Some Effects of Computer Technology on Human Interaction and Individualization in the Teaching of Deductive Logic(1)
Marvin J. Croy
Michael G. Green
James R. Cook
4.1 Predictions and Actual Outcomes
The research hypothesis put forward was that the difference between human-supplied and computer-supplied feedback would be associated with changes in student behavior and attitudes. Specifically, it was expected that having human as opposed to computer-supplied feedback would do more to promote learning, positive in class activity, and favorable student attitudes toward the course, their classmates and their instructor. Given the previous experience just described, the general pattern of results expected is shown in figure 2. (In this figure, each time frame is collapsed into a single point representing the results obtained for that entire time frame.) Group A receives human-supplied feedback in time frame 2 and computer-supplied feedback in time frame 3. This sequence is reversed for group B. Neither of these conditions were administered to either group during time frame 4. Notice in figure 2 that group A’s mean is higher than group B’s during time frame 2 but that the reverse is true for time frame 3. This outcome would support the hypothesis that having human-supplied feedback results in higher scores on the measure taken. In any event, the essential feature is the “crossing pattern” of the lines generated by the particular measure, and it is this pattern that will be searched for. It may well be the case that other patterns of results can support the hypothesis being explored, particularly given that carry over effects from one time frame to another may occur. Nevertheless, the crossing pattern between time frames 2 and 3 would provide the clearest indication of support for the hypothesis and of the sensitivity of the measures being used.
One hypothesis put forward on the basis of previous experience, then, is that students receiving human-supplied feedback would be more active in the classroom than students receiving computer-supplied feedback. “More active” has been defined here in terms of three behaviors: (1) raising questions during class; (2) answering questions posed by the instructor during class; and (3) consulting the instructor immediately after class concerning some aspect of the subject matter. A record of the frequency of each of these behaviors for each student was maintained throughout the semester. The results of these measures for each semester are shown in figure 3. The time frames of most interest in each chart are time frames 2 and 3. Recall that in time frame 2, group A students received their diagnoses from the instructor while group B students received their diagnoses from the computer. During time frame 3, the reverse is true. Each of these charts demonstrate that the mean number of in-class responses observed during time frame 2 is higher for group A than for group B. That is, students receiving human supplied diagnoses were more active in class. The same is true in time frame 3 when group B students begin receiving their feedback directly from the instructor. It is clear that the greatest difference observed between groups occurs during time frame 3 and that the overall pattern of the differences observed are similar across both semesters.
Thus far nothing has been said about the extent to which students attended or failed to attend class during these time frames. This is important since attendance rate would affect the total number of in-class responses observed for particular students and their groups. One way to separate this factor out is to divide the frequency of in-class responses by the number of classes attended. This produces a measure of in-class responses per classes attended for each student. The group means which result are shown in Figure 4 for the various time frames. (Because of late student drop-add activity, a seating chart was not established until time frame 2 for the Spring semester. Consequently no observations are given for time frame 1 during that semester.) Again, the “crossing pattern” expected appears between time frames 2 and 3. Time frame 3 continues to produce the largest differences observed. Class attendance rates themselves are provided for each semester in figure 5. The expected crossing pattern appears in the Spring data but not in that for the Fall.
As stated earlier, the effect of the treatment conditions administered on student performance in the course is not the central focus of this research, but Figure 6 provides a comparison of groups in respect to exam scores during each semester. The expected crossing pattern appears between time frames 2 and 3 in the Spring but not in the Fall. This exam performance can be broken down into two component tasks which are related to the two CAI programs used by students. Each exam contained a section in which students assessed given justifications for proposed inferences and a section in which complete proofs were constructed. These tasks correspond to the two CAI programs used and the type of diagnoses provided. The results for the justification task are shown in figure 7 for both semesters. Here, the predicted crossing pattern appears in the Fall but not in the Spring. With respect to the task of constructing proofs (Figure 8), the crossing pattern appears in the Spring but not in the Fall.
