The Effect Of Announced/Unannounced Examinations On Student Retention

A hypothesis that students given unannounced quizzes in engineering classes, ins tead of announced midterms, would have a improved long-term r etention of the c ourse material was tested using two sections of an introductory fluid mechanics class. The hypothesis was based on the assumption that a series of unannounced quizzes min imizes "cramming" and results in more m aterial being placed in long-term mem ory. Long-term retention was found not to be s tatistically different for the two sections and the hypothesis was not proven. When learning and testing occur at high levels in Bloom's taxonomy, substantial m terial is automatically placed in long-term mem ory and r etention is independent of testing met hod, regardless of how much "cramming" occurs. INTRODUCTION Most engineering science classes evaluate student understanding of the m aterial through one or more examinations during the semester and a comprehensive final examination. Many students prepare for announced examinations through an intensive study immediately before the examination, e.g., cra mming. Although this method has proven to be a su ccessful method for preparing for individual examinations, its value for long-term r etention is questionable. Cra mming normally pl aces the learned material in s hort-term memory, where is it available for the imminent examination, but is forgotten shortly thereafter. It was hypothesized that students would r etain an understanding of the m aterial for a longer period of time if their study habits were altered so that they studied the same m terial multiple times over the c ourse of the semester instead of just prior to a nnounced examinations. This behavior would pl ace more of the material in long-term memory, improving retention for use in subsequent classes and after graduation. A number of methods exist to encourage students to study the same m aterial multiple times during the semester. Multiple examinations covering the same material would, at a minimum, force students to "cram" the same material a number of times. Multiple examinations, however, have the limi ation that they replace classroom learning time with evaluation time. This loss of learning time li its the number of in-class examinations that may be given. Another method for encouraging students to study the same m aterial multiple times is to give unannounced examinations. Not knowing which day the examination w ill be given encourages students to be more prepared at all times. Thus, they will likely review the material more times. P ge 154.1 1996 ASEE Annual Conference Proceedings In one study, students in three s ctions of an elementary ec onomics class were evaluated with three different testing procedures. 1 One section was told at the start of the semester that they would be given tests periodically during the semester, but were not told precisely when. The second s ection was given the additional information that they would be given a test "soon" and the test was then administered from between 2 and 7 days after that announcement. The third s ection was told in advance the exact date of the tests. The section with no advanced warning pe rformed significantly better than the other sections. No significant difference in performance was observed when students were told that the test would occur "soon" and when they knew exactly the date of the tests. Further, class attendance was higher for the section having no advanced warning. Bahrick2, Conway et al. 3, and Bahrick and Hall 4 reported mixed results connecting c ourse grade and long-term retenti on. Conway et al. 5 and Bahrick6 reviewed these differences and suggested that course grades may not reflect long-term retention because grading met hods can refl ect different types of learning through the course. They reported that performance on course work was more strongly correl ated with longterm retention than was final examination scores. It is noted however, that the types of c ourses evalu ated in those studies typically involved preparing papers with a final examination consisting of essay questions about course material. The validity of their conclusions is uncertain for quanti tative engineering c ourses where the type of problems given on examinations are similar to those worked during the semester as homework. DESCRIPTION OF STUDY In this study, the performance of students from two s ections of a required int roductory class on fluid mechanics was evaluated using either a series of either una nnounced quizzes or two announced midterms. The lectures for the two s ections were virtually identical. The quiz problems were s imilar to the problems given on the midterms, as was the total number of quiz problems given during the semester. The total time spend on evaluation during the semester was the same for the two s ction . The two sections were given a common final examination at the end of the semester. The students were also given an unannounced followup examination four months later in a subsequent class. The section receiving the una nounced quizzes had 14 students completing the semester and the section receiving the a nnounced midterms had 19 students. The cumulative GPA of the students entering the class was 2.91 for the quiz s ection and 2.80 for the midterm s ection. The cumulative probab ility distribution functions for the entry GPAs for the two sections are plotted in Fig. 1. To determine whether the higher GPA of the students in the quiz section was statistically significant, the t' test was condu cted. This test is similar to the t test except that it allows the standard deviation of the GPAs in the class to be used instead of the standard deviation of the underlying student population from which the students came. For this test, the variable t' is defined as 7 t ' = M q − M m S q 2