Effects of learner control and hypermedia preference on cyber-students performance in a Web-based learning environment.
As expected, students' task performance is mainly affected by learner control (with/without advisement). However, although students' hypermedia preference has no direct relation to their performance in WBLE, when hypermedia preference works together with learner control, there is a significant joint influence on students' task performance. Students with a low hypermedia preference level benefited significantly from the presence of learner control with advisement. Students with a high hypermedia preference level were not significantly affected by the presence or absence of learner control with advisement.
There is a close kinship among hypertext, hypermedia, and the World Wide Web (WWW or Web). The Web has its roots in hypertext. In 1945, Vannevar Bush first envisioned a hypothetical device called a "memex" that would allow people to select by association and automatically access notes, photographs, memoranda, books, records, and communications stored in a dispersed fashion (Bush, 1945). Although this memex machine was never created and the words "hypertext" and "hypermedia" lay decades in the future as he worked, his visionary image eventually evolved into today's Web.
Tim Berners-Lee (1996, 1998) was ultimately responsible for bringing Bush's idea into the modern world. In 1990, he wrote a program called "WorlDwidEweb," a point-click hypertext "wysiwyg" (what you see is what you get) editor which ran on the "NeXTStep" machine. In 1991, he presented to the public his first Web server (info.cem.ch, now replaced by www.w3.org), which stored nodes of information that could be accessed by nonlinear hyperlinks. Thus, with the Web, a user could use any hypertext link to point to anything, whether personal, local, or global, in draft or final form. The Web has since become so pervasive that it is indeed a reflection of our concrete world today, in both personal and cooperative work, recreation, and socializing.
Without a doubt, the concept of hypertext (or hypermedia) is the core of the Web. The original concept of nodal capacity in hypertext only stored information in text format. Later on, it expanded its nodal capacities to allow designers to store and access "non-text" forms of data in audio and video representation in a non-sequential (i.e., hyper) fashion. Thus, the term "hypermedia" evolved, referring to allowing users to access other media than text in a "hyperarchical" fashion. In fact, in 1988, Marchionini portrayed hypermedia as a "nonlinear representation" of hypertext which allowed users "access to graphics, sound, animation, and other forms of information transfer." While the terms hypertext, hypermedia, hyperdocument, and/or multimedia appear throughout the literature with varying meanings and connotations, Altun (2000) noted that nodes and nonlinear linkage are the core of all definitions. In fact, hypertext and hypermedia, hypermedia and multimedia have been used interchangeably for years. Hypermedi a is preferred in this study.
What is a Web-based Learning Environment (WBLE)?
Five major functions of electronic media are communication, entertainment, motivation, education (Romiszowski, 1994), and storage. Today's Web is fulfilling these purposes. The Web carries an increasingly heavy duty as a collection of media for teaching and learning, and Web usage is a daily commodity for many educators and learners. Web-based instruction (Khan, 1997) and web-based learning (Hackbarth, 1997; Stan, 1997; Windschitl, 1998) are growing exponentially. Some researchers have reported that students in a web-based course outperform students in a traditional classroom (e.g., Liou, 1997; Jones, 1999; White, 1999). With more and more teaching and learning occurring on the Web, the web-based Learning Environment (WBLE) is developing associatively. What is a WBLE?
Because hypertext and its descendant hypermedia are the core of the Web, WBLE was developed based on the notion of a hypertext learning environment. What is a hypertext learning environment? Marchionini (1988) described the hypertext learning environment as a self-directed, information-fluid environment with high teacher/learner interaction. Heller (1990) emphasized that it is incidental (unplanned learning acquired from the learning environment) and discovered (full of opportunities for exploration). Shin, Schallert, and Savenye (1994) further added that it is a learner-activated, self-motivated, self-directed, non-sequential, dynamic, and multipath environment. Sweeters (1994) concurred that it offers a dynamic, exciting, and powerful learning environment. Altun (2000) summarized that it is a "nonlinear, electronic (i.e., digitized), multimedia, hypermedia, and interactive environment." Khan (1997) described web-based instruction learning environments as including the following key features of interaction: multimedia, device/distance/time independence, global accessibility, world-wide uniformity, online resources, cross-cultural interaction, multiple expertise, learner-control, convenience, authenticity, nondiscriminatory, cost-effectiveness, collaborative learning, formal and informal environments, and virtual cultures. Ultimately, the WBLE is an open-ended environment in which any digitized multiple-media forms of information are stored in powerful information units (nodes, web pages, and/or server). The information can then be flexibly accessed electronically and globally through the multiple links in a hyperarchical interactive fashion by any appropriately connected users at any time.
