Visualization for middle school students' engagement in science learning.

This mixed-methods study explored the effects of student-generated visualization on middle-schoolers' science concept learning. We compared students who visualized during study time with those who did not and found that visualization as a study strategy led to students' improved test performances (p=.02). However, middle schoolers' scores on a test of science concepts did not improve as a result of a computer-based visualization workshop (p=.03). In fact, workshop participants scored lower on the test than did those who did not receive the workshop. Qualitative analysis identified elements in the school setting that interfered with instructional effectiveness in the computer-based workshop. Findings indicated that across groups students were quite unskilled at visualization. We found that visualization is a difficult but powerful study strategy and recommend that science curriculum focus on visualization of concepts.

**********

In Visual Communicating, Wileman (1993) defined visualization as, "the process of graphically or pictorially representing facts, directions, processes, data, organizational structures, places, chronologies, generalizations, theories, feelings, and attitudes." According to Wileman, visualization techniques range from providing graphic emphasis in text through highlighting to creating purely graphic, nonverbal representations.

In our study, we were interested in the power of student-generated visualization to facilitate learning. As Dewey remarked early in the 20th century, the work of instruction would be "infinitely facilitated" if teachers would see to it that their students were "forming proper images.... The image is the great instrument in instruction. What a student gets out of any subject ... is simply the images which he himself forms with regard to it." Dewey went on to say that teachers would be wise to spend time "training the student's power of imagery and in seeing to it that he is continually forming definite, vivid, and growing images of the various subjects with which he comes in contact in his experience" (as cited in Wileman, 1993, p. 7).

When asked what intellectual factors are most important for understanding physics, college physics professors, researchers, and teachers selected four factors from Guilford and Hoepfner's (1971) Structure-of-Intellect model of intelligence. They were the abilities to (a) visualize, (b) calculate mathematics problems, (c) reason logically, and (d) solve problems (Peltzer, 1988). Most scientists agree that visualization facilitates thought. Clark stated:

    Visualization has been the cornerstone of scientific progress
    throughout history. Much of modern physics is the result of the
    superior abstract visualization abilities of a few brilliant men.
    ... Virtually all comprehension in science, technology and even art
    calls on our ability to visualize. In fact, the ability to visualize
    is almost synonymous with understanding. We have all used the
    expression "I see" to mean "I understand." (as cited in Earnshaw &
    Wiseman, 1992, p. v)

One visualization method for knowledge construction commonly used by students is concept mapping. Concept mapping refers to the "process for representing concepts and their relationships in graphical form, providing students with a visually rich way to organize and communicate what they know" (Anderson-Inman & Ditson, 1999, p. 7). While concept-mapping, students post word labels and map their network of interrelated topics and subtopics by drawing lines and arrows between word labels. Previous research provides evidence that students can study efficiently by generating electronic concept maps, which demonstrate their understandings of how concepts are related to one another (Anderson-Inman & Zeitz, 1993).

Other less structured visualization methods include drawing and labeling images, drawing and highlighting distinctive features, graphically representing interrelationships among concepts (this includes outlining), and creating visual metaphors. Cifuentes and Hsieh (2003a, 2003b) found that college students who visualized interrelationships among science concepts in their hand written study notes performed significantly higher on a test than those students who did not show such interrelationships (p=.02, effect size=. 57). In that study, successful students identified and represented the underlying structures of text, (hierarchy, sequence, opposition, chronology, category, comparison, and cause-and-effect) using pens and paper.

In addition to helping students construct understanding, students' visualizations manifest the content and the structure of their knowledge of concepts, which provides their teachers with access to their levels of understanding. Given that "the purpose of the science class should be to encourage a fuller understanding of the working of the physical world, one that requires the articulation and investigation of one's own ideas" (Julyan & Duckworth, 1996, p. 56), then articulation of ideas through visualization is appropriate.

According to Gardner, (1999), one important way of knowing that students understand a concept is that they can generate multiple modes of representations. As Perkins and Unger (1999) explained "Understanding a topic is a matter of being able to think and act creatively and competently with what one knows about the topic" (p. 97). Student generated visualizations are one mode of representation that can inform teachers of student understandings and therefore become products for teacher feedback. Teacher feedback regarding students' levels of understanding empowers students to reflect upon their level of understanding and revise their visualizations while constructing new understanding. When accompanied by teacher mentoring regarding relevance, and opportunities for continuous reflection and revision, student-generated visualizations may be a powerful instructional approach.

The validity and appropriateness of a student-generated visualization are associated with the students' abilities to recognize and identify the underlying structures of text. Cook and Mayer (1988) indicated that because students are often unaware of the understandable structures of text, such as compare/contrast, classification, enumeration, generalization, and cause-and-effect, they are more likely to learn by rote instead of construct their own representations. They concluded that students who are unable to construct representations of a concept to display their understanding probably do not deeply understand that concept. Perkins and Unger (1999) went so far as to say that to constitute a clear understanding of concepts, students must construct their own mental representations.

Student-generated visuals surpass illustrations in their effectiveness for instruction because they are more personally relevant to students' understandings and prior knowledge, and because they contribute to construction of meaning (Anderson-Inman & Zeitz, 1993; Finke, 1990; Gobert & Clement, 1999; Papert & Harel, 1991). Tools for generating visuals include paper and pencil, pen, colored markers, or computer graphics tools.

Purpose of Study and Research Questions

This study explored the effects of student-generated visualization and a workshop on computer-based visualization on middle-schoolers' science concept learning. The research questions were (a) did middle school students who visualized concepts in learning materials during study time perform better on a test of science concepts than the students who did not? (b) Did middle school students who received a computer-based visualization workshop perform better on a test of science concepts than the students who did not? and (c) what elements in the school setting contributed to the effectiveness or lack of effectiveness of computer-based visualization as a study strategy for middle school science concept learning?

