The Effects of Prior Knowledge and Goal Strength on the Use of Hypertext.

The study examined the influences of a student's prior knowledge and desired goal on the difficulties and benefits associated with learning from hypertext. Twelve students from an undergraduate course in educational psychology participated in the study. Students used a hypertext program (SKEIM, Kelly, 1993) to explore course content related to tests and measurements. The program automatically collects information about students' choices of topic, subtopic, and time on topic. Students were assigned to either a high or low prior knowledge category, based on whether they had prior exposure to the material, and were further assigned to subgroups with strong or weak goals. Students with strong goals were required to use the program to find the answers to specific questions whereas students in the weak goal condition were simply required to use the program for the general purpose of acquainting themselves with the material. Students were interviewed about their reactions to the program. They were also shown the pat h they had taken through the material and were asked about their choices. Results indicate that prior knowledge and goal strength are important contributors to students' cognitive and affective reactions to the hypertext system.

Vast, yet centralized, online hypermedia systems that can provide enormous amounts of information are now readily available within and outside of education, yet their design, particularly for learning, is not well understood. The hypermedia. environment is generally structured so that a user can click on a hot word or icon and explore a related topic. A user can potentially investigate an extensive amount of information in a relatively short period of time by simply following a path of interconnected concepts, creating a sort of journey (McAleese, 1989). In hypertext systems, a text-based form of hypermedia, the user reads a screen of text, decides what topic to explore further, and then attempts to pursue that topic.

Acquiring knowledge from a hypertext database can be a complex activity, placing great demands on a learner's capacity to understand new material white simultaneously navigating through a system. A commonly recognized difficulty with hypertext is the possibility of getting disoriented within a morass of data and information (Conklin, 1987; Dias & Sousa, 1997; Dias, Gomes, & Correia, 1999; Edwards & Hardman, 1989; Hammond & Allison, 1989). Navigational disorientation often occurs when a user suffers from cogritive overload as a result of being confronted with a mass of links through which to navigate with little structural support (Girill & Luk, 1992). A number of different approaches have been adopted to reduce the risk of navigational disorientation (Dias et al., 1999). Many of these efforts have focused on designing the hardware or software to minimize overload (Dias et al., 1999). Recent research showed that a more treelike or hierarchical text structure limits navigational difficulties as compared to a p urely heterarchical structure (Girill & Luk, 1992; Lin & Davidson-Shivers, 1996; Stanton, Taylor, & Tweedie, 1992; Billingsly, 1982).

While carefully structured hypertext can alleviate some navigation difficulties, disorientation is not just a function of the size or complexity of a database. Design characteristics of the hypertext environment (e.g., linking structures) interact with learner characteristics to enhance or limit performance (Beishuizen, Jeskijk, & Zanting, 1996; Dias & Sousa, 1997; Lin & Davidson-Shivers, 1996; McGregor, 1999). Among the variables known to interact with structural characteristics of software such as linking structures are users' prior knowledge and cognitive styles (Beishuizen et al., 1996; Lin & Davison-Shivers, 1996; McGregor, 1999).

Kelly (1993) showed that disoriented search behavior could be observed even when students were provided with a strict hierarchical hypertext structure and a very narrow choice of topic. Disoriented search behavior could still be observed, especially for students with little prior knowledge. Akanabi and Dwyer (1989) also noted the importance of prior knowledge over text structure in a study in which students with varying levels of prior knowledge benefited from different types of text structure. Students with high prior knowledge profited most from a less structured, inductive learning environment while low prior knowledge students benefited most from a more structured, deductive environment. Similar results were demonstrated by Pazzini (1991). Students' prior knowledge influenced their rate of concept learning more than by a text's content structure. Shapiro (1999) also showed that the use of interactive overviews during hypermedia-aided learning benefited those with little prior knowledge but had no effect on those with prior knowledge.

In general, prior knowledge can determine how well a learner acquires information from hypermedia since it can supply a mental framework, providing both perspective and context for new and inter-connected concepts, allowing a learner to fill in gaps in knowledge and make appropriate inferences (McNamara, Kintsch, Songer, & Kintsch, 1996). It can also guide motivated, self-directed learning (Alexander, Kulikowich, & Jetton, 1994; Alexander, Jetton, & Kulikowich, 1995). Furnishing a sort of road map of the material, prior knowledge allows a learner to focus more directly on information presented by reducing the amount of overhead associated with the task of navigating through hypermedia. Without some kind of mental map of the material already in place, working memory can become overloaded as a user tries to understand all the information presented and how it is structured.

Alexander et al. (1994), concluded that while subject matter knowledge is assumed to influence the nature of a learning activity, it is important to consider how additional factors affect learners. One such factor that was not considered in the review, but has drawn attention intermittently in the educational psychology literature since at least the 1960s, is the effect of goal specificity on learning. Goals influence many facets of learners' behavior. In a series of studies in which students reading the same passages were given different learning goals, goals greatly increased intentional learning (Kaplan 1974; Kaplan & Rothkopf, 1974; Kaplan & Simmons, 1974; Rothkopf & Kaplan, 1972). Further, learning was usually greater when the goals were more specific and less dense. Duchastel and Merrill (1973) came to similar conclusions in their review of literature on learning objectives. They claimed that presenting behavioral objectives to students can provide direction to their learning by helping the students di scriminate between relevant and incidental content. While learning objectives can serve to direct students' reading, objectives can also affect what information students recall. For example, depending on the specific goal and perspective assigned to students when reading, students can recall substantially different information from a reading passage (Anderson & Pichert, 1978).

