasing gain with increasing class size became apparent. A similar, though less robust, trend is seen with more widely–distributed, non-paired data. Although the data was limited, we formed the hypothesis that a target gains of 0.1-0.2 is reasonable in introductory geoscience classes (where the classes showed statistically significant differences in pre/post GCI scores).

We tested this hypothesis using published and unpublished reports for four large classes (80+ students) using teaching methods that research suggests should yield learning gains. Two classes were within the range we identified (g = 0.14, 0.16) while another was lower (g=0.06) and the remaining class had a higher score (g=0.28). We will consider the factors contributing to different student learning gains as measured using the GCI.