The Effect of Computer-Assisted Instruction on Mathematics Achievement of Fourth Graders

Kay Richardson and Ilju Rha

With a majority of United States elementary schools having five or more computers for student use, the effectiveness of computer-assisted instruction is an issue being re-examined. The study was conducted for an Indiana school corporation which had recently installed a computer lab in one of four elementary buildings. This study was to examine the effect of computer-assisted instruction upon math achievement of fourth graders.

In the experimental group, the fourth graders worked primarily on computer drill and practice three times a week as part of their regular mathematics instruction. The remaining two days of the week were spent in the traditional instructional environment. Students in the control group received traditional instruction only for the five days. The experimental group was composed of 84 students and the control group contained 70 subjects.

Differences in pre- and post-tests score were examined through a series of ANOVAs to determine if significant differences existed between the CAI experimental group and the Non-CAI control group; between boys and girls; among the three pre-treatment achievement levels; and among the four different components of mathematics achievement.

A three-way factorial analysis of variance was employed in examining the data on achievement in the four components of mathematics: number concepts, calculation, problem solving, and total math on Stanford Achievement Tests. Initial findings indicated significant differences in achievement between the experimental group and the control group. Using the results from performing separate ANOVAs for each battery of tests indicated a significant difference among the ability levels in number concepts and calculations.

With the quadrupling of the number of computers in schools and tripling of the number of students using computers since 1983 (Beck~1986), the old question of the early seventies has re-emerged - the effectiveness of computer-assisted instruction. Computer-assisted instruction is the most widely use of computers in elementary schools. With the increase number of computers, is it time for schools to begin to expect significant mprovement in student achievement and is it the time to begin to measure the difference that computers are (or are not) making?

In the seventies, Jamison, Suppes, and Wells (1974) reviewed a wide range of CAI research and concluded that "at the elementary-school level, CAI is apparently effective as a supplement to regular instruction." In 1975, Edwards, Norton, Taylor, Weiss, and Dusseldorp found that CAI as a supplement to regular instruction uniformly effective, but results equivocal when used as a substitute for traditional instruction. This early CAI was delivered by mainframe computers, and the high cost, the lack of evidence on the effectiveness, and the general lack of acceptance by the educational community stymied the increase use in the public elementary and secondary schools (McDermott & Watkins, 1983).

With the advent of the microcomputer, the educational community is taking a second look at computer assisted-instruction. Many of the same issues are being researched as in the seventies and additional attention is being focused upon particular groups of students such as the effect of CAI upon disadvantaged pupils (Geller & Shugoll,1985) and (Mevarech & Rich, 1985); and upon learning-disabled students (Trifiletti, Frith, & Armstrong, 1984) and (McDermott & Watkins, 1983). Unfortunately in reviewing the literature on these projects it is evident that many of the confounding variables which led to conflicting findings in the past are still present in the research of the eighties. (Clark, 1985)

The Trifiletti, Frith, & Armstrong study involved twenty-eight learning disabled children with a median age of 11-10 in a private school in Florida. Subjects were randomly assigned to two groups: Computerized Mathematics Instruction and Resource Mathematics Instruction. The resource instruction was conducted by two teachers in two different rooms. The CAI group worked forty minutes a day on The SPARK-80 computerized instructional system. This particular system incorporated computerized tutorial instruction, drill instruction, skill games and assessment. The groups were compared with respect to the number of skills gained and yearly achievement scores. The results indicated a significant difference in the number of skills gained and in the mean achievement gain between the experimental group and the control group. One of the problems with this study was authors statement that the same material was "presumably" taught in the control groups. Another problem with the research was the analysis of the data based upon the actual differences between pre- and post test. Using the pre-test as a co-variant may have better reflected the effect of the CAI.

