Bera and Robinson (2004)

Bera and Robinson (2004) Exploring the boundary conditions of the delay
hypothesis with adjunct displays. Journal of Educational Psychology. Vol. 96, No. 2, 381–388.

Problem/Purpose
This article was selected due to its dual merit within the graphic organizer domain. First, it highlights the enduring debate surrounding the use of graphic organizers (GOs) and outlines within the instructional domain. Second, Bera and Robinson makes a significant contribution to the current knowledge base by not aiming to discount one or the other presentation format, but rather by attempting to identify individual context specific strengths of both formats by using the delay hypothesis as their theoretical anchor.

The present study was conducted in order to explain the contradictory results of two previous seminal research studies, by exploring the boundary conditions of the delay hypothesis with adjunct displays. That is, the authors wanted to determine what types of adjunct displays would result in optimal retention when different lengths of text and numbers of displays are used. To explore the possibility that certain types of adjunct displays lead to more distinctive encoding, they tested students who viewed outlines or GOs either immediately following acquisition or after a short delay.


Methods
Bera and Robinson performed two experiments. In the first experiment they wanted to replicate the findings of Robinson and Schraw (1994) but in a more realistic setting. Bera and Robinson made a total of two modifications; first, they used a modified longer text (1,000 vs. 200 words) in order to create more realistic classroom reading material; second, they reduced the testing delay (5 versus 25 minutes) to determine whether differences in retention could be observed when the interval between study and testing was similar to those experienced in traditional classrooms . In the second experiment, Bera and Robinson wanted to replicate the findings of Robinson and Kiewra (1995) using the original 6500 words text with multiple displays but with a reduced delay (5 minutes versus two days).


In the first experiment, the participants were 124 undergraduates all enrolled in educational psychology courses at the University of Texas at Austin. The design was an A2 (display: text plus outline vs. text plus GO) 2 (testing: immediate vs. 5-min delay) factorial. Students were randomly assigned to one of the four between-subjects conditions and were presented with a 956-word text, which was adapted from an undergraduate developmental psychology text focusing on different explanations of human development, then the students received two different displays , an outline and a GO. Assessment consisted of ten multiple-choice questions which were constructed to measure the knowledge of hierarchical and coordinate concept relations( See Appendix A,B).

In the second experiment both the 72 participants and the design were identical to those used in Experiment 1. The text of 6,500 words was taken from two widely used undergraduate introductory psychology textbooks. In total, seven outlines and seven GOs were constructed on two sets of displays, both showing equivalent information. The authors constructed two tests; the first, a relation-test contained 10 four-option, multiple-choice items that required knowledge of hierarchical and coordinate concept relations that were presented in both the text and the displays. The second, an application-test, contained 10 matching items that required the identification of disorders by giving novel examples of symptoms and behavior.


Results
In experiment 1, all of the statistical tests were conducted using an alpha of .05 and an A2 (display) X 2 (testing occasion) factorial analysis of variance (ANOVA) was conducted on the test scores. Here Bera and Robinson found that the main effect of display was not statistically significant and thus could not be ruled different from zero, F(1, 120)=1.00, MSE =2.87; meanwhile the main effect of testing occasion F(1, 120)=5.06, indicated that students scored higher when tested immediately than when tested after 5 min. This testing occasion main effect was qualified by the Display X Testing interaction, F(1, 120)=7.29. Tests of the simple effect of testing occasion within both of the display conditions indicated that the outline groups’ performance did not change from immediate to delayed testing, whereas the GO groups’ performance decreased from immediate to delayed testing.

In experiment 2, an A2 (testing occasion) X 2 (display) X 2 (test) repeated measures ANOVA was conducted using the relations and application scores as repeated measures. Here Bera and Robinson found that only two of the seven main and interaction effects were significant. The F values for the other five effects (all with 1 and 68 degrees of freedom), including the main effects of testing (1.25), test (1.32), and the interaction effects of Display X Test, (1.39), Testing Occasion X Test, (1.08), and Display X Testing Occasion X Test (0.01), were not statistically significant and could not be ruled different from zero. In addition the main effect of display, F(1, 68)=17.47, MSE=3.41, indicated that students who viewed GOs (M=12.00) scored higher than those who viewed outlines (M =9.41). This display main effect was compromised, however, by a Display X Testing Occasion interaction, F(1, 68) =4.95. And finally, tests of the simple effect of testing occasion within each of the display conditions revealed that for the outline groups performance did not change from immediate to delayed testing, but for the GO group, delayed testing was superior to immediate testing.

Discussion
Bera and Robinson state that the results of both experiments are consistent with the findings of Robinson and Schraw (1994) and Robinson and Kiewra (1995) but indicate that because the length of the texts and number of adjunct displays differed in both the previous two experiments and in the present two experiments, the inconsistency in results may be due to these differences. They describe the possibility that the length of the learning materials may explain these inconsistencies and posit that the distinctiveness of encoding hypothesis may explain why outlines are optimal with short text and a single display, whereas GOs are optimal when text is longer and multiple displays are used.

In other words, when the study process becomes too difficult with a long text and multiple outlines because GOs allow readers to grasp these relations more quickly and easily, the difficulty is lessened as the GOs are now permitting readers to encode the information more distinctively. This results in a more durable encoding compared with what is remembered from the more difficult-to-comprehend outlines.

Finally, the authors suggested future research that would investigate other populations in order to test whether differences between outlines and GOs in terms of the delay hypothesis do occur with elementary or high school students.


Critique
This paper was a very well researched, organized and written piece where the authors used clear and easy to understand language. One critique that was not acknowledged by Bera and Robinson was the less than optimal number of participants for factorial data analysis. All other shortcomings/ limitations were either addressed by the authors or were likely to be left out due to common editorial constraints.

Theoretical and practical implications

One theoretical implication of the Bera and Robinson study is that the results are likely to produce or at least demand that future GO research further explore relations and interdependence between encoding efficiency and encoding distinctiveness, the two factors that can produce the efficient and effective comprehension of main ideas.

Practical implication are likely to include the incorporation of Bera and Robinson efforts which call on educators to capitalize on students’ active involvement in the learning process into everyday in-class instruction and may lead to more content-appropriate handouts and displays in schools in all levels.