Because prior knowledge and goals influence learning from linearly organized texts, it is important to consider how these factors affect learning in the hypermedia environment as well. In an ideal constructivist scenario, hypermedia users would search for complex information to meet their own goals (Jonassen & Wang, 1993). However, external goals are often imposed on learners. Beishuizen et al. (1996), showed that the use of hypertext provides no advantage over a linear text when the task is to study for an exam. In contrast, hypertext was advantageous when the task involved locating specific information. Structural features of the learning material thus interacted with the goal structure of the task. Because so much time using hypermedia is actually spent searching for information, it is worth examining how prior knowledge and goals affect information search and retrieval. A familiar example relating these concepts is searching for a book in the library. In the scenario in which one is in search of a specif ic book (a strong goal), the first step is to locate the correct stack of books, and the next is to locate the book of interest. If a searcher is unfamiliar with the library (low prior knowledge), it would be possible to wander up and down the aisles, but the searcher is likely to have to utilize the resources in the library to even begin to locate the correct stack of books. On the other hand, if the searcher is familiar with the library (high prior knowledge), he or she may have a general idea of which section contains the desired book and be able to locate the book fairly quickly. In the case where searchers have only a weak goal, such as to take out some leisure reading, they are more likely to wander then in the previous scenario. Further, if searchers are familiar with the library, they may be better able to narrow their wandering down to specific sections of the library for which they know they are interested.

The metaphor of a library search is appropriate except in one important aspect: hypermedia users are not usually provided with the same kind of diverse searching tools and resources as found in a library. An optimal hypermedia system must accommodate the user's need for guidance while preserving the quality of independent learning (Small & Grabowski, 1992, p. 446). Without the right kind of guidance and resources, it is possible for users to become frustrated and either abandon their learning task or gain knowledge they did not need.

METHOD

While goals and prior knowledge have been shown to affect linear text processing, very little research has been conducted on how these factors affect learners in a hypermedia environment. Because these factors have not been carefully considered in relation to hypermedia learning, research is still at the stage where rich descriptions are necessary in order to identify and explore relevant factors. Utilizing a qualitative methodology and small sample size, this study examined the influences of a user's prior knowledge and desired goal on the difficulties and benefits associated with hypermedia.

Participants

Twelve participants (eight female, four male) from an undergraduate course in Educational Psychology at a major East Coast university were recruited to participate in the study in return for course credit.

Materials

The study used a hierarchically structured hypermedia system, SKEIM, which runs on the Macintosh platform and was developed using HyperCard software. The program contains material from an undergraduate course in educational psychology on the subject of tests and measurement. Concepts were characterized using the sets of relationships identified by Dansereau and his colleagues (e.g., Holley & Dansereau, 1984). This classification of concepts was instantiated within SKEIM and is meant to provide an extra measure of structure and support for hypertext users. Students' performance while using the system is automatically collected by the program in the form of a data trace. The data collected in the trace recorded the student's review strategies by choice of theme, concepts searched, level of detail of the information accessed, and relationships among concepts. Traces contain data on theme, card name, level of depth for card, and time spent on card.

Material in SKEIM is linked together in a hierarchical structure, which can help users avoid some of the navigational issues associated with hypertext (Girill & Luk, 1992; Lin & Davidson-Shivers, 1996). SKEIM's hierarchical structure is set up such that no links between themes exist; in order to change themes, the student must return to the highest level card. Using SKEIM requires a user to begin at the highest level of detail for a theme, and through the use of pop-up menus, browse through several levels of detail about the theme; the deeper a student browses, the more detail about the topic is encountered. In all, SKEIM contains five themes (e.g., Test Scores) and 82 subtopics (e.g., definition of a Z Score) distributed through nine levels of depth. Depth of a card is defined as the number of screens a student passes through in accessing information about a particular theme; the content is structured so that very deep searches are less crucial for understanding main concepts (Table 1).

To navigate through SKEIM, a user is required to click on a hot word and is asked to choose one of several link types associated with that topic. For example, if a user decides to explore the topic of frequency distribution, he/she would be presented with a pop-up menu and have to choose one of the following: Example, Characteristics of, Leads to, Analogy, Part of, Type of, or Definition. This student might be familiar with the topic of frequency distribution and only interested in an example from which to refresh her memory. In SKEIM, this student would simply click on the frequency distribution. hot word and then choose Example. In a typical hypertext environment, the student would only have a choice such as "more information" and would have to search out an example (or other illustrative characterization) on his/her own.

Procedure

Students recruited for the study, who varied in prior knowledge of the program's content, were assigned to either the low or high prior knowledge category based on the following criteria: students in the high prior knowledge category had taken a course covering the subject of tests and measurement while students in the low prior knowledge group had little exposure to the content. Within the high and low prior knowledge groups, students were randomly separated into two sub-groups: strong and weak goals. Students in the strong goal group were given a specific task to complete. They were required to fill in partially completed sentences found in the hypertext database. Students in the weak goal group were given a nonspecific task. They were asked to review the material in the system, and told they would be asked questions about the material when they where done. The content was part of the students' course in educational psychology. They had not yet covered it in class at the time of this research but they woul d be accountable for the content on the final exam. Thus, the activity in this study was relevant to all students from a grading perspective.

