2005 Salt Lake City Annual Meeting (October 16–19, 2005)

Paper No. 10
Presentation Time: 4:10 PM

ASSESSMENT OF DIVERSE GEOSCIENCE COURSES THROUGH RASCH DEVELOPMENT OF GEOSCIENCE CONCEPT INVENTORY SUB-TESTS


LIBARKIN, Julie C., Geological Sciences, Ohio Univ, 316 Clippinger Labs, Athens, OH 45701 and ANDERSON, Steven W., 903 Pinedale Dr, Spearfish, SD 57783-1624, libarkin@ohio.edu

The Geoscience Concept Inventory (GCI) is a multiple-choice assessment instrument developed for use in entry-level college Earth Science courses. A pool of 69 test items cover topics related to general physical geology concepts, as well as underlying fundamental ideas in physics and chemistry, such as gravity and radioactivity, that are integral to understanding the conceptual Earth. Each question has gone through rigorous reliability and validation studies (Libarkin et al., 2005; Libarkin and Anderson, 2005), with careful analysis of item stability and bias via Rasch approaches. Rasch analysis was completed with the one parameter logistic model, with analysis of bias via Mantel-Haenszel differential item functioning (DIF). A Rasch scale and methodology were developed for creation of 15-item GCI sub-tests that can be scaled to equivalent scores. Five items were removed due to gender and/or racial bias determined via DIF. Overall, students in the test population found the test difficult. For example, during 2002 testing students had similar but statistically different pre- and post-test means (pre = 42.2±12; post = 5.8±13; n=1295 student). Analysis of sub-populations indicates that students with low pre-tests (<40%, n = 388) dominate this effect (pre = 32±7, post = 41±10), with extreme significance on a t-test (tstat = 1.96 < tcrit =15), and an effect size of 0.46. Faculty nationwide can now create GCI sub-tests that are aligned with course content, facilitating the comparison of learning gains for a wide range of geoscience courses.