1. Background Information
 

Author(s)	Angela M. O'Donnell, Donald F. Dansereau,Thomas R. Rocklin, Velma I. Hythecker, Judith G. Lambiotte, Celia O. Larson, and Michael D. Young
Title	"Effects of Elaboration Frequency on Cooperative Learning"
Source	Journal of Educational Psychology
Year	1985

 

2. Abstract

This study investigated the effects of various methods of studying a passage of text on subsequent recall. Eightynine students in an introductory psychology course were divided into four treatment conditions, either as individuals or as dyads elaborating on the material being read once or more than once. Recall of two passages measured five days later was analyzed by using a series of ANCOVA's and repeated measures designs. Results suggest that the effectiveness of both individual versus cooperative learning and single versus multiple elaboration depends upon the type of information being learned. Descriptive information is likely best learned by frequent elaboration in cooperative learning situations. Procedural information is most likely best learned by rote memorization either individually or cooperatively.

3. Null hypothesis, -level 3rd sample size per group.

I. Measurement of understanding of the MURDER strategy

Ho: no treatment effect due to differences in the four treatment conditions
Treatment conditions:

1. Dyad - multiple elaboration n=20
2. Dyad - single elaboration n=27
3. Individual - multiple elaboration n=20
4. Individual - single elaboration n=22

Treatment conditions:
A. Dyads versus individuals
B. Single versus multiple elaboration
C. MURDER strategy versus individual plan
Descriptive scores
same list as above

III. Measurement of the elaboration scores
First passage
Ho: no treatment effect for A
Ho: no treatment effect for B
Ho: no interaction for A*B
Treatment conditions:
A. Dyads versus individuals
B. Single elaboration versus multiple elaboration
Second Passage
same list as above

4. Independent and dependent variables.

I. Independent variable was experimental treatment condition

Covariates were the Delta score, a measure of verbal ability and the GEFT score, a measure of field dependence/independence

II. A. Independent variables were dyads versus individuals, single versus multiple elaboration, and MURDER versus individual plan.

Dependent variable was the score for the procedural information in the second passage.

Covariate was the procedural score on the first passage.

B. Independent variable was dyads versus individuals, single versus multiple elaboration, and MURDER versus individual plan.

Dependent variable was the score for the descriptive information in the second passage.

Covariate was the descriptive score on the first passage.

III. A. Independent variables were dyads versus individuals and single versus multiple elaboration.

Dependent variable was the elaboration score on the first passage.

B. Independent variables were dyads versus individuals and single versus multiple elaboration.

Dependent variable was the elaboration score on the second passage.

5. Instrument, briefly comment on its reliability and validity.

Delta Reading Vocabulary Test and Group Embedded Figures Test, the two test instruments which were used ac covariates, were merely described. The only reference psychometric properties was that Delta was said to be moderately correlated with the verbal ability to the Scholastic Aptitude Test.

The strategy called MURDER is an extension of the SQ3R study method. It was developed by one of the authors, Donald F. Dansereau, in a previous study. The features of the strategy are described as being effective using different cited research studies. No validation of this strategy as a whole was discussed.

6. Experimental Procedure (brief)

The study was conducted over three sessions. During the first session, the subjects were randomly assigned to one of the four treatment conditions. Each subject then individually studied the MURDER strategy, developing his(he own elaborations. Each subject was then tested on the strategy to be sure that the subjects were equally familiar with it.

Following the individual testing, the subjects from the dyad conditions were then randomly assigned to same-sex dyads to practice the MURDER strategy on a passage about taking blood pressure.

During the second session, the subjects, who remained in the treatment groups to which they had been previously assigned, studied passage number one, a description of intravenous therapy. After studying that passage, they were all asked to study a second passage, a description of intramuscular injection, individually. However, half the subjects were told to use the MURDER strategy and half were told to use their own plan.

Five days later, a third session was conducted. All subjects were given a free-recall test on both passages and the Delta and GEFT tests.

7. Statistical analysis and conclusion.

Analysis of the dependent measures:

1. MURDER test - two scorers achieved 95% agreement
2. Free-recall tests - propositions were scored by three judges. Interrater reliability was .90-.91. Further, the propositions were classified as procedural or descriptive. Again, three judges were required to reach a consensus on the classification.

1. homogeneity of within-cell variances.
2. homogeneity of regression series of delta and GEFT and dependent measures

Difference due to assignment to use MURDER strategy versus ones own plan in view of previous assignment to treatment group:

Separate three-way ANCOVA's on the second passage, one for procedural scores and one for descriptive scores, each using their respective scores on the first passage as a covariate. The only significant factor was the Dyad/Individual X MURDER Individual plan. F(1,80), = 6.4 p > .05.

Post hoc analysis showed that of those using the MURDER plan, subjects who had worked as individuals outperformed those who worked in dyads.

Differences in elaboration scores due to treatment groups: Separate two-way ANCOVA's were performed on the first and second passages. No significant effects were reported.

8. If you were the researcher. how would you improve the study? (Be specific.)

The purpose of the study is to investigate ways to improve recall. We have no way of knowing whether or not the authors found improvement because they did not establish any base recall rate against which to compare their results. They report that the maximum mean recall demonstrated by any of the groups was only 30% of the available total score." (p. 578) While they may be able to justify that the difference between groups is "genuine" and that one could predict getting similar differences between groups in real-life, the group differences do not mean anything if the overall recall for across groups is lower than it would be if each person were left to his own devices. I would want to establish a base recall rate for material of equal difficulty.

Interpretation is further jeopardized by the lack of information about the subjects. They are 89 college students who enrolled in an introductory psychology course at a private religiously affiliated institution in the South. No attempt is made to describe the representativeness of the sample of any population. As a researcher, I would want to supply information which could be applied to a given population. Therefore, I would be more careful about subject selection and description.

The authors did not sufficiently explain several areas:

1. They presented no summary tables for their ANCOVA's. Thus, I would question the accuracy of the degrees of freedom for the F(1,80) reported on page 576.
2. The authors failed to specify the covariate used for the two-way ANCOVA's reported on page 577.
3. The authors described randomly assigning subjects to treatment conditions in the first session and then asking each subject to INDIVIDUALLY study the MURDER strategy and make their own elaborations. However, the first ANCOVA compares the test scores on the MURDER strategy by treatment groups. Why compare the groups if each person studied the strategy individually?
4. Before the subjects practice using the MURDER strategy on the blood pressure passage, the subjects previously assigned to dyads are reassigned to same-sex dyads. The authors fail to explain this move. They also fail to make clear whether or not the subjects remain in these same-sex dyads. If so, did the authors believe that sex was a factor. They did not measure it. As a researcher, I would hope to write clearly enough so that questions like these do not arise in the reader's mind.