Each student, on an individual basis, was shown how to use the program and then given specific or nonspecific goal to accomplish. Students were allowed to browse through the hypertext database program uninterrupted for approximately 40 minutes. After the task, each student was interviewed for approximately 45 minutes.

To gain insight into the students' thought processes while using the system, the researcher interviewed students using a structured interview. The first questions were fairly general and not trace specific (Appendix A). Subsequent questions asked of the student depended on the behavior reflected in the trace. For example, if a student switched a major theme after spending only a short time in the theme, the student was asked questions relating to that decision.

The first part of the interview involved questions related to the students' feelings about use of the program in general: In what ways was navigation difficult or simple, what kind of strategy did they use, and so forth? Students were then shown every step they made during their navigation of the hypertext database. The program showed them a replay of their search strategies, and students had the opportunity to reflect on their activity. At branching points, which represented a shift in topics, the researcher asked questions to gain an understanding about the student's immediate goal or purpose, if any, as well as any reflections on cognitive and metacognitive activity relating to the branching decision. All interviewing followed a basic protocol of questions using an interview guide described in Appendix A.

RESULTS

General Characteristics of Low and High Prior Knowledge Students

High prior knowledge. A total of six students were considered to have high prior knowledge. High prior knowledge students reported navigational problems. These problems were different than those reported by low prior knowledge students. Several students wanted a faster, more efficient way, to navigate and none complained about feeling lost or in need of assistance while using the program. These students all reported that the material was exactly what they expected; none felt their prior knowledge interfered with searching but rather made navigation simple. In response to questions about navigation, almost all believed they were able to find what they wanted easily and keep a model of the program in mind:

It was easy to navigate--the information was where I expected. I wanted a short cut though, so I didn't have to keep going back and forth. (Madelena)

I didn't get lost--I could remember where I was and figure out where to go. (Danielle)

Navigating was very simple. I had no trouble keeping everything in my head. Everything was where I expected. (Cory)

A few students complained about dead ends, but in reference to content rather than navigational issues. Several students expressed disappointment at dead ends if the material was not what they wanted or insufficiently fulfilled their interests; they seemed intrinsically motivated by their interests in the material and were discouraged by attaining what they considered inappropriate or incomplete information:

"There were times where I would go to a screen and there was only one sentence--it felt disjointed when that happened. I was sometimes disappointed with the lack of information on a screen." (Danielle)

Low prior knowledge. Six students were considered to have low prior knowledge. Students falling in the low prior knowledge category (i.e., with little familiarity with the concepts in tests and measurement) tended to find the system difficult to use. The most common complaint was disorientation. Several students claimed that the farther they went into the program, the more lost they felt; the more details they encountered, the harder it was to maintain a navigational perspective. One student claimed that the system was like a big matrix of information, while another referred to it as a maze. Some felt that the content was broken up in an outline format making the material hard to remember, limiting their comprehension of the material. Several claimed they would have benefited by some kind of map or navigational tool. The students were especially disconcerted when a search path dead-ended (i.e., there were no other forward branching directions) because they were then forced to make more explicit and thoughtful navigational decisions rather than simply continuing along a path.

An interesting difficulty encountered by the low prior knowledge students related to using and understanding hypertext terms. A commonly perceived advantage to hypertext is the opportunity for a user to create mental associations while traversing material. The problem for content novices is the likelihood of encountering unfamiliar hypertext words. Unfamiliar words are likely, in turn, to lead to screens with still more unfamiliar words. Several low prior knowledge students felt that hypertext terms were really only helpful when the terms were familiar:

The [hypertext] terms that were more familiar helped more--they were a lot more useful. (Laurie)

The links seemed ambiguous at times--they didn't always give me what I expected. (Lisa)

When I already knew the material, the links were good n they usually brought me where I expected. (Benny)

I was so unfamiliar with the content and the terms, I just didn't know which way a hypertext term would take me. (Peter)

Although the content in the program was highly organized, students who had very little prior knowledge could not seem to use the structure effectively, having no cognitive map of the domain that would allow them to exploit the structure. Several students had limited and inaccurate models of the material, which ended up interfering with their review. In describing her strategy during a certain part of her trace, Laurie commented that:

When I thought I knew what to expect, it usually tamed out different than I expected--and that threw me off. I had preconceived notions about the grades topic-I expected material about "A's B's etc" but found frequency distribution, mean and mode instead.

Laurie claimed that her own expectations of the material were so far off from the program's structure, that she became quite frustrated and gave up searching through the section.

Another possible problem with providing a highly structured learning environment is that students were unable to organize material according to their own growing comprehension of the material. For example, activities such as constructing content diagrams or schematic maps when first learning a subject often prove helpful. These kinds of study strategies helped students with little prior knowledge create a model of the material, which is personally meaningful. On this point, Benny felt that the content in the hypertext database was: "...organized in a way that made it seem structured--I didn't have to structure it myself I actually would rather have created my own structure since I wanted the material organized differently."

