STUDENT DEMOGRAPHIC CHARACTERISTICS AND LEARNING GAINS on THE GEOSCIENCE CONCEPT INVENTORY (GCI) IN AN INTRODUCTORY GEOLOGY LAB COURSE

A 15-question Geoscience Concept Inventory (CGI; Libarkin and Anderson, 2005) was assigned as pre- and post-tests for students in 62 Physical Geology lab sections taught during the Fall 2009 and Spring 2010. We collected demographic data on 780 students completed who completed both pre- and post-GCIs and gave permission to for their data to be analyzed (74% of all students taking the pre-test). Here we consider the aggregate responses of a variety of demographic groups using pre/post GCI scores over both semesters.

Data analysis revealed that the average pre/post GCI scores were similar between the two semesters - Fall 47%/52% and Spring 46%/51%. Normalized gain scores (G = postscore%-prescore%/100-prescore%) were identical for the two semesters (9%). Pretest and post-test scores showed a significant difference on the basis of gender. Female students generally averaged a pre/post score that was 5% lower than that of male students but the average normalized gains were the same for both genders indicating that both male and female students learned a similar amount of new material.

We also saw some differences in student pre/post scores and normalized gains on the basis of major and academic rank. Students from some colleges (e.g., Engineering, Physical Sciences) recorded higher incoming GCI scores (>50%) while undeclared students averaged lower pretest scores (43%). However, with the exception of one college, normalized gains showed only modest variations on the basis of major (8-11%). Pretest scores increased steadily with academic rank from freshmen to seniors. However, normalized gain scores for sophomores were approximately 30% lower than all other academic ranks.

These results can be interpreted to suggest some implications for how we interpret GCI scores for introductory geoscience courses. A single test at the conclusion of a course may yield high or low scores depending on the demographic characteristics of the student population and may be a poor reflection of actual learning. The normalized gain is a more robust value to use to interpret learning gains. Even with the benefit of pre- and post-scores, students from specific colleges or academic ranks may underperform their peers, resulting in lower than expected learning gains in a course.