Abstract:

In this work, we describe an alternative assessment scheme for individual, out-of-class assignments whereby students are graded on their ability to self-identify and reflect meaningfully on mistakes rather than on instructor provided, absolute measures of correctness. We describe the implementation of this scheme in two university computer science courses and compare with offerings of the same courses where students completed assignments and were graded in a traditional manner. The two courses require students to construct proofs or apply mathematical concepts and formulas involving many sequential steps. This type of multi-step problem is similar to those commonly encountered in other fields, like physics, engineering, and mathematics, and we believe that the methods described here are broadly applicable in STEM disciplines.

Complex, multi-step problems can be challenging for students because there are many places where a mistake may be made, and a small, early mistake can lead to highly divergent results. Traditionally, such assignments are completed by the student and submitted to the instructor to be graded for correctness. Our scheme instead requires students to first demonstrate that they completed the work, but their work is not evaluated for correctness. Instead, students are then given the solutions and must evaluate their own work for correctness. Students are required to identify the type of mistake(s) they made from a list of possibilities, then must reflect on the cause of the mistake and strategize about how a similar mistake could be avoided in the future. Their grade is based on their ability to find and diagnose mistakes and the quality of their reflection.

We describe several potential benefits to our approach of self-assessed, reflective homework. First, as students are still learning, this scheme allows them to try and fail without penalty. While the adage “mistakes are opportunities to learn” may be often repeated by teachers, traditional STEM assignments can sometimes harshly penalize, and therefore discourage mistakes in the early phases of learning a new concept. Second, our scheme requires that students critically evaluate their own work, a skill that becomes necessary once students complete their education and may no longer rely on an instructor to determine the quality of their work. Finally, our method requires students to meaningfully engage with the material a second time, doubling their exposure to challenging concepts.

We found that measures of educational effectiveness were statistically equivalent between course offerings that used traditional, instructor-graded assignments and those that used self-assessed, reflective assignments, thus demonstrating the effectiveness of our approach. These measures include student self-assessment of learning outcome attainment, student rating of utility of assignments to their learning, instructor subjective assessment of learning outcome attainment, and objective measures of learning outcome attainment. Instructor subjective assessment also noted that the method appeared to increase the uniqueness of student work (indicating less cheating) and decrease student anxiety around grades. Future work will seek to more formally assess and quantify these potential benefits.

Keywords:

Self-assessment, Higher education.