General Characteristics of High and Low Prior Knowledge Students Performing the Weak Goal Task

The weak goal condition required students to review the material and would be asked questions about the content later. The most apparent differences among low and high prior knowledge students performing the weak goal task related to strategy. The low prior knowledge students tended to go through the material methodically, in the order it was presented while the high prior knowledge students tended to bounce around looking to fill in gaps in their knowledge. The students deviated from their strategies only when they encountered a subject with which they claimed they were more or less familiar; when novice students encountered familiar material, they would follow a strategy more typical of the high prior knowledge students. Similarly, when high prior knowledge students encountered unfamiliar material, they would follow a more methodical approach typical of the novice students.

Low prior knowledge/weak goal. The students in the low prior knowledge category utilized a tentative and methodical strategy, not exploring the more detailed concepts until becoming familiar with a topic:

If I was seeing something for the first time, I didn't go too deep into detail but tried to get a more general idea. I was afraid that if I went too deep I would lose sight of what I had just learned. (Laurie)

I didn't go too deep at first. I tried to get a general understanding and then went deeper into detail. I would sometimes get lost when I looked into the details. (Lisa)

I went through in order and thoroughly read everything. Even if I knew a concept, I would check it Out to make sure there wasn't anything I didn't already know. (Benny)

A good example of the thorough and methodical nature of a low prior knowledge strategy is shown in Table 2. It contains a list of all possible nodes in the Test Scores theme, the depth for each node, and the relative order that Laurie traversed the nodes. Laurie claimed that she was extremely unfamiliar with the test scores topic, and as Table 2 reveals, she visited most of the cards in that theme and navigated basically in the order presented by the system. When questioned as to why they followed such a thorough strategy, the students explained in terms of their low prior knowledge; they generally had poor cognitive models of the material and as a result followed the model or structure presented to them by the program. This strategy was adaptive since by following the program's model, there was less cognitive overhead related to navigational factors therefore reducing the risk of becoming lost.

High prior knowledge/weak goal. Unlike the novice students, the high prior knowledge group performed the weak goal task in a fairly non-methodical manner. Their strategy, although not haphazard, was much more unpredictable. The students tended to search out of order looking for material that was either new or of interest. This tactic, however, only pertained to material that was familiar. When the students discovered new material, their strategy changed, appearing similar to the novice users' as they searched in an orderly and thorough fashion:

I usually tried to start with a topic less familiar and then went through it carefully. I realized I knew most of the material in the program so I tried to find topics I hadn't already seen. (Cory)

I went through and looked for information I didn't already know. Towards the end, I went back over stuff that I had not gone to at first just to make sure I hadn't missed anything. (Jamie)

For the topics I already understood, I usually looked at definition first to refresh my memory. In general I would skip the examples if I already knew the material. (Danielle)

The trace data in Table 3, a section of the Test Types theme, reveals some of the behavior described by the students in the previous quotes. Both Cory and Jamie visited nodes sporadically and out of order, revisiting a few nodes after exploring other topics. Danielle's behavior in this section does not match the expected high prior knowledge strategy displayed by Gory and Jamie, but rather is more indicative of a low prior knowledge student; she visited the nodes in this section basically in order and thoroughly. Although Danielle met the criteria for high prior knowledge she turned out to be low in prior knowledge for this local section of content. She explained her change in strategy as a result of the fact that: "The sections on standardized tests had the most new material in it. I had never learned about norm-referenced or criterion-tests before." In general, students with high prior knowledge were better able to take advantage of the program's assets since, for most topics, they were not learning the su bject for the first time. They were able to search for new and interesting information without the fear of navigational difficulties and generally seemed more enthusiastic and encouraged by their experiences.

High and Low Prior Knowledge Students Performing the Strong Goal Task

The strong goal task assigned in this study requires to navigate through the system seeking answers to the questions listed in Appendix B. The average number of correctly completed answers out of a total of ten questions was 6.0 for the low prior knowledge group and 6.7 for the high prior knowledge group. Of the six students in the strong goal group, the lowest score of four correct answers occurred in the low prior knowledge group and the highest score of nine correct occurred in the high prior knowledge group.

Navigation and the structure of the material in the hypertext database presented many more serious problems for the students in the strong goal group. Students with little familiarity of the material found the task extremely difficult, used poorly defined strategies, and became quite frustrated. Students with higher prior knowledge were better able to perform the task, had only slightly better strategies, and were much less likely to become frustrated.

Low prior knowledge/strong goal. Students with low prior knowledge had little idea where to begin searching for the information needed to complete the sentences listed in Appendix B. They tended not to follow a well defined strategy once they did start searching, and generally stumbled upon answers by chance. This inefficient searching method could be attributed to the student's poor cognitive model of the subject; even if a student understood the question, he/she did not understand the topic well enough to know where to look for the answer:

I tried to use my knowledge to figure out what to do. I compared what I already knew with what was there-but it didn't end up being that easy. My own knowledge actually interfered. It seemed like so many choices could lead to an answer. (Domenica)

I spent a lot of the time wondering where I was in the program. I usually tried to pick a topic to start with, and then went through everything looking for the answer. (Beth)

I was only familiar with very little of the material-I didn't know which way to go to get an answer. I ended up finding answers randomly. (Peter)

Consistent with the previous quotes, these students often displayed a methodical and exhaustive search strategy similar to the low prior knowledge students in the weak goal group. A trace example illustrating this point is shown in Table 4. The answers to questions three and four were located in the Test Scores theme and, as Table 4 illustrates, the low prior knowledge students traversed the Test Scores section thoroughly and basically in order (each student successfully answered these two questions).