THEORETICAL PERSPECTIVE

Constructivist learning theory suggests that quality instruction provides learners with opportunities for practice with feedback, revision, and reflection (Cognition and Technology Group, 1999). Students need to actively engage in learning activities that provide them with opportunities to negotiate meaning internally as well as socially, rather than to passively receive explanations of phenomenon from those in the know. In addition, students need to build understanding in the context of mentoring from their teachers who help them find connections among concepts (Bransford, 2000; Fosnot, 1996). "Connections that are obvious to a scientist, may be far from obvious to a pupil" (Driver, 1983, p. 2). Such connections are made through experience over time. Student-generated visualization is one strategy for providing students with quality learning experiences that also give teachers access to the students' understandings. When students visualize concepts on paper or on computers, teachers have concrete representations of their understandings.

Computer-Generated Visualization: A Constructivist Strategy

Computer-based software such as Inspiration[TM] or Visio[TM] support visualization processes for constructivist learning. Additionally, images can be manipulated with image-processing software such as AppleWorks[TM] or PhotoShop[TM] drawing and painting tools to show interrelationships. These visualization tools can be regarded as "mindtools" to extend and reorganize learners' cognitive structures during learning (Jonassen, 2000). Learners can use graphic conventions for organizing their thoughts as they construct knowledge of concepts (Dodge, 1998). Computer-generated graphics created by learners offer several advantages over pen and paper, such as ease of subsequent revision and generation of sophisticated looking graphics by students with undeveloped artistic skill.

Computer graphics software and image processing software have become pervasive in today's schools. Such software allow students to access a variety of tools that might help them draw and paint objects to visually organize and represent what they know. Student-generated interpretative illustrations, or visualizations, can help students clarify for themselves the concepts expressed in texts and facilitate their comprehension of abstract concepts. When students are able to manipulate images during knowledge construction, they tend to engage in the meaning-making process and understand and remember concepts better than through the traditional transmission approach of instruction (Jonassen, 2000). Computer graphics tools allow students to create and manipulate shapes and images. These images can be edited, analyzed, filtered, enhanced, and animated to reveal relationships and information not visible to the human eye. "[Image processing] is an ideal tool for encouraging student discovery, promoting a constructivist science, or math experience" (Mann, 1997, p. 1).

Several research studies have provided evidence that computers can be used to support and enhance independent studying (Anderson-Inman, 1992; Anderson-Inman, Redekopp, & Adams, 1992; Anderson-Inman, Knox-Quinn, & Horney, 1996; Tenny, 1992). These studies explored the effects of computer-generated concept-maps and outlines on learning. The use of computers in the externalization of students' knowledge structures enriched individuals' mental models for organizing, retrieving, and using knowledge (Williamson, Jr., 1999).

Further investigation into the effect of computer graphics on learning needs to be conducted. Constructivist learning theory indicates that students need opportunities to form mental representations and receive feedback from experts regarding the accuracy of those representations. Student generated visualizations provide for constructivist learning. Previous research indicates that representing underlying structure of text by concept mapping, outlining, or showing interrelationships, helps students learn concepts and prepare for tests. In addition, student generated computer graphics have been shown to enhance learning. Therefore, we hypothesized that given training, students would benefit from using computers to represent underlying structure of text.

METHODS

We adopted mixed research methods, including both experimental and naturalistic analyses of data collected in a school setting. Teachers participated in the delivery and management of the treatment of students who received those treatments in the context of their typical class periods. Following treatment, students' scores on a test were compared across groups. In addition, qualitative data was used to gain understanding of the natural classroom learning environment that contributed to or diminished the generalizability of the findings (Shulman, 1997).

Participants

The students engaged in this study were junior high school students in science classes at a rural, public, junior high school. They had previous experience with AppleWorks[TM] drawing and painting software, although none claimed to have used graphics tools, visualization, or concept-mapping to support their study time. The potential participants were the entire 7th and 8th grade student body (N = 110). However, some of those students did not turn in consent forms, some were absent for part of the treatment, and others were absent for testing. Therefore, only 88 students participated in the complete study. Science classes were randomly divided in half from both the seventh and eighth grades so that approximately half of the participants were assigned to the no-workshop group (N = 45) and approximately half of the participants were assigned to the workshop group (N = 43). The groups were comparable across age, gender, and ethnicity. The no-workshop group consisted of 22 seventh graders and 23 eighth graders. Twenty-six were male and 19 were female. Twenty-five were Caucasian, 10 were African-American, and 10 were Hispanic. The workshop group consisted of 20 seventh graders and 23 eighth graders. Twenty-one were male and 22 were female. Twenty-two were Caucasian, 15 were African-American, four were Hispanic, and two were identified themselves as a mixture of races.

After treatment, participants were asked to rate the extent that they had previously been exposed to the information over which they were tested. To determine if the two groups varied in their prior knowledge of the textual material, a t-test was conducted. No differences were found; therefore, we could assume that participants in both groups had similar entry knowledge regarding the concepts over which they were tested.

Design

Each of the participating science courses were listed on strips of paper, placed in a hat, and drawn to randomly assign them to one of the two groups:

* The no-workshop group received print-based, verbal material on a science concept (General Properties of Matter) and was given 50 minutes for unguided and independent study prior to the test. Students did not have access to computers during classroom study time. At the end of the 50 minutes, students kept both their study notes and reading material. The next day, and prior to taking the test, students handed in their study notes. These activities were led by both researchers.

* The workshop group participated in three 50-minute training sessions on how to manipulate and generate computer graphics during study time. Students then received the same print-based, verbal material on the General Properties of Matter that the no-workshop group received. Students were given 50 minutes to study prior to the test. Students were given access to computers during their study time. At the end of the 50 minutes, students kept both their study notes and reading material. The next day, and prior to taking the test, students handed in their study notes. These activities were led by both researchers.

The no-workshop group met in the science classrooms. The workshop group met in one of the school's three computer labs that are continuously made available for class use by teachers. Teachers sign up for the labs and frequently take their classes to the lab to conduct activities in their content area. The lab and the classrooms provided naturalistic, familiar settings for the students to conduct their work.

