The effects of student self-assessment on students' attitudes and academic performance in a college mathematics course

The professional literature in mathematics education is filled with recommendations for changing the way mathematics is taught and how student performance is evaluated. Appeals to reform curriculum, instruction and evaluation are gaining acceptance. Unfortunately, there is little experimental evidence that supports the efficacy of these suggested reform measures. This research was designed to determine if student self-assessment, a form of alternative assessment, can enhance student academic performance and improve student attitudes in the mathematics classroom. The study involved four Finite Mathematics classes at The American University; two classes made up the experimental group, the other two comprised the control group. The experiment was designed so that instructional and evaluation procedures were the same for all students participating in the study. Student self-assessment supplemented other evaluation procedures for the experimental group: students in the experimental group assessed their own performance on a variety of tasks throughout the semester. Test grades and attitude surveys were used to determine if student self-assessment had a measurable impact on the experimental group's performance. Qualitative evidence was also collected during the course of the study. Conclusions were based on both statistical and qualitative evidence. One-way analysis of variance, the Mann-Whitney U test, and the t-test for paired samples were used to statistically test the study hypotheses. Statistical evidence indicates that end-of-study attitudes of experimental students towards mathematics assessment were significantly more positive than their pre-study attitudes. Statistical evidence also indicates that there was a significant difference in the academic performance of male and female experimental students at the end of the experimental treatment. Males scores on the final exam were significantly better than those of females when pre-study scores were taken into consideration. There were no other statistically significant findings that can be attributed to the experimental treatment. However, qualitative evidence and statistical testing of selected subgroups of the study population point out the benefits of the experimental intervention.