As with the students in the weak goal group, these students found navigation difficult. However, navigational issues were only of minor annoyance to students in the weak goal group; they could simply follow an exhaustive and methodical approach to counteract the problem. For the low prior knowledge students performing directed searches, navigational problems hindered performance. Because they were searching for specific information, they had to rely on their own poor models of the material to help guide them through the system and as a result, were much less effective. Domenica claimed that using the program was like trying to remember a maze: "You can go so many ways--it was hard to find my way back. I couldn't keep it all in my memory. The more specific questions were harder to figure out--you had to go deeper (into the maze) to get the answer and could get lost easier. A result of the difficulties mentioned so far was frustration, directed by the students both at themselves and the program:

I expected to be able to get more information from the program--I would get frustrated if the information I got wasn't what I wanted. (Peter)

I became frustrated like I do on exams. Especially when I got lost or when I couldn't figure out where to go--or whether I had been someplace before. (Domenica)

High prior knowledge/strong goal. For the students in the high prior knowledge group, performing directed searches was a considerably more manageable task. The strategy used most commonly by these students was to look for familiar material in a non-methodical approach, matching a set of associations formed with what they found in the system. These students generally had a good idea where to start searching, and often quickly realized when they were off track. The students tended not to go through the material thoroughly, but rather seek appropriate information:

I would look for stuff based on what I already knew. I didn't usually read everything. I already knew a lot of the material. If I couldn't find something, I would go back and read some more. (Holly)

I would look at a question and could usually tell where to start out. I would keep going until I found what I wanted. I'd narrow it down by going through and looking for something that was familiar or made some sense for the question. I had an idea of what I was looking for--

I wasn't just bouncing around. (Madelena)

I looked for key words in the question I was trying to answer. Then I would look for stuff in the system, which matched the key words. (Daniel)

Unlike the typically exhaustive strategy used by the low prior knowledge students, the trace created by these students tended to be unpredictable, probably as a result of their nonmethodical strategy. Differences between the low and high prior knowledge students can be easily observed by comparing Table 4 and Table 5, which both contain trace data from the Test Scores theme. While Table 4 shows that the low prior knowledge students traversed this section basically in order, Table 5 reveals that the high prior knowledge students tended to jump back and forth between the Test Scores and other sections.

This strategy was ineffective at times; it occasionally caused a student to just barely miss an answer. Unlike the students performing the weak goal task, these students could suffer directly and immediately by not reading a topic they thought they already knew. Several of the students realized that this was happening and actually changed their strategy:

I switched strategies after having trouble finding the answer to an easy question. I got less focused on finding specific answers--I figured I'd go through and stumble on the answer. I read all the branches so I wouldn't miss anything. (Madelena)

I didn't check the holistic topic well and didn't realize how close I was to the answer. So I was more careful after that; I would go to each link so as not to miss anything. (Holly)

DISCUSSION

The purpose of this study was to examine the influences of prior knowledge and goal specificity, focusing on students' search strategies and navigation styles but also considering students' affect toward the hypertext learning experience. The ability to trace students' search behavior can provide valuable information about the learning process if interpreted correctly. By conducting indepth interviews, which are unusual in this style of research, The researchers were able to go beyond assumptions about how people use educational hypertext from the trace data alone. As a result, they were able to investigate some of the affective as well as cognitive reactions to nonlinear text processing.

An important finding yielded by the methodology employed in this study related to the interpretation of trace data. In this study, a number of the participants appeared to make unsystematic and even random selections of content whereas others exhaustively and systematically searched the links available within a section. The trace data alone do not allow the interpretation of the quality of the strategy employed as evidenced by the interviews conducted in this study. The apparently unsystematic patterns of choices deployed by students with high prior knowledge were a consequence of "filling in" information they already knew. An examination of the trace data alone might simply have concluded that the student was non-strategic. The findings reported here related to the alignment of the trace data and students' comments about their intentions and goals suggested caution in relying on trace data without due regard to the students' goals.

One advantage of a hypermedia system such as that described here is that it allows for the presentation of highly structured and organized material. In processing text, high prior knowledge students scored better on tests of meaningful learning if they have worked with less coherent text (McNamara et al., 1996). The opposite is true for low prior knowledge students. Similar results have been found with the processing of lecture material with low prior knowledge students benefiting from an explicit exposition of the relationships between ideas and high prior knowledge students benefiting when those relationships were not explicated (Lambiotte & Dansereau, 1992). In contrast, in a hypertext environment, students with prior knowledge can take advantage of the structure of the content because they appear to use their own cognitive models to guide their activity. Students with prior knowledge of the content were better able to navigate easily, remember where they had been, and decide how to get to where they want ed. They reported more positive feelings about using the system than did the low prior knowledge students and seemed to suffer much less from frustration while performing their tasks. The low prior knowledge students often suffered from disorientation, even though the database was highly structured. These results are generally consistent with research on the effects of prior knowledge and text structure on learning (Pazzini, 1991; Tennyson & Bagley, 1992).