The Computer-based Visualization Workshop

The objectives of the workshop were for students to be able to (a) recognize underlying structure of text (interrelationships), (b) illustrate underlying structure, (c) relate new concepts to prior knowledge, (d) highlight distinctive features, and (e) use graphics for review. We, the researchers, facilitated the workshop and modeled visualization of concepts using seven underlying structures: (a) cause-and-effect, (b) hierarchy, (c) chronology, (d) sequence, (e) opposition, (f) comparison, and (g) category. We then gave the students a turn practicing generating visuals illustrating each of the seven interrelationships by creating causal chains, flow charts, timelines, sequential diagrams, tables, charts, and matrices. We provided feedback and encouraged revision. We also modeled direct representation of concepts, showed how to highlight distinctive features of concepts, and gave the students opportunities to directly represent concepts and highlight distinctive features. Additionally, we gave students 15 short texts to use to practice identifying underlying structure and visualizing.

The Text, the Posttest, and Procedures

The day after the trained group completed the workshop, all participants from the two groups were given eight double spaced pages of text describing the "General Properties of Matter." The textual information was an excerpt from "Chapter 2: General Properties of Matter" of the book Prentice Hall Exploring Physical Science (Maton, 1997), designed for middle school students at the ninth-grade level. Ninth grade-level reading material was used to assure a high level of difficulty and lack of previous student exposure to the content as well as to increase the variability in scores.

Students were encouraged to keep their text describing the "General Properties of Matter" and their study notes and use them for study and review overnight prior to taking the test. We chose to give the students flexible study time because of a recommendation from previous research (Cifuentes & Hsieh, 2003a, 2003b). Student behavior in that earlier study indicated that, given flexible study time, students trained in visualization skills would likely choose to spend more time-on-task. The natural setting of schools generally offers students flexible study time and we wanted to use a setting that was as natural as possible.

The 30 multiple-choice posttest items came from the instructional support materials for the textbook (split-half reliability, r=.73). All items focused on the relationships among important concepts contained in "General Properties of Matter." Additionally, they were validated by the science teachers and were considered appropriate for the study. An example of a test item on the concept of mass follows:

Among the following places, an object would have the greatest weight

a) on the surface of the Earth.

b) on the moon.

c) at the bottom of a mine.

d) out in space

The test was scored by assigning 1 point for each correct answer and 0 points to each incorrect answer.

After the completion of the posttest, participants filled in a "Study Strategies Questionnaire." In both groups, students were asked on the questionnaire to describe in detail the steps that they took to prepare for the test. All participants' study notes, printouts of computer graphics, and graphics saved to discs were collected.

Using a posttest-only-control-group design, posttest scores were compared across groups. The design of the study avoided effects of initial bias, previous testing, maturation, instrumentation, regression, selection, and mortality because groups were randomly assigned and were not pretested (Huck & Sandler, 1979).

Data Analyses

The five data sources included: (a) posttest scores, (b) researchers' journals, (c) students' study notes, (d) students' computer graphics, and (e) the "Study Strategies Questionnaire." In a blind analysis, student's study notes and computer graphics were compared qualitatively and categorized as including visualization elements or not including visualization elements. If a student included visualization elements, we categorized them as visualizers; if they did not, we categorized them as nonvisualizers. Then, we matched student's study notes and computer graphics with their questionnaires. Students' self-reports of whether or not they visualized during study time on the "Study Strategies Questionnaire" were used to validate conclusions regarding our classification of students as visualizers or nonvisualizers.

We then estimated the effects of (a) type of study strategy (visualizer or nonvisualizer) and, (b) group (no-workshop and workshop) on the posttest by comparing scores using planned contrasts in a general linear model. We conducted a two-way analysis of variance (ANOVA) with the test score as the dependent variable. We estimated effect size for standardized differences using Cohen's d (1965) that defines effect size as difference between means divided by the pooled within group standard deviation (Gall, Borg, & Gall, 1996).

We applied content analyses approaches, as described by Emerson, Fretz, and Shaw (1995), to our journal entries, students' study notes, computer graphics, and study strategy questionnaire results. During and upon completion of data collection, we used the two-phase process of content analyses, open coding and focused coding, to analyze the data and identify factors contributing to the effectiveness or lack of effectiveness of computer-based visualization as a study strategy for middle school science concept learning.

RESULTS

Results of data analyses provided answers to the research questions regarding the effect of visualization, the effect of computer-based training in visualization, and elements in a school setting that contribute to effect of visualization training.

The Effect of Visualization

Middle school students who visualized science concepts in learning materials during study time performed better on the test of those concepts than the students who did not. In the no-workshop group, 9 students visualized and 36 did not visualize while preparing for the test. In the workshop group, 22 students visualized and 21 students did not visualize while studying for the test on the "General Properties of Matter." Visualizers yielded a mean score of 14.19 (SD = 3.71) while nonvisualizers yielded a mean score of 12.46 (SD = 5.30) (Table 1). There was a statistical difference between the scores of visualizers and nonvisualizers (p = .030, d = .38). Specifically, the students who visualized concepts while studying performed better on the test with a small effect size (Table 2).

The Effect of the Computer-based Visualization Workshop

Middle school students who received the computer-based visualization workshop produced lower scores on the test of science concepts than the students who did not participate in the workshop. The no-workshop group yielded a mean score of 14.18 (SD = 5.27), while the workshop group yielded a mean score of 11.91 (SD = 4.10) (see Table 1).

The results of the ANOVA revealed significant differences between study strategy application and between treatment groups (see Table 2). Further, the workshop created a negative treatment effect (p = .029, d = .48). The group that received the visualization workshop performed significantly worse (with a medium effect size) on the posttest than the group that did not receive the workshop.

There was no interaction between group and study strategy on students' performances (see Figure 1). In the no-workshop group the visualizers (M = 14.44) and nonvisualizers scores (M = 14.11) did not differ (p = .06). But, in the workshop group, visualizers (M = 14.09) scored much higher than did nonvisualizers (M = 9.62, p = .000). The low scores of the nonvisualizers in the workshop group and the high scores of the no-workshop group overall explain the discrepancy between the scores of the no-workshop and workshop groups. In addition, the visualizers in the workshop group and both visualizers and nonvisualizers for the no-workshop group scored significantly higher than the nonvisualizers in the workshop group explaining the lack of interaction between group and study strategy.

