1991 Paper

A researcher wished to determine the effectiveness of a new computer-based method for teaching elementary statistics. The computer-based material consisted of information and examples which students can work through on their own. She decided to compare the performance of participants using the computer-based materials with that of participants working through a chapter from a standard textbook covering much the same material. She therefore recruited 24 students attending a Summer School (on the music of Java) at York University and assigned them to one of two groups (termed "Computer group" and "Textbook group"). Participants in the Computer group were tested four at a time (each working independently on a separate computer in a microcomputer laboratory), while participants in the Textbook group were tested individually in a quiet room. To ensure a random allocation of participants to groups, participants invited to attend in the morning were assigned to the Computer group, those attending in the afternoon were assigned to the Textbook group.

Participants in both groups were told to study the materials carefully as their knowledge would be tested afterwards. Participants in the Computer group were given brief instructions on how to use the keyboard, and then allowed 45 minutes to study the materials. The researcher estimated that at least ten minutes were required to become familiar with using the computer. Participants in the Textbook group were therefore allowed only 35 minutes to study the materials. Immediately following this, participants were given a written test for their understanding of the materials studied. Finally they were given a "learning style" questionnaire yielding a measure termed exploratory index (measuring a tendency to adopt an exploratory rather than a passive approach to studying).

After starting testing, the researcher realised that an additional group might provide further useful information. She was worried that the computer materials and the textbook materials did not cover exactly the same ground, so that it would be unclear whether any differences would be due to the materials themselves or to some other factor. She therefore printed out each computer screen on a separate sheet of paper (using a high quality printer) and stapled the pages together into a booklet. Twelve more participants were recruited for the third group. Since the University term had now started, first year Psychology students were used (who had not yet studied statistics). Participants in this group ("Booklet group") were tested using the same procedure as the Textbook group, but with the booklet replacing the textbook chapter.

The researcher predicted that the computer group would perform best, the textbook group worst, with the performance of the booklet group falling in between.

Details of the participants are presented in Table 1.

Table 1: participant details
GroupAge (years)Sex
Computer28.25M, 7F
Textbook32.17M, 5F
Booklet23.83M, 9F

A one-way between-subjects analysis of variance was conducted on the ages of participants, F(2,33) = 2.86, p = .07. The researcher concluded that there were no problems with the assignment of participants to groups as the ages did not differ reliably.

The researcher then analysed the results from the written test. The mean scores for the three groups were: Computer group, 59.7%; Textbook group, 54.8%; Booklet group, 71.1%. She conducted a one-way between-subjects analysis of variance which showed that the groups differed significantly (F(2,33) = 5.56, p < .01). She then conducted a series of t-tests to determine how the groups differed. The results are shown in Table 2.

Table 2: Pairwise t-tests comparing groups
Computer vs Booklet2.1622.042
Computer vs Textbook1.1622.256

The researcher concluded that, contrary to expectations, the booklet version of the materials was superior to both the computer version and the textbook version, and the last two did not differ.

Finally, she analysed the questionnaire results by correlating (Pearson product-moment) the exploratory index for each participant with his or her performance score on the written test. A separate correlation was computed for each of the three groups. A positive correlation indicates high exploratory tendency associated with better performance. The results were as follows: Computer group, r = +0.53; Textbook group, r = -0.26; Booklet group, r = +0.41. Only the first of these correlations is significant (p < .05, one-tailed test). The researcher concluded that performance on the written test depended on learning style as measured by the exploratory index, but only for participants using the computer.

On the basis of these findings, the researcher made the following recommendations to the course organiser. Students should first be screened for learning style using the exploratory index questionnaire. Those scoring highly on the index should be advised to study the materials on computer. Other students should be advised to study the same materials in booklet form.

[Back to Criticise-the-Experiment List] [Back to Main Page]