Results show that students tended to have more than just a cognitive reaction when learning from hypertext. High levels of anxiety were common for the low prior knowledge students especially when they were required to perform a specific learning task. These findings extend those of Beishuizen et al. (1996) who showed that task type influenced the effectiveness of hypertext. In this study, it was observed that the goal structure of a task interacts with the prior knowledge of a student and influences both strategy use and affect. The implications of these findings are that features of hypermedia design, which in interaction with specific individual characteristics such as prior knowledge and goals, can promote negative affect that is known to be nonproductive for learning.

While research has considered the effects of prior knowledge on the hypertext environment (Alexander et al., 1994) students' goals remain relatively unexamined. Students are usually left to browse without specific learning goals. Because self-determined goals are difficult to judge and research has shown that goals influence learning (Duchastel & Merrill, 1973), research considering only prior knowledge may be incomplete. Limitations of the hypermedia learning environment which do not emerge when a student simply wanders (e.g., using conventional search engines), may play a critical role when a student is assigned to learn a specific task. For instance, what is intended as a review task may end up as a knowledge acquisition task when goals are introduced. Because knowledge acquisition tasks are harder for students with low prior knowledge, hypertext may not be the best format to create an initial schema for a knowledge base. On the other hand, for students with high prior knowledge, the elaboration of a sche ma may be greatly facilitated by this format.

This study contributes to the previous literature by considering how learning goals affect hypermedia users. Students' goals and prior knowledge influenced their search strategies and navigational styles. The low prior knowledge students performing the strong goal task had complaints about navigation and generally stated that they would have benefited by the use of a navigational aid, such as a map or record of nodes visited. They were basically unsuccessful at their task, suffering more seriously as a result of cognitive overload, frustration, and poor ability to navigate than did those novices simply traversing through the system. The high prior knowledge students suffered much less from negative affect and were somewhat more successful at completing the strong goal task. The task was easier for these students since they already had a mental model of the content and had some idea where to look for an answer.

These findings suggest that developers of educational software should seriously consider both characteristics of the learner and of the tasks before committing to a method of instructional delivery. A system designed to teach novices might be more effective if it used techniques geared toward introducing students to a subject rather than a hypertext format. In general, hypermedia systems should be designed to accommodate users of different levels of prior knowledge. A hypermedia system designed for an audience varying in content expertise should provide as many resources as possible, since novices will probably require them and experts might benefit from their presence. Finally, with expert students a prestructured hypertext database may impose an analytical and organizational scheme at odds with their own learning. These students may benefit from malleable hypertext databases. In short, these data suggest that a one-size-fits all approach to the construction of hypertext databases for learning is not tenabl e.

References

Akanabi, M.R., & Dwyer, F.M. (1989). Effects of students' prior knowledge level on their ability to profit from visualized inductive and deductive instructional strategies. International Journal of Instructional Media 16, 69-85.

Alexander, P.A, Jetton, T.L., & Kulikowich, J.M. (1995). Interrelationship of knowledge, interest, and recall: Assessing a model of domain learning. Journal of Educational Psychology, 87, 559-575.

Alexander, P.A, & Kulikowich, J.M., & Jetton, T.L. (1994). The role of subject-matter knowledge and interest in the processing of linear and nonlinear texts. Review of Educational Research, 64, 201-252.

Anderson, R.C., & Pichert, J.W. (1978). Recall of previously unrecallable information following a shift in perspective. Journal of Verbal Learning and Verbal Behavior, 17, 1-12.

Beishuizen, J., Jesdijk, E., & Zanting, A. (1996). Using hypertext for studying and information search. Journal of Educational Computing Research, 15, 289-316.

Billingsley, P.A. (1982). Navigation through hierarchical menu structure: Does it help to have a map? Proceedings of the Human Factors Society 26th Annual Meeting (pp. 103-107). Santa Monica, CA: Human Factors Society.

Conklin, E.J. (1987). Hypertext: An introduction and survey. IEEE Computer, 2, 17-41.

Dias, P., Gomes, M.J., & Correia, A.P. (1999). Disorientation in hypermedia environments: Mechansims to support navigation. Journal of Educational Computing Research, 20, 93-117.

Dias, P., & Sousa, A.P. (1997). Understanding navigation and disorientation in hypermedia learning environments. Journal of Educational Multimedia and Hypermedia, 6, 173-185.

Duchastel, P.C., & Merrill, P.F. (1973). The effects of behavioral objectives on learning: A review of empirical studies. Review of Educational Research, 43, 53-69.

Edwards, D.M., & Hardman, L. (1989). Lost in Hyperspace': Cognitive mapping and navigation in a hypertext environment. In R. McAleese, (Ed.) Hypertext: Theory into Practice, pp. 105-125. Oxford: Intellect Books.

Girill, T.R., & Luk, C.H. (1992). Hierarchical search support for hypertext on-line documentation. International Journal of Man-Machine Studies. 36, 571-585.

Hammond, N., & Allison, L. (1989). A learning support environment: The hitch hiker's guide. In R. McAleese (Ed.), Hypertext: Theory into practice. (pp. 62-74). Oxford: Intellect Books.

Holley, C.D., & Dansereau, D.F. (1984) (Eds.). Spatial learning strategies: Techniques, applications, and related issues. New York: Academic Press.