WBLE, Learner Control, and Hypermedia
As the literature review implies, a WBLE is itself an open-ended learning environment. In 1994, Hannafin, Hall, Land, and Hill stated that learners situated in an open-ended learning environment can determine what is to be learned, how it is to be learned, in what sequence, and when the learning goal has been reached. As evidenced, the WBLE is a constructivist learner-centered environment in which learners have choices to make about their own flow of learning. They have full control of how, where, what, and when to access and use the given information. This notion matches the concept of learner control. That is, learner control plays a key role for learner performance in the WBLE.
Hypermedia, being central to the Web, is part of the process for efficient learning in the WBLE. Web-based instruction today is designed with more advanced hypermedia features (streaming video, Virtual Reality, MP3 music, etc.) in addition to the traditional hypermedia elements of graphics, sound, and animation and (with improving connectivity) can better be accessed than before. Among these hypermedia elements, images, animation, sound, and digital video were used to design this study. To measure learners' task efficiency in the WBLE, possible characteristics of learning in hypermedia should be considered. For example, Marchionini (1988) claimed that hypermedia provided learners with a great opportunity for learning "how to learn." Hypermedia can force learners to decide the next step, to evaluate their own progress, and to make their own decisions in what they want to learn. Achievement of these points requires higher-order thinking skills and a high level of learning control from learners. Hannafm, Hall, Land, and Hill (1994) further stated that noneffective learner control causing learner disorientation worsens the learning effectiveness of the traditional CAI hypermedia environment. Hence, learner control is a key facet of hypermedia learning. Ultimately, hypermedia characteristics (thinking skills, learner control, disoriented navigation, etc.) are important factors that need to be considered for successful learning outcomes in the WBLE. Effective learner control is of the essence for learner achievement in both hypermedia learning and WBLE. What is learner control?
Learner Control and Learner Control With Advisement
Learner control can be broadly defined as the degree to which a learner can direct his or her own learning process (Milheim & Azbell, 1988).
That is, the learner has choices to make in his or her own flow of learning. Goforth (1994) further contended that learner control equals decision-making plus information. What decisions can a learner make during the presentation of instructional information? What should be controlled? Reigeluth and Stein (1983) noted that according to Merrill's report in 1980, the learner has freedom to take command of the selection and sequencing, which are: (a) pace control (the rate at which a learner will learn), (b) display control (the order in which a learner will learn), (c) content control (the material a learner will learn), and (d) conscious cognitive control (the cognitive strategies a learner employs in learning). Meanwhile, the learner can make decisions to control feedback in instruction as well (Friend & Cole, 1990). Shyu and Brown (1992) added the variables of difficulty and amount of practice. Arnone, Grabowski, and Rynd (1994) added remediation and other choices, such as exiting and reaction to advisement.
In summary, a learner can possess decision-making capacity for instructional pace (also called speed and rate), sequence (also called display, order, and path), content, cognitive strategy, feedback, difficulty, amount of practice, remediation, choice of exiting, and reaction to advisement. Among these variables, the three essential ones for which a learner can make decisions are pace, sequence, and content. These three learner control elements are heavily implemented in the design of this study.
What are the effects of learner control on the effectiveness of learning? The results of learner control researches in CAI have been somewhat contradictory. Some researches conducted by Papert (1980); Merrill (1980); Gay (1986); Steinberg (1989); Dalton (1990); Shyu and Brown (1992); and Reeves (1993) showed increased effectiveness of learning with learner control. Conversely, other research studied by Atkinson (1972); Fry (1972); Fisher, Blackwell, Garcia, and Greene (1975); Gray (1987); and Morrison, Ross, and Baldwin (1992) showed decreased effectiveness of learning with learner control. Yet some studies in Carrier, Davidson, Higson, and Williams (1984); and Goetzfried and Hannafin (1985) showed no difference with learner control.
One reason for the inconsistency in research findings on the effectiveness of learner control might be that learner control is not a single construct, but rather many factors that function together to make it effective. Learner control may be influenced by the types of learners being granted control and the conditions under which it is being granted (Goforth, 1994).
What types of learners should be granted control? The effects of learner control may vary because of learners' individual differences such as age (e.g., Steinberg, 1989; Arnone & Grabowski, 1992; Arnone, Grabowski, & Rynd, 1994), gender (e.g., Dalton, 1990; Campbell, 2000), prior knowledge (e.g., Fry, 1972; Gay, 1986; Steinberg, 1989; Shyu & Brown, 1992; Shin, Schallert, & Savenye, 1994), cognitive style (e.g., Carrier, Davidson, Higson, & Williams, 1984), curiosity (e.g., Arnone, Grabowski, & Rynd, 1994), and preference (e.g., Carrier, Davidson, Higson, & Williams, 1984; Freitag & Sullivan, 1995).