The McDermott and Watkins study involved 205 learning-disabled first- through sixth grade children in an Arizona school district. Ninety-six students in two elementary schools were randomly assigned to a mathematics CAI or spelling CAI group. The district's remaining students served as the control group. The experimental CAI instruction was provided through multifunctional microcomputer programs covering from fundamental mathematics to advanced elementary mathematics. The experimental program proceeded through a complete school year. Data on two different pre- and post tests were analyzed using simple factorial analysis and a co-variance analysis on repeated measures. The achievement gains of all three groups were found to be equivalent.

In the Mevarech and Rich study, the impact of CAI on mathematics achievement was investigated using students (376) in the third, fourth, and fifth grades. The results indicated that the those students in a traditional program supplemented with 2 days of CAI scored significant higher in math achievements as measured by Arithmetic Achievement Test developed by the Israeli Ministry of Education. The confounding factors in this study were a lack of correlation between the experimental activities and the control activities and a lack of pre-test data on the control and experimental students' achievement level.

The studies cited do not provide conclusive evidence to support or reject the effectiveness of computer assisted instruction but were selected because of the similarities to this study. The study was to examine the effect of computer-assisted instruction upon mathematics achievement of fourth graders. The main questions of this research were as follows:

1. Did a significant difference in mathematics achievement gains exist between the CAI supplemented math instruction and the non-CAI math instruction?
1. Did a significant difference exist for CAI supplemented, instructed 4tudents in achievement between the different areas of mathematics as measured by the Stanford Achievement Test?
1. Did a significant difference exist for CAI supplemented, instructed students in the achievement gains of students of different ability levels?
1. Did a significant difference exist for CAI supplemented, instructed students in achievement gains between boys and girls?

Method

Subjects

One hundred forty-four fourth grade students were involved in the study. All four of the fourth grade classes in one elementary school were selected as the experimental group (84) because of the installation of a computer lab within the school. The computer lab contained 15 microcomputers with single drives and were networked to access a single set of dual disk drives and two printers. The lab contained several tables which accommodated an additional 15 students and an overhead projector. A full time lab assistant was available to help the students in the operation of the equipment.

The classes were scheduled into the lab three times a week during their regular math instructional period beginning in the fall of 1984. During a forty minute period, they received teacher centered instruction and used the computers for drill and practice of the presented concepts. Some tutorial material was utilized; however the majority of the software would be classify as drill and practice. The remaining two math periods were conducted in the traditional instructional setting. All of the students participated regardless of their placement in special programs such as the gifted and talented program and the remedial title programs.

The control group (74) was designated by the director of elementary education of the school corporation. These students represented all of the fourth graders in a second elementary school which did not have a computer lab. Although the school did have several mobile computers in the building, the fourth grade students were not exposed to regular computer drill and practice as was the experimental group.

Instrumentation and Procedure

Test data from the Stanford Achievement Test administered in the spring of 1984 was used to classify experimental and control student's pre-treatment achievement level and served as the pre-test for the study. The test is used extensively throughout the United States to assess student achievement. Students who scored at least one grade level about the average (3.8) were grouped as high achievers and those who scored below 3.0 were grouped as low achievers. The Stanford Achievement Test was administered again in the spring of 1985 as a regular part of the district's evaluation of student achievement, and the results were used as the post test data for this study. The study investigated the differences in gains between pre- and post test.

Data Analysis

A three-way factorial analysis of variance was employed in examining the data on the differences in achievement in number concept6, calculation, problem solving, and total math. Separate ANOVAs were performed on the gains between pre- and post test scores for each component of the test and for total math.

Results

Number Concepts

Table 1 presents the ANOVA summary for achievement gains in number concepts. No significant effect was evident between the achievement of boys and girls and no significant interaction effect between sex and the other factors was revealed. This finding was repeated throughout the study. A significant effect was discovered between the experimental and control group indicating that gains of the control group were significantly higher than the computer-assisted instruction, experimental group (F 1,143) = 6.07, p < .01. The analysis revealed a significant difference in achievement means (Table 2) between the ability groups (F 2,143) = 6.56, p < .001. The post hoc analysis (Table 3) for different ability levels shows that there were significant differences between the high ability group and medium ability group and between the high ability group and low ability group with a 95% confidence level.