Elements in the School Setting that Contributed to the Lack of Effectiveness of Computer-based Visualization

The elements in the natural school setting that contributed to the lack of effectiveness of the computer-generated visualization workshop included--motivational problems among students, distractions of the computers and software, innate difficulty of the skill of visualization, and the limited time for learning the difficult skill in a three hour workshop.

Students' responses to the workshop on computer-generated visualization partially explain the lack of effectiveness of computer-based visualization. Additionally, study strategies applied during test preparation across the workshop and no-workshop groups differed dramatically impacting the effectiveness of the computer-based visualization workshop.

[FIGURE 1 OMITTED]

Students' responses to the workshop on computer-generated visualization. Student participants did not demonstrate positive motivation toward the workshop. Several students asked if the work in the workshop would be graded. When we told them "no," their engagement in the workshop decreased. Very early in the workshop two students asked if they could leave in voices for all to hear. One student said, "When can I get out of here," and another asked the whole class sarcastically, "Are we having fun yet?" Students noisily entered the workshop each day and visited with friends a great deal throughout the workshop.

Some students talked and visited with each other rather than focus on the learning tasks. Others left in the course of the workshop because of illness, or extracurricular activities. Although data for those students were excluded from the statistical analyses, we believed that the disruptions impacted results for other students.

Moreover, those students who were at computers were highly distracted by the computer software before them. Most drew graphics that were totally irrelevant to the science concepts being explored, or played with computer games. For instance, there was an electronic bomb that could go off on the screen and several students enjoyed setting-off that bomb frequently during the workshop. They enjoyed creating sounds and playing them loudly for the class to enjoy at unexpected moments.

Several students said that identifying the underlying structures of the text was quite difficult for them, especially the hierarchical, oppositional, and cause-effect interrelationships. We found that most students were often unable to create appropriate representations of new concepts. We also noted that most students were incapable of or unwilling to thoroughly and accurately represent texts. When, for instance, they were asked to represent the periods included in the Mesozoic Era, one student neglected to draw arrows in her timeline to indicate continuity, and did not include the label "Mesozoic Era" in her graphic. She also placed the periods in the wrong order on a timeline indicating her lack of understanding (Figure 2).

Text to visualize: Scientists divide the Mesozoic Era into three periods. The oldest period is called the Triassic Period. The middle period is called the Jurassic Period. The youngest period is called the Cretaceous Period.

[FIGURE 2 OMITTED]

Another student misrepresented the visual of an iceberg. He was enthusiastic about drawing an iceberg, but ignored the text's main idea that only a small part of an iceberg is above the waterline and the rest of an iceberg is under water (Figure 3).

[FIGURE 3 OMITTED]

Nevertheless, some workshop participants successfully identified and visualized the underlying structure of some of the texts they were studying. For instance, most students were able to successfully represent the sequence of the moon phases as well as accurately place periods within eras on a timeline. Also, many students were able to directly represent concepts. For instance, one student generated a direct representation of how the earthworm breathes (Figure 4).

[FIGURE 4 OMITTED]

At times in the second day of the workshop, we noted that only a few students were on the same page that we were working from as we modeled how to visually represent the concepts. Those times corresponded to our modeling of causal, hierarchical, and oppositional interrelationships in concepts and students' subsequent generation of graphic representations. Those interrelationships were the most difficult for students to understand and we suspect that this was why we lost their interest. Those who were not on the right page were not working on activities associated with learning how to visualize. They were instead, playing on the computer or fixated on working on a graphic representation that they created for the previous task in the workshop.

Several of the middle school students struggled to understand concepts of interrelationships being presented in the workshop. For instance, when given the following short text and then asked to identify the underlying structure of that text (cause & effect, chronology, sequence, comparison, opposition, category, or hierarchy), none of the students were able to indicate the text structure as cause and effect:

    The earth's climate has cooled and warmed naturally with irregular
    fluctuations over millions of years. However, man's activities are
    contributing to climatic changes. As a result of man's activities
    during the industrial and nuclear ages, the rate of climatic change
    is predicted to increase dramatically. (Cifuentes, 1992, p. 131)

Interrelationships being explored in the workshop needed to be defined, exemplified, and explored. We found that none of the students could define "hierarchy" and none believed that they had ever heard the word or idea before. We observed that the only type of interrelationship that students could identify from text with ease was the sequential relationship. Visualizing concepts appeared to be utterly new to the middle school students.

The provided workshop time appeared to be insufficient and students had great difficulty internalizing visualization as a study strategy within the time-frame and experiences of the workshop.

Participants' study strategies during test preparation. After the workshop, the workshop participants were asked to study the text regarding the "General Properties of Matter." We encouraged those students to use the visualizing techniques that they had explored in the workshop to represent the interrelationships among science concepts during their study time. However, when left on their own to use visualization as a study strategy, the workshop students failed to make good use of their time for studying the entire assigned text. In class, students who received the workshop studied from 10 minutes to 50 minutes, with the majority of students spending less than one half-hour. Twenty-six (65%) of those workshop participants reported that they studied at home with the time ranging from 10 minutes to 2 hours.

Many workshop students appeared to lack motivation to use their time for study. Several students expressed that they already knew most of the information; therefore, they spent very little time studying either in class or at home. However, test scores indicated that those students did not know the information. A few students chose not to study the material at all. Most of them were not anxious about the test.

Many students who participated in the visualization workshop attempted to highlight or underline the important ideas in the text during their study time. Several of them typed their study notes on the computer or wrote them down on paper. When they encountered difficulty with understanding certain concepts, two students drew examples to help them comprehend those concepts. One student reported that he drew in his mind while studying rather than drawing on the computer, so that he could remember those key concepts more efficiently. Another student expressed his concern about the appropriateness of his drawings, and indicated that the representations that he generated could be helpful for the preparation of the test if they were valid. Some of the students reported that note taking during the study time did not actually help them retrieve the information better while testing, and they perceived the test to be quite challenging.

The processes of (a) identification of the underlying structure of concepts presented in text, and (b) subsequent creation of visual representations associated with those concepts required a lot of time and effort from students. Most of the middle school students were unable to identify or represent underlying structure on their own during study time. None of them created causal chains, flow charts, timelines, sequential diagrams, tables, charts, or matrices to organize the material that they were studying.