Under what conditions should a learner be granted control to achieve optimal effectiveness in learning? Some researchers have combined aspects from both learner control and program control and suggested an answer by providing a latitude (Reeves, 1993) within a "learner control continuum" (Milheim & Martin, 1991). There are two extremes in this continuum, one end having total learner control and the other end having total program control. Learners in the former extreme will encounter the dangers of lack of motivation and disinterest while learners in the latter extreme will encounter the danger of anxiety, avoidance, disinterest, and inefficiency (Day 1982; Arnone, Grabowski, & Rynd, 1994). These researchers have produced an alternative moderate formula of learner control, called "learner control with advisement," in which learner control is placed in the middle of this continuum, thereby using learner control in combination with program control. For the purposes of this study, "advisement" will refer to the presence of a limited degree of program control (in the form of access to an online glossary) combined deliberately with a corresponding degree of learner control.
The effectiveness of learner control with advisement has been under study for over two decades. Learner control with advisement has been found to be effective in many studies. For example, in an experiment using a CAL system, Minnesota Adaptive Instructional System (MAIS), Tennyson (1981) found that learner control with advisement can successfully keep students actively involved in the management and assessment of their learning, and thus keep the students on the task until they master the learning. Furthermore, in an experiment involving 101 first and second grade students who attended a public elementary school in upstate New York, by using an interactive video lesson of a visit to an art museum, Arnone and Grabowski (1992) found that students having learner control with advisement achieved the greatest amount of learning and scored higher than the learner control without advisement group did. Learner control with advisement also provided the most chance for interaction. Likewise, in investigating the moti vational and instructional effects of advisement in computer-based instruction for at-risk high school students, Clariana (1993) stated that females with advisement outperform both females without advisement and males, with or without advisement. Moreover, Arnone, Grabowski, and Rynd (1994) also found that all children in their sample, regardless of curiosity level, perform better in the advisement condition than in the no-advisement condition. In addition, Shin, Schallert, and Savenye (1994) indicated that learner control becomes more effective in instruction when advisement is provided.
In summary, the studies on learner control with advisement in CAI have reported that students reach mastery and achieve a higher score than a learner control group without advisement. Learner control with advisement could provide a moderate and effective method to help a learner achieve optimal learning. As such, the strategy of learner control with advisement is the preferred implementation in this study because it is an effort to take advantage of the best features of both program control and learner control. It can remedy problems inherent in environments of total learner control or total program control. Therefore, learner control with advisement is used as the experimental treatment in this study.
Purpose of the Study and Research Questions
As the literature review makes clear, hypermedia is central to the Web, and thus to a WBLE. In addition, learner control is an integral design component affecting learner achievement with both hypermedia learning environments and WBLE. Are the inconsistent results which characterize learner control studies in CAI also true in a WBLE? If this is the case, could some of the variability be accounted for by other factors, such as the learner's preference for this particular learning environment? Freitag and Sullivan (1995) began to explore this possibility with CAI, and Altun (2000) indicated its possible application to a hypertextual environment.
The purpose of this study was to investigate whether or not the factor of "hypermedia preference" in the WBLE affects cyber-students' task performance under "learner control with and without advisement" conditions (for the purposes of this study, cyber-students are defined as students learning exclusively through the Web.) Four questions are studied: (a) Can cyber-students test scores be affected by the treatment (learner control with/without advisement), by the hypermedia preference of the student (high/low), or the interaction of the treatment and preference in WBLE? If so, (b) Is the main effect of treatment (learner control with/without advisement) significant for cyber-students' task performance in WBLE? (c) Is the main effect of high/low hypermedia preference level significant for the cyber-students' task performance in WBLE? and (d) Is there a significant interaction effect between "learner control with/without advisement" treatment and "high/low hypermedia preference" in terms of cyber-students' task performance in WBLE?
In this study, a web-based lesson, "American shopping malls," was created as the medium to deliver the instruction. Every subject had equal access to the hypermedia elements. All subjects were "cyber-students." The web-based Glossary used in this study was created as a treatment to differentiate the experimental group (learner control with advisement) from the control group (learner control without advisement). The Glossary was provided only to the experimental group; that is, "advisement" in this study consisted of open access to the Glossary. Data were analyzed using factorial analysis of variance (two-way ANOVA) and t-tests.
Subjects were 81 undergraduate volunteers from two sections of the computer education course Application of Instructional Technology in Special Education at National Taiwan Normal University, a leading teacher education university, in Taipei, Taiwan. The sample was comprised of a mix of preservice and inservice teachers. The inservice teachers already held associate degrees. The 11 males and 70 females ranged in age for 20 to 48 years, with a mean age of 25 years. The substantial dominance of females over males in this sample is typical of teacher education programs in both Taiwan and the United States. Mean self-reported English score in the National Collegiate Entrance Exam (NCEE, similar to the SAT in the USA) for this group was 57, with a minimum of 15 and a maximum of 95 points. Possible scores on NCEE range from 0 to 100, and scores indicate percent correct. Everyone had used a computer before with an average of three years of usage experience. Of the 81, only nine subjects had been to the United States, with medium length of stay about two weeks. The rest had no experience in the United States. In other words, this was a sample with limited English proficiency.
