Computers and statistical methods are often applied in evaluating student performance. However, both the quantity and quality of information which a teacher uses in evaluation student performance is beyond that normally input into the computer. This means that the teachers' subjective values and decision-making processes should also be included in as parameters of any computerized system for evaluation of student performance. This paper describes a system for simulating student performance evaluation based upon previously stored objectives, test items and student responses. This system may be characterized as follows : 1) It is able to automatically generate rules for evaluation based upon relationships between interactively defined elements. 2) It is able to refer to subjective teacher decisions whenever it finds that the students' responses correspond to the expected degree of subordinate objective achievement. 3) The system can refer to and re-apply the rules for decisions developed in 2), above. 4) The system uses natural Japanese in interfacing with the teacher, thereby making his or her work easier. In order to determine the practicality of this system, it was used during 34 classroom hours of instruction in junior high school mathematics. The results were nearly identical with those predicted by the teachers. It was also found that since this is not an "expert system" based upon rules for evaluation developed by experienced teachers, but rather is a method for simulating the teachers' decision-making processes by using their own decision rules, it can function as a training system which permits teachers to clarify and more clearly recognize their own decision-making processes.

View full abstract

A survey of the computer literacy of 3,452 6th, 8th and 11th grade students was completed in 1984, at a time when few schools had installed computers for educational use. Intended to provide a baseline, it measured the levels of computer literacy of Japanese students before schools began to use computers in the classroom. The results were used to divide the students into four categories : (1) Program writers, (2) Non-program writers, (3) Future interest in computers, and (4) Indifferent/avoidant. Students in the first of these two groups could operate computers while those in the latter two could not. The percentages for the groups were 4, 13, 46 and 34%, respectively. These ratios varied little between types of schools. Most of the students had positive attitudes towards computers but the indifferent/avoidant group was less familiar with them. Students who already knew how to use microcomputers were more interested in mathematics, science and social studies. They were also better adapted to school life and activities. Their primary reasons for using computers were to play games and to learn programming. In these groups, the most frequently mentioned negative effect was that computers were hard on the eyes, while the most frequently mentioned positive effects were intellectual training and the making of friends. The authors also make a comparison of the results of this study with one done in California.

View full abstract

The author proposes a new format for answering two-choice questions, questions which are frequently used when introducing instruction, predicting the results of scientific experiments, promoting student discussion and so on. Using this one-dimensional multiple-choice format makes it possible to get a answer frequency distribution in two-choice problems. The author also proposes a selection model for the information processing steps involved in such decision making acts as choosing one answer out of a multiple answer problem. This model may be applied to the new answering format as a means for interpreting the frequency distributions of the answers. The results are as follows : 1) The frequency distribution of the answers in the deduced model coincide with measured data. 2) Using computer simulation permits surmising the process through which students choose one out of a set of multiple answers. 3) The model shows some degree of validity of "confidence," one measure for testing student cognitive behaviors.

View full abstract

Many kanji (Japanese-Chinese characters) can be resolved into meaningful components. This study tested the assumption that familiarity with such components makes memorizing and reproducing of kanji easier. Twelve unfamiliar, 12-stroke kanji were given as a learning task to 5th grade students. The subjects were divided into three experimental groups. Group 1 simply copied the kanji. Group 2 copied them using the stroke order proposed By SHIMOMURA (1978). Group 3 learned the meaning of the components while coping the kanji. Two sub-experiments were conducted. In the first, the students underwent four 10 minute practice periods followed by 3 minute tests, all done without interruption. In this sub-experiment, Group 3 performed significantly better than did the other two groups. In the second sub-experiment, three new groups of children were given the same learning tasks, but a one-week delay was inserted between the 2nd and 3rd practice periods. Compared to the other two groups, Group 3 did not indicate as great a drop in retention during this one-week delay. This indicates that kanji can be more effectively learned if broken down into meaningful components. It was also shown that the performance of lower achievers in Group 3 (as determined by a pretest) was much closer to that of the higher achievers in the same group than were those of lower abilities in Groups 1 and 2.

View full abstract