Jonassen, D.H., & Wang, S. (1993). Acquiring structural knowledge from semantically structured hypertext. Journal of Computer-Based Instruction, 20. 1-8.

Kaplan, R. (1974). Effects of learning with part vs. whole presentations of instructional objectives. Journal of Educational Research, 66, 254-256.

Kaplan, R., & Rothkopf, E.Z. (1974). Instructional objectives as directions to learners: Effect of passage length and amount of objective-relevant content. Journal of Educational Psychology, 66, 448-456.

Kaplan, R., & Simmons, F.G. (1974). Effects of instructional objectives used as orienting stimuli or as summary/review upon prose learning. Journal of Educational Psychology, 66, 614-622.

Kelly, A.E. (1993). Designing instructional hypertext for use in lecture note review: knowledge engineering and preliminary testing. Journal of Educational Multimedia and Hypermedia, 2, 149-176.

Lambiotte, J.G., & Dansereau, D.F. (1992). Effects of knowledge maps and prior knowledge on recall of science lecture content. Journal of Experimental Education, 60, 189-201.

Lin, C., & Davidson-Shivers, G. (1996). Effects of linking structure and cognitive style on students' performance and attitude in a computer-based hypertext environment. Journal of Educational Computing Research, 15, 317-329.

McAleese, R. (1989). Navigation and browsing in hypertext. In R. McAleese (Ed.), Hypertext: Theory into practice. (pp. 6-44). Oxford: Intellect Books.

McGregor, S.K. (1999). Hypermedia navigation profiles: Cognitive characteristics and information processing strategies. Journal of Educational Computing Research, 20, 189-206.

McNamara, D.S., Kintsch, E., Songer, N.B., & Kintsch, W. (1996). Are good texts always better? Interactions of text coherence, background knowledge, and levels of understanding in learning from text. Cognition and Instruction, 141, 1-43.

Pazzani, M.J. (1991). Influence of prior knowledge on concept acquisition: Experimental and computational results. Journal of Experimental Psychology: Learning, Memory. and Cognition. 17, 416-432.

Rothkopf, E.Z., & Kaplan, R (1972). Exploration of the effect of density and specificty of instructional objectives on learning from text. Journal of Educational Psychology, 63, 295-302.

Shapiro, A.M. (1999). The relationship between prior knowledge and interactive overviews during hypermedia-aided learning. Journal of Educational Computing Research, 20, 143-167.

Small, R.V., & Grabowsky, B.L. (1992). An exploratory study of information-seeking behaviors and learning with hypermedia information systems. Journal of Educational Multimedia and Hypermedia, 1, 445-464.

Stanton, N.A., Taylor, R.G., & Tweedie, L.A. (1992). Maps as navigational aids in hypertext environments. Journal of Educational Multimedia and Hypermedia, 1, 43 1-444.

Tennyson, R.D., & Bagley, C.A. (1992). Structured versus constructed instructional strategies for improving concept acquisition by domain-competent and domain-novice learners. Journal of Structured Learning, 11, 255-263.

                    Distribution of Content Material
Theme                     % of Total Content
Reasons for Testing:              5%
Basic Concepts:                  16%
Qualities of Good Tests:         19%
Test Types:                      49%
Test Scores:                     12%
Note. Content is found in 187 screens (cards), and
is distributed among five themes.
                     Trace Data of the Test Scores
                    Section of Laurie's Performance
Level    Theme        Topic                    Link Type
3      Test Scores  Standard Scores            Type of
3      Test Scores  Standard Scores            Characteristics
3      Test Scores  Standard Scores            Definition
3      Test Scores  Grade Equivalent Scores    Example
3      Test Scores  Grade Equivalent Scores    Characteristics
3      Test Scores  Grade Equivalent Scores    Leads to
3      Test Scores  Grade Equivalent Scores    Definition
3      Test Scores  Percentile Ranks           Leads to
3      Test Scores  Percentile Ranks           Example
3      Test Scores  Percentile Ranks           Definition
4      Test Scores  Stanine Scores             Characteristics
4      Test Scores  Stanine Scores             Definition
4      Test Scores  T Scores                   Example
4      Test Scores  T Scores                   Leads to
4      Test Scores  T Scores                   Characteristics
4      Test Scores  T Scores                   Definition
4      Test Scores  Z Scores                   Example
4      Test Scores  Z Scores                   Leads to
4      Test Scores  Z Scores                   Characteristics
4      Test Scores  Z Scores                   Definition
4      Test Scores  Interpretation Percentile  Example
                    Ranks
Level  Traversal Order
3            32
3            31
3            30
3            29
3            27
3            28
3            26
3
3
3            25
4            39
4            38
4
4            37
4            36
4            35
4
4
4            34
4            33
4
Note From left to right, columns
contain node level, theme, topic,
link type, and
relative traversal order of
Laurie's trace through the test
scores section of SKEIM.
                         Trace Data of the Test
                         Types Section for the
                            Three High Prior
                               Knowledge/
                           Weak Goal Students
Level    Theme     Content                     Link Type
  4    Test Types  Formative                   Characteristics
  4    Test Types  Formative                   Leads to
  4    Test Types  Formative                   Example
  4    Test Types  Formative                   Definition
  4    Test Types  Criterion-Referenced Tests  Characteristics
  4    Test Types  Criterion-Referenced Tests  Example
  4    Test Types  Criterion-Referenced Tests  Leads to
  4    Test Types  Criterion-Referenced Tests  Definition
  4    Test Types  Norm-Referenced Tests       Example
  4    Test Types  Norm-Referenced Tests       Characteristics
  4    Test Types  Norm-Referenced Tests       Leads to
  4    Test Types  Norm-Referenced Tests       Definition
  4    Test Types  Achievement Tests           Example
Level  Dannielle  Cory Jamie
  4
  4
  4
  4
  4       37
  4       38          24
  4
  4       36          21
  4       35
  4       34
  4       33
  4       32
  4                   22
4  Test Types  Achievement Tests   Type of          31      20
4  Test Types  Achievement Tests   Characteristics  30      23
4  Test Types  Achievement Tests   Definition       29      19
4  Test Types  Aptitude Tests      Example
4  Test Types  Standardized Test   Characteristics  24
4  Test Types  Standardized Test   Example          39
4  Test Types  Uniform Conditions  Characteristics  26
4  Test Types  Uniform Conditions  Example
4  Test Types  Uniform Conditions  Leads to         27
4  Test Types  Uniform Conditions  Definition       25
4  Test Types  Essay Tests         Characteristics  47   3   6
4  Test Types  Aptitude Tests      Characteristics
4  Test Types  Aptitude Tests      Definition       28      18
5  Test Types  Multiple Choice     Characteristics  44  19
5  Test Types  True False          Characteristics      17
5  Test Types  Multiple Choice     Leads to             22
5  Test Types  Multiple Choice     Part of          43  18
Note. From left to right,
columns contain node level,
theme, topic, link type,
and relative traversal
order for Danielle's, Cory's,
and Jamie's trace through
the Test Types section of
SKEIM.
                     Trace Data of the Test Scores