Subjects were assigned to treatment using stratified random sampling (Kerlinger, 1986). For each class section, the mean of the students' self-reported NCEE English scores for that section was used as a cut point to divide the students into high and low English skill group. For section one, the mean (n = 38) was 54.2; for section two, the mean (n = 43) was 59.9. Each subject in each one of the four groups then was randomly assigned a number generated by Microsoft Excel. Based on the random number, about half of the subjects in each group were assigned to the experimental group and half to the control group. Consequently, there were 41 subjects assigned to the experimental group and 40 subjects assigned to the control group. Heterogeneity was thus created among group subjects with respect to English skill level.
Hypermedia lesson. Cyber-students were provided with the lesson "American shopping malls." The hypermedia was designed to be attractive and provide additional relevant contextual information. To increase the validity of this study, four pilot studies were conducted before the experiment. These prior tests indicated that "American shopping malls" was not a topic familiar to college students in Taiwan and that they would prefer to learn about malls rather than about American national parks, American sports, American music, American holidays, or American garage sales. The text, at less than two pages long with 670 words, was designed for English-as-a-second-language students to read completely and effectively in 40 minutes.
The design of this lesson was aimed at creating a "virtual immersion" in an English-learning environment, with the aspect of American culture attractively presented as hypermedia elements in images (pictures or graphics), sound, digital video, and animation. To create a lively, interactive hypermedia learning environment, three American shopping malls and one neighborhood were visited to create these hypermedia materials. Consequently, the author produced four rolls of film for image development, 61 audio sound files, and 104 digital video files for use in the study. However, due to the consideration of the speed limitation of downloading hypermedia elements from a Web site, only 10 image files (gift box, Halloween, masquerade, trick-or-treating, gingerbread house models, Christmas displays, Sears, Kaufmann's, J.C. Penney, and Dillard's), four sound files (Oh! Susannah; Dixie; U.S. Flag; and American shopping malls), and three digital videos (Title of the Shopping Mall, running 27.7 seconds with a file size o f 2,811 KB; Parking Lot, running 20.9 seconds with a file size of 3,484 KB; and Hard to Find A Place to Park running 13 seconds with a file size of 1,625 KB) were implemented in the study. In addition, animations in the five varying marquees in the web page symbolized the multiple facets of American culture, and a colorful but light gift box background was designed and used. Additionally, a poetic couplet in Chinese, written by a relative of the author, was used in designing the background of the survey web pages. The purpose was to provide a more intimate and familiar environment for the subjects so that they might pay more attention. Hyperlinks were created and allowed every subject to access the hypermedia elements. Explanations were provided to accompany the hypermedia elements (Figure 1).
Glossary. A glossary of 63 entries was designed with a mix of vocabulary (e.g., masquerade, gingerbread, trick-or-treating, etc.), slang (e.g., to yak on, gotta-go, an ear-piercing scream, etc.), and figures of speech (e.g., swim and whistle "Dixie," in the whole deal, to make one's round, etc.). A beta test from a different group of subjects with similar backgrounds indicated that these entries were unknown words and phrases that cyber-students in Taiwan wanted to learn. The software Random House Webster's Electronic Dictionary and Thesaurus, College Edition, (1994). Version 1.5 by WordPerfect Corporation was used in constructing some of the content of the Glossary.
Underlined vocabulary entries were hyperlinks to the Glossary. When a subject clicked a vocabulary link in the lesson, the vocabulary explanation was shown at the top of the Web page in the list of the glossary. Thus, the design helped cyber-students reduce the degree of possible disorientation in the WBLE.
Pretest and posttest. Based on the glossary, 10 identical multiple-choice pretest and posttest questions were designed. The balance of degree of difficulty was taken into account in designing the test questions.
Additionally, to increase the effectiveness of the experiment, the direction "It is alright if you don't know the right answers now" for the pretest was designed to make the subject feel more comfortable while answering the questions. In the directions of the posttest, "Remember! You can send your answers only once" and "Before you click the button 'Send Answers,' you can correct all the answers if it is necessary" were used as reminders. "Please do not use a dictionary" was indicated on both tests as well.
Survey. Survey questionnaires were designed to measure a cyber-student's hypermedia preference for images (pictures or graphics), animation, sound, digital video, hyperlinks, and the nonlinear capacity of hypertext in a WBLE. For example, "In this lesson, I like the digital video 'Parking Lot' because it helped me to understand better what an American 'parking lot' is;" and "In this lesson, the navigation between hyperlinks allows me to go back and forth as much as I want, I like this kind of characteristic because it gives me the chance to have freedom (i.e., flexibility) in navigating between the hyperlinks."