They claimed that visualization was too hard for them. However, for some students, the visualization process helped them to think hard about what they needed to learn, and learning that was not measured on the test resulted from such thinking. For example, a student generated a visual representation to help himself understand the conceptual relationship between inertia and weight (Figure 5).

[FIGURE 5 OMITTED]

When students were able to extract meaning from text and generate relevant representative images, they built their complete understandings. For example, as one student tried to understand that weight of an object changes according to altitude, but that mass remains the same, he generated a computer graphic accurately conveying an object with the same mass on a mountaintop and in a deep mine (Figure 6).

[FIGURE 6 OMITTED]

Some of those students who received the computer-based visualization workshop spent time visualizing what they already knew rather than grappling with a new concept. Their visual representations were often inappropriate or misrepresented concepts, which failed to facilitate concept learning. Detailed visualization was entirely new to all of the students and very few of the students who visualized during study time appeared to use visualization to construct meaning.

The introductory paragraph of "General Properties of Matter" used a moon rock to introduce the concept of matter. Most of the students read the first paragraph and began generating a complex visualization including the moon with astronauts, rocketships, stars, the earth, and finally an image of a moon rock. The students spent a lot of time beautifying their representations. Creating such complex images distracted learners from their studies and students did not focus attention on the main ideas of the content. Several of the students in the workshop group never read beyond the first page of the eight-page text because they so busily worked on their illustration.

As in the workshop, students were distracted by the fun computer software and the computer graphics tools during study time. Most of the students spent their time creating irrelevant images and generated artistic visuals that involved inefficient use of time. While studying the "General Properties of Matter," the workshop group simply read the first two or three pages of the text, and then either played on the computers or selected images from the text to draw, rather than focusing on illuminating understanding through visualization.

The no-workshop group contrasted dramatically with the workshop group during study time. They studied quietly the entire study time in class. Students did not talk or visit or disrupt other students' studies. Thirty-two (70%) students continued to study at home beyond the 50-minute class time provided. The studying time of the no-workshop participants ranged from 50 minutes to 3 hours, with the majority of students spending more than one hour. Most students spent 30 to 50 minutes of class time studying quietly and approximately 1 half to 2 hours studying at home. One student even self-reported that he studied for three hours at home.

Twenty percent of these students highlighted or underlined the important messages in the text while studying. Several of them took notes and wrote down the definitions of some keywords on paper for better recall. Those who visualized created verbal study notes with structural elements such as highlighting and indention for subtopics. Most of them read the material repeatedly for memorization. One student indicated that he generated representations in his mind while he was studying. Another student reported that he was drawing while studying and indeed we observed that he was the only student in the no-workshop group drawing during study. Two students expressed that during the study time, they would ask themselves questions, which they thought would be on the test. One student had her mother ask her questions pertaining to the content of the material after her studying.

A few students reported that because they already knew a lot of information regarding the "General Properties of Matter," they only read in class and did not spend any time studying at home. Additionally, two students said that they did not spend much time studying the text because they assumed that the test would be quite easy.

Summary of Qualitative Findings

The naturalistic environment of the study affected the comparative impact of visualization as a study strategy across groups. Observation and qualitative data from the groups provided insight into why the computer-based visualization workshop was ineffective. Workshop participants had difficulty making good use of their study time to read and visualize the important concepts underlying the text. They were not adequately skilled in visualization to be able to visualize effectively on their own to prepare for test taking. In addition, they appeared to be unmotivated to learn visualizing and study quietly for the test. It may be that the new strategy interfered with more effective study strategies that they would have used had they not been told to visualize. Tools on the computers distracted the students from their studies. Only one of the students who had taken the workshop described application of a metacognitive strategy that he used while studying for the test. That strategy was visualization. On the other hand, the no-workshop group spent their entire study time working quietly to prepare for test taking. Several of the students described application of metacognitive strategies while they studied.

SUMMARY AND DISCUSSION

Among middle school students, visualization, ranging from simple highlighting to computer graphics generation had a positive effect on students' test scores. However, training in the use of computer graphics tools for visualization during study time had a negative effect and impact on learning as measured by test scores and observation. Computers and the technical environment of the school computer laboratory proved distracting for the participants in this study. Students in this study found both identification of underlying structure of text and visualization to be quite difficult.

The Positive Effect of Visualization on Learning

In spite of problems associated with the natural setting of a school, some students engaged actively in the meaning-making process of studying while we provided scaffolding. Those who visualized, scored significantly higher on the test than those who did not. Students who successfully formulated mental representations of concepts and then concretized those representations as computer graphics and those who highlighted and structured study notes while studying applied a strategy for spending time thinking in order to learn.

The students in the workshop group who did not visualize were the students in the study who scored poorly on the test. Visualizers in both groups and nonvisualizers in the no-workshop group scored the same statistically. The nonvisualizers in the no-workshop group did not have computers in the room to distract them. However, the nonvisualizers in the workshop group (those who explain the difference between groups) were encouraged to use computers and were seated at computer workstations. They had difficulty using the new study strategy and chose to play on the computers rather than to use the computers as a tool for their learning. Therefore, computer graphics for visualization interfered with their learning. The visualizers in the workshop group benefited from the training. It is quite possible that given enough experience with the strategy, they would become comfortable with the process of computer graphics for visualization and it would not interfere.

Qualitative data provided explanations for why the nonvisualizers in the workshop group scored much lower than both the visualizers and nonvisualizers in the no-workshop group. Nonvisualizers in the workshop group might have had (a) less motivation for studying the "General Properties of Matter," (b) insufficient ability to visualize the content in the "General Properties of Matter" because of the short period of the workshop, or (c) inability to generate relevant graphics during study time due to time and effort required for visualization.

The Negative Effect and Impact of the Use of Computer Graphics Tools on Student-Generated Visualization

When asked to use computer graphics tools to create visual representations of concepts, seventh and eighth graders spent a lot of time engaged in playful episodes, which were the result of being distracted by the perceived computer affordances. Downes (2000) indicated that young children often perceive the computer as both a tool and a toy, and when they are thinking and using the computer as a tool, it is "playable." Such a conception of the computer enables children to complete their work-related tasks more through playful means. Downes claimed that without adult guidance, young children who use computers during study time are more likely to spend their time fiddling around instead of concentrating on purposeful tasks. In our case, students played in spite of explicit instruction and supervision from two adults. Playing on computers was quite compelling for our participants.

