Student exam creation as a learning tool

College Student Journal, June, 2004 by Jan Brink, Emerson Capps, Al Sutko

The purpose of this study was to determine if there is a relationship between: 1) comprehensiveness of student developed model tests and final exam score, 2) the quality of student developed model tests and answer keys combined, and final exam score, 3) student developed model test similarity to instructor final exam and final exam score. Students in a freshman fluid power class, in Manufacturing Engineering Technology, were given a homework assignment to create a final exam, and the corresponding answer key, over the material that they would be examined over on the final exam given by the professor. Student constructed exams and answer keys were handed in before the actual final exam. Two professors used rubrics to independently evaluate student tests and answer keys, and compare them to the instructor's final exam. Results were correlated with student performance on the final exam. Support was found for the first two hypotheses.

Introduction

Numerous studies have examined the effects of student developed tests on student performance and motivation with mixed results. Pack (1999) found that although student developed tests did not reduce test anxiety or increase academic achievement, they did help students focus more on learning and less on grade performance. Frase and Schwartz (1975) determined that college students recalled significantly (p<.005) more prose material when they constructed questions than when they just studied. However, student performance improved only on test items similar to the student constructed items. Lehman and Lehman (1984) found no difference in posttest performance between experimenter and student-generated questions over a museum "cave" exhibit. Students who responded to questions written by either themselves or the experimenter performed better on a posttest than those who merely observed the exhibit and read the case exhibits. Pearson (1991) trained college general biology students to generate and answer their own questions. This resulted in improved performance on weekly quizzes, but was not more effective than teacher generated questions in improving final exam performance. Smith (1990) concluded that having students in English as a Second Language classes write tests and answer keys over material studied in class and homework resulted in increased motivation. Brown (1991) stated that requiring students to develop an eight question final exam as preparation for the instructor developed take-home final exam resulted in the best final exam essays he had read for any course he had taught in ten years of college teaching. Ward (1973) found a significant (p <.01) improvement in exam scores for students who wrote multiple-choice items over course content, as compared to those who participated in a research project or did neither. Rash (1997) asked college computer science students to construct problems, and their solutions over an entire semester. The student-constructed problems, as a set, replaced one test. Although Rash did not directly measure the learning effect of this experience, students indicated, through course evaluations, that they expended more energy than usual in such a course. Students also reported that the assignments were worthwhile, resulted in increased understanding and were, sometimes, fun to do.

Research Method

Fifteen students in a freshman, four year Manufacturing Engineering Technology program were given a homework assignment at the end of the semester. They were instructed to create a final exam and the corresponding answer key over the material assigned by the professor.

This study was approved by the Human Subjects Review committee (#02112611) at the University. The study was designed to test three research hypotheses. The hypotheses examined whether there is a relationship between: comprehensiveness of student developed model tests and final exam score, the quality of student developed model tests and answer keys and final exam score, similarity between student developed model test and instructor final exam, and final exam score.

The final exam was the last exam of four exams and was not comprehensive. The material taught after the third exam was explicitly listed in the instruction sheet given to all students. It included material from the textbook, lab notes and homework assignments. All fifteen students participated in the assignment. Students were instructed to print their name on a blank page stapled on top of their exam and answer key and not to write their name on any of the other pages. This was done to facilitate a "blind" evaluation of their assignment. The department secretary placed numbers on the backs of their exams and created a cross-reference sheet with student names and removed the cover sheet. A team of two professors, both of whom have previously taught this class, evaluated the exams and answer keys of the students. Both instructors also compared the student's exams and answer keys to the actual final exam given.

The students were asked to create an exam that followed the format of the three previous exams, which consisted of essay questions, calculation questions and the graphing of hydraulic and pneumatic circuits. Multiple choice questions or fill in the blank questions were not used and the students were explicitly instructed not to use these question formats. Students were instructed to design their exams as follows: "design an exam so that a student who knows the material can complete your exam in approximately 60 minutes". Students were to create the answer key to their exams. This homework assignment was turned in before the final exam given by the professor. Most students turned in this homework assignment five minutes before the exam, although some turned it in a few days before the final exam given by the professor. None of the students missed the deadline and therefore all exams were eligible to be evaluated as homework assignments.