In summary, these materials were composed of a pretest, the hypermedia format lesson "American shopping malls," a glossary, a posttest, and survey. Hyperlinks were used to navigate these materials, with the only required tool being a standard web browser. All the materials and survey data including information sent and received were stored in a mirror site web server, http://icon.ntnu.edu.tw:2369/ that existed in Taiwan to alleviate heavy Internet traffic between the United States and Taiwan.
One treatment with two levels was used in this study: learner control with advisement (experimental group) and learner control without advisement (control group). The difference was that the subjects in the experimental group could choose to consult the Glossary from the lesson "American shopping malls" while the control group could not. That is, subjects in the "learner control with advisement" group had full freedom to select any vocabulary they wanted to learn at any pace and in any sequence. They could read the information provided in the Glossary for as long a time as they desired until they mastered the learning. They could also skip any vocabulary they already knew.
However, within the same lesson, all the subjects could control the learning pace, choose the sequence of learning, select the presentation of learning material, and navigate back and forth between the hypermedia elements as many times as desired. To navigate the elements, subjects used hyperlinks provided in the lesson or by the Web browser (e.g., the "Back" button).
The study was conducted during a three-week period to resolve geographical constraints. Two stages with a total of six steps were conducted. Stage one included a survey on personal information (10 minutes), a pretest (20 minutes), and a survey on prior knowledge (20 minutes) while stage two consisted of the lesson (40 minutes), a posttest (20 minutes), and a wrap-up survey (40 minutes). A two-week delay was incorporated between stages to minimize participant retention of the questions in the pretest. Each subject received a brief introduction prior to each stage. The contents of each stage were posted on the web site only during the specific experimental week. An easy point-click mouse function was mainly used by the subjects for answer selection, submission, and web page navigation.
The study was completed remotely in a computer lab at the National Taiwan Normal University. The lab had 50 multimedia personal computers with an Internet connection and web browser software. A responsible faculty member of NTNU with advanced computer experience was chosen to monitor the process in Taiwan.
Before the experiment, by using a computer video projector, the course instructor explained to the subjects on the screen and in Chinese what they should expect in the experiment. A password and ID were distributed to subjects to login throughout the course of the study.
In stage one, each subject was asked to login and key in answers to the questionnaires of "Personal Information" (Step 1-1), "Pretest" (Step 1-2), and "Prior Knowledge" (Step 1-3). The computer program guided the subjects through the steps in sequence.
Two weeks later, the subjects studied the lesson "American shopping malls" (Step 2-1). In the process of reading the lesson, the subjects in the experimental group could further choose to consult the section Glossary while the control group could not. After reading the article, each subject immediately took the follow-up posttest (Step 2-2) of 10 multiple-choice questions. After taking a break, they took a wrap-up survey (Step 2-3).
After submitting the pretest or posttest answers, subjects instantly received their individual score report from the computer program. A warning message (e.g., "you have already taken this test. Here is your test information...") signaled previously submitted answers, thus screening out redundant answers and surveying.
The procedure occurred and was administered online remotely at NTNU. The researcher never met with the subjects face-to-face.
The mean difference between pretest and posttest was used as a dependent measure to assess cyber-students' performance in WBLE. The pretest and posttest were identical. Each test was administered online and consisted of 10 multiple-choice questions with three response choices for each item. To protect the study's internal validity, subjects were consistently given 20 minutes to complete each test. Time remaining was displayed in the lower left-hand corner of the testing web pages. Additionally, subjects were asked to not use any dictionaries during the tests. Each question was worth 10 points. The internal consistency of Cronbach's alpha reliability for the test instrument was determined to be .74. Test questions were extensively field tested.
Every subject had an equal chance to access all the hypermedia elements. Hypermedia preference for the study was assessed with 11 items on a student attitude survey administered online. A five-point Likert-type response format was employed in the survey (5 = strongly agree, 4 = agree, 3 = neutral, 2 = disagree, 1 = strongly disagree, respectively). The inter-item reliability of Cronbach's coefficient alpha on hypermedia preference was determined to be .86.
The computer program recorded all the data while the subjects were using the learning materials. The data included user identification, password, login and out time and validation, the answers a subject submitted in all the surveys and tests (in all six steps), and score reports for the pre and posttest that each subject took. All data were collected at the mirror site in Taiwan at http://icon.ntnu.edu.tw: 2369/. Then all data were transferred using FTP to the computer in the U.S.A. for storage and analysis. Data were imported into Microsoft Excel by using the worksheet function TRANSPOSE.
Microsoft Excel and SPSS were used to analyze the data. The mean score (3.92) of 81 subjects resulting from the eleven-question hypermedia preference survey was used as a cutpoint to classify cyber-students as "high" or "low" in hypermedia preference. The study used a 2 X 2 factorial design with two levels of learner control (with and without advisement) and two levels of hypermedia preference (high and low). A two-way analysis of variance (ANOVA) was used to test for main effects and the interaction assumption between conditions of control and hypermedia preference on means difference between posttest and pretest scores. A follow-up t-test was further used to investigate the significance of the degree of the disordinal interaction between treatment and hypermedia preference.