Children's tendencies to play on the computer while doing school-related work may be explained by their lack of metacognition. Metacognition is "being aware of our thinking as we perform specific tasks and then using this awareness to control what we are doing" (Marzano, 1988, p. 27). Metacognition requires students to monitor activities during studying, such as checking their level of understanding, goal setting and planning, evaluating the effectiveness of their efforts, budgeting time, and organizing and transforming information (Baker & Brown, 1984; Weller, Repman, Lan, & Rooze, 1995). Application of metacognitive strategies requires training, discipline, and lifelong practice. Training on metacognitive strategies in a computer-based learning environment can help children use computers wisely and study more efficiently (Lee, 1997).

When students activate computer software during their study time, they should have an idea of how that software will facilitate their studying activities. Their time and effort should be invested in activities that help them process what they read in a deep way, rather than in the features of the computer software. When learners are able to monitor their own learning processes at the computer, they can become responsible learners. They are more likely to gain knowledge while working at a computer, and computers become less of a distraction to them during study time (McKeachie, 2000).

Difficulty of the Use of Visualization as a Study Strategy for Middle Schoolers

Middle schoolers are unsophisticated learners and require guidance toward effective visualization. The seventh and eighth grade students who participated in our workshop were unskilled visualizers. They needed cognitive apprenticeship (Collins, Brown, & Newman, 1989) for identifying the underlying structure of concepts and representing them as well as using computer graphics. Our workshop participants were unable to internalize the visualization methods as part of their study strategy. The three days of training did not afford the opportunity to give all students the continuous cycle of visualization, feedback, and revision necessary for constructing valid representations of concepts. Most students were not ready cognitively to generate meaningful computer graphics while they studied the textual information. They were more likely to highlight the important points by typing those words or sentences on a computer than to produce visuals concretizing those concepts. This research indicated that middle school students might need extensive practice in constructing their own concept representations while receiving expert feedback regarding their appropriateness.

Although it may be easy for students to represent one concept described in one sentence, it is much more difficult to represent interrelationships among concepts in a paragraph, and harder still to represent interrelationships among concepts in an essay. Instructors should teach students in the sequence of increasingly complex tasks. When the learning task becomes more and more difficult for students to handle independently, the aids of peers and the close scaffolding of the instructor are of great importance.

We found that students had little trouble representing sequential concepts and concepts that were well described so that they could be represented directly. However, hierarchical, oppositional, and causal relationships were quite difficult for students to identify. Identification of such interrelationships among concepts indicates understanding. We concluded that students need more experiences identifying relationships among concepts in order to build understanding (Perkins & Unger, 1999). This finding supports the theoretical notion that construction of meaning is a lifelong, cognitive, and social process.

CONCLUSIONS

Further investigations into the effects and impact of computer graphics for student-generated visualization on learning should be conducted in schools with more reliable computers and technical support. To minimize the distraction of the computers, the lab in which future studies are conducted might have mouse locks so students cannot play during instruction. This would help the students focus on the materials being studied and on the task of learning how to visualize.

The negative effect of the visualization training in this study and the interference created by computers might be explained by inadequate training. We suspect that to be successful visualizers, students might need more lengthy training than the three class periods of training that they received in this study. We recommend continued development and testing of training that includes several examples of visualization and extensive practice generating visualizations. Students need opportunities to generate visualizations of progressively longer excerpts from texts to approximate the true study environment of school classes that typically involves chapter study.

The visualization training in this study did not interfere with learning for the visualizers; only for the nonvisualizers. Further studies should explore the possible differential effect of visualization across different types of learners. It may be, for instance, that visualization works well for visual learners, but does not work for tactile or auditory learners (Gardner, 1999). In addition, the effectiveness of the use of visualization as a study strategy may differ across developmental stages.

Researchers over decades, such as Lesgold, Levin, Shimron, & Cuttmann (1975), Novak (1990), Peeck (1980), Sodamann (1991) indicated that young children, such as elementary and middle school students, usually have difficulty in spontaneously generating useful images that are relevant to the content, and in representing the conceptual structures adequately even though they are explicitly trained to do so. High school and college students are more likely to have positive learning outcomes as a result of the use of visualization as a study strategy (Cox, Smith, & Rakes, 1994; Lipson, 1994; Schmid & Telaro, 1990). More research should be conducted to investigate the differential impact of visualization on children's learning across various developmental stages.

Although this study was conducted in the natural setting of the school, researchers taught the workshop and administered the test. It is possible that our presence disrupted the natural school setting. In future studies, a trained teacher might deliver the workshop in his or her classroom thereby diminishing the possibility of disruption.

In addition to interaction with mode of learning and developmental stage, culture may play a role in the impact of visualization on learning. Future research should be conducted in different cultural settings to investigate the effect of language structures and social contexts on the learning process of how to visualize, and compare the impact of visualization on science concept learning among different cultural groups. Unlike the participants in this study, certain ethnic groups, such as Chinese learners, who have been highly exposed to pictographic characters, often use imagery in recall and tend to think visually (Sinatra, 1981). Researchers need to explore whether such learners will be more capable than other cultural groups to generate meaningful representations of concepts with or without instruction in visualization.

Recommendations for Educational Practice

Visualization is a factor in scientific understanding (Earnshaw & Wiseman, 1992; Guilford & Hoelfner, 1971; Peltzer, 1988). Given the power of visualization as a study strategy and the lack of visualization skill among learners, we conclude that students need expert modeling (Collins, Brown, & Newman, 1989) of identifying the underlying structure of concepts. The process of translating textual information to graphical representations may be difficult for middle school students to accomplish alone. Adolescents' limited cognitive structures can be effectively supported and expanded through extensive interaction with teachers or more capable peers. With appropriate scaffolding, students should be able to construct their own concept representations and then internalize visualization as part of their repertoire of study strategies (Bruner, 1986; Bliss, Askew, & Macrae, 1996).