                       Section for the Three Low
                  Prior Knowledge/Strong Goal Students
Level     Theme     Content                    Link Type
3      Test Scores  Standard Scores            Type of
3      Test Scores  Standard Scores            Characteristics
3      Test Scores  Standard Scores            Definition
3      Test Scores  Grade Equivalent Scores    Example
3      Test Scores  Grade Equivalent Scores    Characteristics
3      Test Scores  Grade Equivalent Scores    Leads to
3      Test Scores  Grade Equivalent Scores    Definition
3      Test Scores  Percentile Ranks           Leads to
3      Test Scores  Percentile Ranks           Example
3      Test Scores  Percentile Ranks           Definition
4      Test Scores  Stanine Scores             Characteristics
4      Test Scores  Stanine Scores             Definition
4      Test Scores  T Scores                   Example
4      Test Scores  T Scores                   Leads to
4      Test Scores  T Scores                   Characteristics
4      Test Scores  T Scores                   Definition
4      Test Scores  Z Scores                   Example
4      Test Scores  Z Scores                   Leads to
4      Test Scores  Z Scores                   Characteristics
4      Test Scores  Z Scores                   Definition
4      Test Scores  Interpretation Percentile  Example
                    Ranks
Level  Domenica  Peter  Beth
3         15      12      9
3         26      11      8
3         14              7
3
3         28              5
3         27      10      4
3         18       9      6
3                  7      1
3         29              2
3         13       8      3
4         17      13     10
4         16      14
4
4         31
4
4
4
4         30
4
4
4
Note. From left to right,
columns contain node level,
theme, topic, link type, and
relative traversal order for
Domenicas, Peter's, and
Beth's trace through the
Test Scores section of SKEIM.
                     Trace Data of the Test Scores
                       Section for the Three High
                  Prior Knowledge/Strong Goal Students
Level  Theme        Content                  Link Type        Daniel
3      Test Scores  Standard Scores          Type of             8
3      Test Scores  Standard Scores          Characteristics     7
3      Test Scores  Standard Scores          Definition
3      Test Scores  Grade Equivalent Scores  Example            47
3      Test Scores  Grade Equivalent Scores  Characteristics     6
3      Test Scores  Grade Equivalent Scores  Leads to           36
3      Test Scores  Grade Equivalent Scores  Definition         15
3      Test Scores  Percentile Ranks         Leads to           46
3      Test Scores  Percentile Ranks         Example            16
3      Test Scores  Percentile Ranks         Definition
4      Test Scores  Stanine Scores           Characteristics    10
4      Test Scores  Stanine Scores           Definition          9
4      Test Scores  T Scores                 Example
4      Test Scores  T Scores                 Leads to           12
4      Test Scores  T Scores                 Characteristics    13
4      Test Scores  T Scores                 Definition         11
4      Test Scores  Z Scores                 Example
4      Test Scores  Z Scores                 Leads to
4      Test Scores  Z Scores                 Characteristics    45
4      Test Scores  Z Scores                 Definition         14
4      Test Scores  Interpretation           Example
                    Percentile Ranks
Level  Holly  Madelena
3       11       13
3       32       11
3       31       12
3
3
3                10
3       10        7
3        8        9
3        5       22
3       33        8
4       12       18
4                17
4
4                21
4                20
4                19
4
4                16
4                15
4                14
4
Note. From left to right, columns
contain node level, theme, topic,
link type, and relative traversal.