Two-way full factorial model ANOVA results revealed a significant effect for the overall model using treatment (learner control with/without advisement) and hypermedia preference (high/low), F(3, 77) = 4.91, p< .01.
The results of the ANOVA also indicated significance in main effect for treatment, F(l, 77) = 5.87, p < .05. It was as expected that cyber-students' task performance is mainly affected by the treatment of learner control (with/without advisement) in the WBLE.
On the contrary, the main effect for hypermedia preference was not significant at the .05 level, F(1, 77) = 0.02, p = 0.88. This finding comes as a surprise. For the cyber-students, the preference toward hypermedia elements has no direct relation with their performance in present study.
However, Figure 2 shows the interaction between treatment and hypermedia preference was statistically significant, F(1, 77) = 8.30, p< .01. It was significant to find that cyber-students' task performance is affected by the joint influence of the treatment of learner control (with/without advisement) and the factor of hypermedia preference. The low hypermedia preference group (n = 42) has a higher mean value of 22.00 under learner control with advisement than the mean value of 1.82 under learner control without advisement. Meanwhile, the high hypermedia preference group (n = 39) has a lower mean value of 10.48 under learner control with advisement than the mean value of 12.22 under learner control without advisement. Thus, the interaction is disordinal.
Since in the group of high hypermedia preference, the mean value of the "with" advisement group (i.e., 10.48) is lower than that of the "without" advisement group (i.e., 12.22), it may be that the treatment has a negative effect on high-preference cyber-students. What is the import of the degree of the disordinal interaction between treatment and hypermedia preference? Follow-up t-tests were used to investigate the significant disordinal interaction. For the low hypermedia preference group, Levene's test for Equality of Variances between the two advisement condition (with and without) groups yielded an F(27.52) of 5.13, p< .05, so equal variances could not be assumed. Then the two-tailed t(27.52) = 4.09, p<.001. This is highly significant. This further supported that in the low hypermedia preference group, learner control with advisement condition does benefit cyber-students more than without advisement. Meanwhile, for the high hypermedia preference group, Levene's test also yielded significant results, F(32.30) = 7.43, p< .05, so equal variances could not be assumed here either. The two-tailed t(32.30) = .30, p = .76, which is not significant at the .05 level. The result revealed that for the high hypermedia preference group, the negative effect of treatment was insignificant. In other words, there is no counter effect resulting from the treatment for the cyber-students who were in the high hypermedia preference group.
CONCLUSIONS AND DISCUSSION
Cyber-students' task performance was not directly affected by their preference toward hypermedia elements in the present study. The outcomes may depend on the natural complexity of the hypertext learning environment itself. Early in 1988, Marchionini stated that several problems may occur in the self-directed hypertext learning environment. These problems include disorientation; distraction; requirement of a high level of learner control, cognitive-overhead, and an increasing navigational decision-making load; confusion and dizziness from the richness of information; and slow processors and inadequate storage for rapid execution of links, high-resolution graphics, and sounds. These problems may decrease learners' critical learning. Recently, more learners are becoming confused by complicated web page designs that are associated with a variety of complex hypermedia elements. It is hard for some novices to use the Web appropriately. Consequently, Altun (2000) pointed out that reading in a hypertext learning en vironment involved various strategies and complex cognitive activities. For example, expert computer users have a tendency to use their own strategies (e.g., opening multi-layered windows to avoid delay for downloading hypermedia elements; using embedded browser's navigation buttons for fast navigation; printing out some pages for later reading) to control their own learning instead of always following the given links. Thus, higher order thinking skills are more involved. Web page designers should pay more attention to user expertise, personal interest, node content, and navigation design. For example, Kacmar and Carlson (1999) suggested that link design with color is an effective style. In addition, when the color is associated with fully bound objects in italics, boxes, shading, and "<>" characters, those links are most preferred by college students.
Among the cyber-students with a low level of hypermedia preference, those in the learner control with advisement group (those having access to the "Glossary") significantly outperformed those in the learner control without advisement group. This result concurred with the previous effective learner control with advisement literature (e. g., Tennyson, 1981; Arnone & Grabowski, 1992; Shyu & Brown, 1992; Clariana, 1993; Arnone, Grabowski, & Rynd, 1994; Shin, Schallert, & Savenye, 1994; Campbell, 2000). Learner control with advisement is a critical factor in promoting task performance in the WBLE, especially for the English-as-a-second-language cyber-students with low hypermedia preference in present study.
