Predictors of Success and Failure in Basic Mathematics Courses: A Quantitative Analysis
by
Nellie Carrier
University of Southern Maine
Coauthors: Sat Gupta, David Harris

Virtually all college programs require that students complete at least one mathematics course for graduation. However, failure rates in these courses can be high (Windham, 2000) and many students report negative attitudes toward the study of mathematics (Wieschenberg, 1994). Failure rate at USM in fall 2001 for three entry-level math courses (MAT 105: Mathematics for Quantitative Decision Making, MAT 108B: College Algebra, and MAT 120D: Introduction to Statistics) was 20.7% compared to the failure rate of 9.6% for all other 100-level courses for the same term. In an effort to identify predictors of success and failure in basic mathematics courses, we studied data collected from 168 students enrolled in these three courses at USM during the spring and summer semesters of 2003. The student grades were added to the survey data at the end of these two semesters. Students provided demographic data, as well as information about previous mathematics experiences in high school and college and their attitude towards math study. Since the grades were somewhat skewed, we used various non-parametric methods such as Mann-Whitney test, Kruskal-Wallis test and the chi-square test to examine the impact on grades of various predictors such as gender, major (math/science vs. other), attitudes toward mathematics study, instructor rank, course location (on campus/off campus), use of technology in the course, current semester load, number of children at home, number of missed classes, and number of hours worked per week for pay etc. We found no difference in attitude toward math study by gender or major. We did, however, find the following significant predictors of higher final course grade: lower rank of the instructor, more use of technology, on-campus course location, missing fewer classes, and accessing fewer hours of tutoring. These results suggest that gender, once thought to be an important predictor of attitude toward and success in math (Chouinard et al., 1999), is not a significant factor in either, but that engagement in learning, as previously reported elsewhere (Smith et al., 1996), is vital to student success. Significance of the instructor rank was bit of a surprise (particularly lower instructor rank associated with higher grades), but it may be attributed to limitation on data size. They also show that merely accessing tutoring is not sufficient to insure student success, and suggest that efforts to identify struggling students early in the semester and to design individualized help programs may be needed. The significance of the use of technology was anticipated. We plan to expand this preliminary study to a much larger student population during the 2003-04 academic year.

Date received: September 3, 2003