The data obtained by questionnaire is the most complex and challenging with respect to systematic analysis. Questionnaires were administered at four different points in the course. These points are marked as Q1 through Q4 in Figure 1. Each questionnaire contained approximately 30 – 40 items depending upon the time frame and semester in which it was given. Being explicit about what should be expected in terms of results here is more difficult than other measures since previous experience with student attitudes and their expression during the course is limited. The hypothesis previously put forward that having special student-teacher sessions would result in greater in-class responsiveness is, however, suggestive. It suggests that the special sessions would result in more favorable attitudes concerning various aspects of the course and students’ experiences in it. As will be seen below, analyses of the questionnaire data produce only partial confirmation of this expectation.
The analysis of questionnaire data proceeded as follows. The intercorrelations between items on the Fall questionnaires were examined in an effort to identify scales of similar content or meaning. Four scales were initially developed. These scales assess student attitudes towards the instructor, other students in the class, computers, and the course. Another scale was constructed to gauge the students’ estimate of the helpfulness of the special sessions with the instructor and the summary feedback from the computer. A final scale assesses student perceptions of the degree to which they interacted with their classmates both within and outside of class. Each of these scales is composed of 2 – 8 Likert-type items (ranging in numeric value from 1 to 5).
Student attitudes toward the instructor were assessed by the following items:
Figure 9 presents the results for each time frame and each semester in respect to expressing favorable attitudes toward the instructor. There is no crossing pattern exhibited in the Fall data between time frames 2 and 3, although there is a slight tendency toward convergence. However, the expected crossing pattern does appear in the Spring.
Attitudes towards fellow students were assessed through items which asked if “other students in this class seem: friendly, supportive, intelligent, helpful and considerate” respectively. The scale indicating the expression of a favorable attitude towards other students in the class is charted for both semesters in figure 10. Nothing revealed in the Fall data supports the hypothesis that having human as opposed to computer-supplied feedback affects this attitude, but the Spring results do show the expected crossing pattern.
Attitudes towards computers were assessed through the following items (except in the Fall semester when only the first 3 items were used):
Figure 11 presents the results from the scale assessing favorable attitudes toward computers. While a convergence appears between time frames 2 and 3 during the Fall, there is no indication in the Spring data of any changes that might be related to the treatment conditions administered.
Students’ attitudes towards the course were assessed by the following items:
Figure 12 exhibits the results for both semesters concerning student attitudes toward the course. These results are of interest because they suggest an opposite effect of that predicted. In the Fall, for example, group A attitudes are more favorable than those of group B as expected in time frame 2, but these differences increase rather than decrease in time frame 3. In the Spring, group B students express more favorable attitude during time frame 2, and this is clearly the opposite of what’s expected.
Students’ assessment of the helpfulness of the special sessions and the computer feedback was derived from the following items. My individual meetings with the instructor (or summary diagnostic report):
Using these same items, the fourth questionnaire was modified to have students make a direct comparison based on their experience as to whether the special meetings or the report from the computer was more helpful. The results show that students’ perceptions about the helpfulness of these alternatives varied from the Fall semester to the Spring semester. During the Fall semester, the students did not perceive the meetings with the instructor as being more helpful than the computer feedback. Instead, one group of students had a slight bias towards giving more positive evaluations at each time period. In the Spring, though, there were large differences between groups, with the special sessions being viewed as more helpful.
During the Spring semester only, the following items were used to assess the degree to which students were interacting with their classmates. The first two items were combined to measure how often students interacted during or near class time, while the last two items were combined to assess their interactions away from class.
Students having special meetings with the instructor (Group A) consistently indicated that they interacted with their classmates more than did Group B. There was also a general upward trend for each group, with more interaction occurring later in the semester. This would, or course, be expected in most classes, as students become familiar with fellow classmates and become more likely to interact with them.
Go to: 5. Assessing the Ethical Implications
Home > Teaching Resources > Computer Ethics Issues in Academic Computing > Some Effects of Computer Technology on Human Interaction and Individualization in the Teaching of Deductive Logic
HOME | IN
THE NEWS | RESEARCH
RESOURCES
TEACHING RESOURCES | STUDENT
RESOURCES | LINKS
The Research Center on Computing & Society
at Southern Connecticut State University
501 Crescent Street | New Haven, CT 06515
Director: (203) 392-6790 | e-mail: webmaster@computerethics.org
© 2000 – 2004 – Research Center on Computing & Society