Teachers need to provide regular feedback about student-generated representations so students can revise their visuals repeatedly until their visuals match the scientific reality. Jonassen (1999) described the kinds of activities in which students can engage for meaning making. Such activities should be incorporated into curriculum to prepare students for scientific thinking:

    The process for building the ability to model phenomena requires [1]
    defining the model, [2] using the model to understand some
    phenomena, [3] creating a model by representing real world phenomena
    and [4] making connections among its parts, and finally [5]
    analyzing the model for its ability to represent the world. (p. 227)

Given the negative effect of training students to use computer graphics for visualization, one might conclude that students should be discouraged from using computers to visualize during study time. However, we draw the opposite conclusion. Because visualization is so important to scientific processes as described earlier, yet so difficult, students need more rather than fewer opportunities to visualize on computers. They need guidance through the processes of scientific visualization. This study showed that most middle school students are unable to create meaningful representations of concepts on computers. They are quite unskilled at the process of visualization. In addition, our qualitative findings indicate that the interference was due in part to poor facilities and lack of preparedness on the part of the learners. If we can minimize negative environmental factors and provide more time for modeling and practice, perhaps students can learn science concepts as many scientists do, through computer-generated visualization.

We agree with Dewey and others who claimed that learning to use visuals for understanding as well as for generating concepts is an important part of becoming a scientific and creative thinker (Dacy, 1989; de Bono, 1995; Peltzer, 1988; Torrance & Safter, 1999). Therefore, it is essential to require students to engage in the diverse practice of learning how to construct their own concept representations while receiving expert feedback regarding their appropriateness. Indeed, this research indicated that scientific visualization of concepts during study time should be a focus of science curriculum. We believe that such curriculum should begin at an early age, should extend across developmental stages, and should include computers as potential study tools. Curriculum needs to include computer graphics activities from an early age so computers can become perceived by students as tools for learning in addition to their function as toys.

Table 1 Descriptive Statistics for Groups and Learner's Study Strategy

Group         Learner's Study Strategy  Mean   Standard Deviation  N

No-workshop   Nonvisualizers            14.11  5.48                36
              Visualizers               14.44  4.61                09
              Total                     14.18  5.27                45
Workshop      Nonvisualizers             9.62  3.54                21
              Visualizers               14.09  3.39                22
              Total                     11.91  4.10                43
Total         Nonvisualizers            12.46  5.30                57
              Visualizers               14.19  3.71                31
              Total                     13.07  4.85                88

Dependent variable: Posttest scores

Table 2 ANOVA for Effects of Treatment and Applications of Visualization
on Test Scores

Source          Type III    df  Mean       F        Sig.   Cohen's d
                Sum             Square
                of Squares

Corrected
Model             329.043    3    109.681    5.374  .002
Intercept       11776.403    1  11776.403  576.955  .000
Group             101.223    1    101.223    4.959  .029*  .48
Study Strategy     99.542    1     99.542    4.877  .030*  .38
Group * Study
Strategy           73.837    1     73.837    3.617  .061
Error            1714.548   84     20.411
Total           17072.000   88
Corrected
Total            2043.591   87

* significance p [less than or equal to]. 05

References

Anderson-Inman, L. (1992). Computer-supported studying for students with reading and writing difficulties. Reading and Writing Quarterly: Overcoming Learning Disabilities, 8, 317-319.

Anderson-Inman, L., & Ditson, L. (1999). Computer-based concept mapping: A tool for negotiating meaning. Learning and Leading with Technology, 26(8), 6-13.

Anderson-Inman, L., & Knox-Quinn, C., Horney, M. A. (1996). Computer-based study strategies for students with learning disabilities: Individual differences associated with adoption level. Journal of Learning Disabilities, 29(5), 461-484.

Anderson-Inman, L., Redekopp, R., & Adams, V. (1992). Electronic studying: Using computer-based outlining programs as study tools. Reading and Writing Quarterly: Overcoming Learning Disabilities, 8, 337-358.

Anderson-Inman, L., & Zeitz, L. (1993). Computer-based concept mapping: Active studying for active learners. The Computer Teacher, 21(1), 6-8, 10-11.

Baker, L., & Brown, A. L. (1984). Matacognitive skills and reading. In P.D. Pearson (Ed.), Handbook of reading research (pp. 353-394). New York: Longman.

Bliss, J., Askew, M., & Macrae, S. (1996). Effective teaching and learning: Scaffolding revisited. Oxford Review of Education, 22(1), 37-61.

Bransford, J. (2000). How people learn: Brain, mind, experience, and school. Washington, DC: National Academy of Sciences.

Bruner, J. S. (1986). Actual minds, possible worlds. Cambridge: Harvard University Press.

Cifuentes, L. (1992). The effects of instruction in visualization as a study strategy. Unpublished doctoral dissertation, University of North Carolina, Chapel Hill.

Cifuentes, L., & Hsieh, Y.C. (2003a). Visualization for construction of meaning during study time: A qualitative analysis. International Journal of Instructional Media, 30(3).

Cifuentes, L., & Hsieh, Y.C. (2003b). Visualization for construction of meaning during study time: A Quantitative Analysis. International Journal of Instructional Media, 30(4).

Cohen, J. (1965). Some statistical issues in psychological research. In B. B. Wolman (Ed.), Handbook of clinical psychology (pp. 95-121). New York: McGraw-Hill.

Cognition and Technology Group at Vanderbilt (1999). Technology for teaching and learning with understanding. Boston: Houghton Mifflin.

Collins, A., Brown, J. S., & Newman, S. E. (1989). Cognitive apprenticeship: Teaching the crafts of reading, writing, and mathematics. In L. B. Resnick (Eds.), Knowing, learning, and instruction: Essays in honor of Robert Glaser (pp.453-494). Hillsdale, NJ: Lawrence Erlbaum.

Cook, L.K., & Mayer, R. E. (1988). Teaching readers about the structure of scientific text. Journal of Educational Psychology, 80, 448-456.

Cox, G. C., Smith, D. L., & Rakes, T. A. (1994). Enhancing comprehension through the use of visual elaboration strategies. Reading Research and Instruction, 33(3), 157-174.