However, among cyber-students with a high level of hypermedia preference, the presence or absence of learner control with advisement neither affects nor harms their task performance in the WBLE in present study. It was surprising to find that the factor of hypermedia preference in itself is not significant as a main effect on cyber-students' task performance in the WBLE. However, when it works together with the learner control treatment, there is a significant joint influence on cyber-students' task performance in the WBLE. Interestingly, cyber-students with a low level of hypermedia preference significantly benefited from the treatment of learner control with advisement. Cyber-students with a high level of hypermedia preference did not benefit nor become harmed when participating in the treatment of this study. Based on the results of this study, one could clearly recommend that web-based instructional designers make their first priority the quality of learning content rather than the cosmetic details of th e web pages involved. The effectiveness of the WBLE will depend upon the extent to which hypermedia elements can be associated with other factors.
The study would appear to indicate two important principles for educators who would teach in a WBLE, though it may be ill-advised to generalize beyond this population. The first is that if the intended audience consists of students who have little experience in a WBLE, and/or a preference for other types of learning environments, it is critical that the instructor provide a high level of what has been termed in this study "learner control with advisement." In practice, this would take the form of navigational assistance, glossary links, and other low-level structural elements which make it particularly easy for the novice to remain well oriented and ontask. On the other hand, if the intended audience consists of WBLE "veterans," this study suggests that beyond the basic elements of good structure, attention paid to providing such assistance may in fact be largely wasted, and that the attention of the instructor in such a case would far better be focused on the instructional content itself rather than the str ucture in which it is embedded.
FURTHER RESEARCH SUGGESTIONS
While the use of video, sound, and animation was restricted in this experiment by bandwidth concerns, some cyber-students specifically requested these components. One area of needed research would investigate the role of increasing levels of such multimedia components on student achievement, possibly in combination with the learner control and advisement variables investigated in this study. Do students actually learn better when presented with higher levels of multimedia in a WBLE? Does it matter in such cases whether they have an initial preference for the WBLE, or to what extent they are granted learner control?
In addition, the effects of learner control may be influenced by cyber-students' individual differences, such as age, gender, prior knowledge, cognitive style, and curiosity. To what extent do these attributes, singly or jointly, predict cyber-students task performance in the WBLE, under different learner control conditions? Further research to investigate this question is also indicated.
The potential of the Web in the area of teaching and learning is only beginning to be explored. The findings of this study provide one small portion of the knowledge which will help us to fulfill this potential. It is our hope that the suggestions for future research mentioned above will encourage other interested educational researchers to join in this exploration of the full potential of multimedia/hypermedia pedagogy within the web-based learning environment.
[FIGURE 2 OMITTED]
The authors would like to thank Lida Allen for statistical support in this study.
Altun, A. (2000). Patterns in cognitive processes and strategies in hypertext reading: A case study of two experienced computer users. Journal of Educational Multimedia and Hypermedia, 9(1), 35-55.
Arnone, M.P., & Grabowski, B.L. (1992). Effects on children's achievement and curiosity of variations in learner control over an interactive video lesson. Educational Technology Research & Development, 40(1), 15-27.
Arnone, M.P., Grabowski, B.L., & Rynd, C.P. (1994). Curiosity as a personality variable influencing learning in a learner controlled lesson with and without Advisement. Educational Technology Research & Development, 42(1), 5-20.
Atkinson, R.C. (1972). Optimizing the learning of a second-language vocabulary. Journal of experimental psychology, 96(1), 124-129.
Berners-Lee, T. (1996). The Web: Past, present and future. [Online]. Available: http://www.w3.org/People/Berners-Lee/1996/ppf.html
Berners-Lee, T. (1998). The Web: A very short personal history. [Online]. Available: http://www.w3.org/People/Berners-Lee/ShortHistory.html
Bush, V. (1945). As we may think. The Atlantic Monthly, 176(1), 101-108.
Campbell, K. (2000). Gender and educational technologies: Relational frameworks for learning design. Journal of Educational Multimedia and Hypermedia, 9(2), 131-149.
Carrier, C., Davidson, G., Higson, V., & Williams, M. (1984). Selection of options by field independent and dependent children in a computer-based concept lesson. Journal of Computer-Based Instruction, 11(2), 49-54.
Clariana, R.B. (1993). The motivational effect of advisement on attendance and achievement in computer-based instruction. Journal of Computer-Based Instruction. 20(2), 47-51.
Day, H.I. (1982). Curiosity and the interested explorer. NSPI Journal, May, 19-22.
Dalton, D.W. (1990). The effects of cooperative learning strategies on achievement and attitudes during interactive video. Journal of Computer-Based Instruction, 17(1), 8-16.
Fisher, M.D., Blackwell, L., Garcia, A.B., & Greene, J.C. (1975). Effects of student control and choice on engagement in a CAI arithmetic task in a low-income school. Journal of Educational Psychology, 67, 776-783.
Freitag, E., & Sullivan, H. (1995). Matching learner preference to amount of instruction: An alternative form of learner control. Educational Technology Research & Development, 43(2), 5-14.