Calculations

Similar findings as reported for number concepts are indicated in the ANOVA summary (Table 4) for achievement in calculations. The control group's gains were higher than the experimental group (F 1,143) = 18.23, p<.001. Computer-assisted instruction is usually described as being an effective strategy for increasing students' calculation skills. The mean gains comparison (Table 5) indicates a negative gain for the high ability group who was involved in the computer-assisted instruction. Again, a significant difference was revealed between the ability groups (F 2,143) = 11.10, p<.001. The post hoc analysis (Table 5) shows a significant difference between the high ability group and the medium ability group and between the high ability group and the low ability group. Both of the post hoc results indicate that the lower ability group gained more than the high ability groups did, regardless of the treatment. No interaction effects were found to be significant.

The summary of the ANOVA findings (Table 6) for problem solving shows no significant effects on achievmenet gains (Table 7) for this component of mathematics achievement.
 

Total Math

The analysis of the total math component which represents a composite of the other three measures of achievement indicated a significant difference between the control and experimental groups (Table 8 and 9) The control groups shows a greater gain in achievement.

Discussion

The study does not provide any evidence to the support effectiveness of computer assisted-instruction in improving mathematics achievement. The results indicated that the control groups shows a greater gain in achievement. This finding was repeated throughout the study. Even calculation skills which are usually described as appropriate area for effective CAI shows the same results. This study is an example of the same type of generalized research about the effectiveness of computer assisted instruction that was conducted in the early 70ís. In fact, the present research may be even more general with the software and microcomputer providing more alternative in respect to the amount of time spent by students on a computer, the instructional activities, and the instructional environment.

This over-generalization of computer instructional activities could easily provide a multitude of conflicting research results. Conflicting results supporting or not supporting computer assisted-instruction will impede the adoption of any type of computer instruction by the education community. It may not be the time to ask the question on the effectiveness of computer instruction upon achievement but it is the time to conduct research which identifies the components of effective computer instruction. If the effectiveness of the software can be validated, other questions arise as to the type of environment in which the instruction is delivered; the correlation between the computer instruction and the conventional instruction; and as to the identification of which students will benefit the most from this mode of instruction.

Conclusion

In addition to the confounding factors, it should be emphasized that the findings are those of a fixed model and should not be generalized to a larger population. Because of the significant differences between the pre-test scores of the control and experimental group, additional analysis should be conducted to determine the correlation of the pre-test scores and the differences between pre- and post test scores to determine the feasibility of using analysis of co-variance to confirm the results. Another factor which might reflect a difference in the results is the criteria for determining the ability groups. However, making these adjustments will not account for the previously discussed confounding factors and the evidence concerning the effectiveness of computer-assisted instruction will remain inconclusive.

Bibliography

Becker, H.J. (1986) Our national report card: preliminary results from the New Johns Hopkins Survey. Classroom Computer Learning. 6(4): 30-33.

Edwards, J. Norton, S., Taylor, S., VanDusseldorp, R., & Weiss, M. (1974) Is CAI effective? AEDS Journal. 7.122-126.

Clark, R. (1985) Confounding in educational computing research. The Journal of Educational Computing Research. 1(20): 137-149.

Geller, D.M. & Shugoll, M. (1985) The impact of computer-assisted instruction on disadvantaged young adults in a non-traditional educational environment. AEDS Journal. 18, Fall:49-65

McDermott, P.A. & Watkins, M. W. (1983) Computerized V8. conventional remedial instruction for learning-disabled pupils. The Journal of Special Education. 17(1):81-88.

Mevarech, Z. & Rich Y. (1985) Effects of computer-assisted mathematics instruction on disadvantaged pupils' cognitive and affective development. Journal of Educational Research. 79(1):5-11.

Trifiletti J.J., Frith, G. & Armstong, S. (1984) Microcomputers versus resource rooms for LD students: A preliminary investigation of the effects on math skills. Learning Disability Quarterly. Vol 7, Winter: 69-76.