Dacy, J. (1989). Fundamentals of creative thinking. Lexington, MA: D.C. Health.

De Bono, E. (1995). Mind power. New York: Dorling Kindersky.

Dodge, B. (1998). A taxonomy of information patterns. Retrieved January 29, 2002, from http://projects.edtech.sandi.net/staffdev/tpss98/patterns-taxonomy.html

Downes, T. (2000, April). The computer as a "playable" tool. Paper presented at the annual meeting of the American Educational Research Association, New Orleans, LA.

Driver, R. (1983). The pupil as scientist? Milton Keynes, England: Open University.

Earnshaw, R. A., & Wiseman, N. (1992). An introductory guide to scientific visualization. New York: Springer-Verlag.

Emerson, R. M., Fretz, R. I., & Shaw, L. L. (1995). Writing ethnographic field-notes. Chicago: The University of Chicago Press.

Finke, R. (1990). Creative Imagery: Discoveries and inventions in visualization. Hillsdale, N.J.: Lawrence Erlbaum.

Fosnot, C. T. (1996). Constructivism: Theory, perspectives, and practice. New York: Teachers College Press.

Gall, M. D., Borg, W. R. & Gall, J. P. (1996). Educational research: An introduction. White Planes, NY: Longman.

Gardner, H.E. (1999). Multiple approaches to understandings. In C. M. Reigeluth (Ed.), Instructional-design theories and models: Vol.2. A new paradigm of instructional theory (pp.69-89). Hillsdale, NJ: Lawrence Erlbaum.

Gobert, J. D., & Clement, J. J. (1999). Effects of student-generated diagrams versus student-generated summaries on conceptual understanding of causal and dynamic knowledge in plate tectonics. Journal of Research in Science Teaching, 36(1), 39-53.

Guilford, J. P., & Hoepfner, R. (1971). The analysis of intelligence. New York: Macmillan.

Huck, S. W., & Sandler, H. M. (1979). Rival hypotheses: Alternative interpretations of data based conclusions. New York: Harper & Row.

Jonassen, D. H. (1999). Designing constructivist learning environments. In C. M Reigeluth (Ed.), Instructional-design theories and models: Vol.2. A new paradigm of instructional theory (pp.215-239). Hillsdale, NJ: Lawrence Erlbaum.

Jonassen, D. H. (2000). Computers as mindtools for schools: Engaging critical thinking. Upper Saddle River, NJ: Merrill.

Julyan, C., & Duckworth, E. (1996). A constructivist perspective on teaching and learning science. In C. T. Fosnot (Ed.), Constructivism: Theory, perspectives, and practice (pp.55-72). New York: Teachers College Press.

Lee, P-L. H. (1997). Integrating concept mapping and metacognitive methods in a hypermedia environment for learning science. Unpublished doctoral dissertation, Purdue University.

Lesgold, A. M., Levin, J.R., Shimron, J., & Cuttmann, J. (1975). Pictures and young children's learning from oral prose. Journal of Educational Psychology, 67(5), 636-642.

Lipson, M. (1994). Effects of a mnemonic imagery strategy on the prose recall of developmental and nondevelopmental college readers. Reading Improvement, 31(1), 9-13.

Mann, L. (1997). Image processing for unlimited exploration. Curriculum/Technology Quarterly, 6(2), 1.

Marzano, R. J. (1988). Dimensions of thinking: A framework for curriculum and instruction. Alexandria, VA: Association for Supervision and Curriculum Development.

Maton, A. (1997). Prentice hall exploring physical science. (Teacher edition) Englewood Cliffs, NJ: Prentice Hall.

McKeachie, W. J. (2000). Helping students learn how to learn. (ERIC Document Reproduction Service No. ED 450 864)

Novak, J. D., (1990). Concept mapping: A useful tool for science education. Journal of Research in Science Teaching, 27(10), 937-949.

Papert, S., & Harel, I. (1991). Constructivism: Research reports and essays, 1985-1990. Epistemology and Learning Research Group, the Media Laboratory, Massachusetts Institute of Technology. Norwood, NJ: Ablex.

Peeck, J. (1980, April). Experimenter-provided and learner-generated pictures in learning from text. Paper presented at the annual meeting of the American Educational Research Association, Boston.

Peltzer, A. (1988). The intellectual factors believed by physicists to be most important to physics students. Journal of Research in Science Teaching, 25(9), 721-731.

Perkins, D. N., & Unger, C. (1999). Teaching and learning to understanding. In C. M Reigeluth (Eds.), Instructional-design theories and models: Vol.2. A new paradigm of instructional theory (pp.91-114). Hillsdale, NJ: Lawrence Erlbaum.

Schmid, R. F., & Telaro, G. (1990). Concept mapping as an instructional strategy for high school biology. Journal of Educational Research, 84, 78-85.

Shulman, L.S. (1997). Disciplines of inquiry in education: A new overview. In R.M. Jaeger (Ed.), Complementary methods for research in education (pp. 3-29). Washington, DC: American Educational Research Association.

Sinatra, R. (1981). Using visuals to help the second language learners. The Reading Teacher, 5, 539-546.

Sodamann, P. (1991). Visual imagery and achievement test outcomes. International Journal of Instructional Media, 18(2), 175-182.

Tenny, J. L. (1992). Computer-supported study strategies for purple people. Reading and Writing Quarterly: Overcoming Learning Disabilities, 8, 359-377.

Torrance, E. P., & Safter, H. T. (1999). Making the creative leap beyond ... Buffalo, NY: Creative Education Foundation Press.

Weller, H. G., Repman, J., Lan, W., & Rooze, G. (1995). Improving the effectiveness of learning through hypermedia-based instruction: The importance of learner characteristic. Computers in Human Behavior. 11(3/4), 451-465.

Wileman, R. E. (1993). Visual communicating. Englewood Cliffs, NJ: Educational Technology.

Williamson, J. W., Jr. (1999). Mental models of teaching: Case study of selected preservice teachers enrolled in an introductory educational technology course. Unpublished doctoral dissertation. University of Georgia.

LAUREN CIFUENTES

Texas A & M University

USA

[e-mail]

YI-CHUAN JANE HSIEH

Texas A & M University

USA

[e-mail]