Friend, C.L., & Cole, C.L. (1990). Learner control in computer-based instruction: A Current Literature Review. Educational Technology, 30(11), 47-49.
Fry, J. P. (1972). Interactive relationship between inquisitiveness and student control of instruction. Journal of Educational Psychology, 63(5), 459-465.
Gay, G. (1986). Interaction of learner control and prior understanding in computer-assisted video instruction. Journal of Educational Psychology. 78(3), 225-227.
Goetzfried, L., & Hannafin, M.J. (1985). The effect of the locus of CAI control strategies on the learning of mathematics rules. American Educational Research Journal, 22(2), 273-278.
Goforth, D. (1994). Learner control = Decision making + information: A model and meta-analysis. Journal of Educational Computing Research, 11(1), 1-26.
Gray, S.H. (1987). The effect of sequence control on computer-assisted learning. Journal of Computer-Based Instruction, 14(2), 54-56.
Hackbarth, S. (1997). Introduction to special issue on web-based learning. Educational Technology, 37(3), 7-15.
Hannafin, M.J., Hall, C., Land, S., & Hill, J. (1994). Learning in open-ended environments: Assumptions, methods, and implications. Educational Technology, 34(8), 48-55.
Heller, R.S. (1990). The role of hypermedia in education: A look at the research issues. Journal of Research on Computing in Education, 22(4), 431-41.
Jones, E.R. (1999). A comparison of an all web-based class to a traditional class. Paper presented at the meeting of the Society for Information Technology & Teacher Education International Conference (pp. 1372-1377), San Antonio, TX.
Kacmar, C.J., & Carlson, J.R. (1999). Increasing link marker effectiveness for WWW and other hypermedia interfaces: An examination of end-user preferences. Journal of the American Society for Information Science, 50(5), 386-98.
Kerlinger, F.N. (1986). Foundations of behavioral research. Orlando, FL: Harcourt Brace College Publishers.
Khan, B.H. (Ed.). (1997). Web-based instruction. Englewood Cliffs, N.J.: Educational Technology Publications.
Liou, H-C. (1997). The impact of WWW texts on EEL learning. Computer Assisted Language Learning, 10(5), 455-78.
Marchionini, G. (1988). Hypermedia and learning: Freedom and chaos. Educational Technology, 28(11), 8-12.
Merrill, M.D. (1980). Learner control in computer based learning. Computers and Education, 4(2), 77-95.
Milheim, W.D., & Azbell, J.M. (1988). How past research on learner control can aid in the design of interactive video materials. In proceedings of selected research papers presented at the annual meeting of the Association for Educational Communication and Technology, 461-472. New Orleans, LA. (ERIC Report ED 295652)
Milheim, W.D., & Martin, B.L. (1991). Theoretical bases for the use of learner control: Three different perspectives. Journal of Computer-Based Instruction, 18(3), 99-105.
Morrison, G.R., Ross, S.M. & Baldwin, W. (1992). Learner control of context and instructional support in learning elementary school mathematics. Educational Technology Research & Development, 40(1), 5-13.
Papert, S. (1980). Mindstorms: Children, computers and powerful ideas. New York, NY: Basic Books.
Reeves, T.C. (1993). Pseudoscience in computer-based instruction: The case of learner control research. Journal of Computer-Based Instruction, 20(2), 3 9-46.
Reigeluth, C.M., & Stein, F.S. (1983). The elaboration theory of instruction. In C.M. Reigeluth (Ed.), Instructional-design theories and models: An overview of their current status, 335-381. Hillsdale, NJ: Lawrence Erlbaum.
Romiszowski, A.J. (1994). Educational systems design implications of electronic publishing. Educational Technology, 34(7), 6-12.
Shin, E.C., Schallert, D.L., & Savenye W.C. (1994). Effects of learner control, advisement, and prior knowledge on young students' learning in a hypertext environment. Educational Technology Research & Development, 42(1), 33-46.
Shyu, H-Y, & Brown, S.W. (1992). Learner control versus program control in interactive videodisc instruction: What are the effects in procedural learning? International Journal of Instructional Media, 19(2), 85-96.
Starr, R.M. (1997). Delivering instruction on the Web: Overview and basic design principles. Educational Technology, 37(3), 7-15.
Steinberg, E.R. (1989). Cognition and learner control: A literature review, 1977-1988.
Sweeters, W. (1994). Multimedia electronic tools for learning. Educational Technology, 34(5), 47-52.
Tennyson, R.D. (1981). Use of adaptive information for advisement in learning concepts and rules using computer-assisted instruction. American Educational Research Journal, 18(4), 425-438.
White, S.E. (1999). The effectiveness of web-based instruction: A case study. Paper presented at the meeting of the Central States Communication Association and the Southern States Communication Association, St. Louis, MO.
Windschitl, M. (1998). The WWW and classroom research: What path should we take? Educational Researcher, 27(1), 28-33.