Students Evaluating Internet Sources: From Versatile Evaluators to Uncritical Readers

The Internet is a significant information resource for students due to the ease of access it allows to a vast amount of information. As the quality of the information on the Internet varies, it is important that students are able to evaluate such information critically. The aim of the study was to investigate how students evaluate Internet sources in an authentic learning task. Upper secondary school students (n = 25) were asked to look for source material on the Internet in order to write an essay. They were asked to verbalize their thoughts during the material gathering process. Their verbalizations and actions on the Internet were recorded and analyzed. The five evaluation profiles emerged: 1) versatile evaluators; 2) relevance-orientated evaluators; 3) limited evaluators; 4) disorientated readers; and 5) uncritical readers.


INTRODUCTION
The Internet is widely used for educational purposes and it is an important information resource for students. Instead of libraries and textbooks, students rely more and more on the Internet in their school work (Jones & Madden, 2002;Lenhart, Simon, & Graziano, 2001;Metzger, Flanagin, & Zwarun, 2003). Because anyone can publish anything on the Internet, the information obtained from this source can be inaccurate or even misleading. As in most cases teachers are unable to control the quality of Internet-derived information used by students, it is important to know how well students are able to evaluate that information.
Many researchers (Liu & Huang, 2005;Metzger et al., 2003;Whitmire, 2004) have investigated university or college students' evaluation of information on the Internet, but fewer studies (Brem, Russel, & Weems, 2001;Lorenzen, 2001) concern the evaluations of adolescent learners. Previous research has indicated that college students have shortcomings in evaluating Internet sources (Grimes & Boening, 2001;Metzger et al., 2003). Therefore, it is important to investigate the evaluation strategies used by younger students to be able to better scaffold them and prepare them for their future studies.
In the present study evaluation of information on the Internet is observed in an authentic learning context. The results provide information about the evaluation strategies used by upper secondary school students when searching the Internet for source material for an essay. Due to the cumbersomeness of the procedure, the number of participants in earlier think-aloud studies has ranged from 10 to 15 (Coiro & Dobler, 2007;Eveland & Dunwoody, 2000;Whitmire, 2004). More students are, however, needed in order to properly identify evaluation profiles among students. The purpose of this study is to meet this challenge.

Literacy Skills and Critical Thinking
Researchers agree that, with the rapid development of information and communication technologies, more diversified literacy skills are needed (Coiro, 2003;Leu, Kinzer, Coiro, & Commach, 2004). Various new concepts, such as information literacy (Bruce, 2002;Doyle, 1994), digital literacy (Gilster, 1997), network literacy (McClure, 1994), and media literacy (Aufderheide & Firestone, 1993) have been used to describe these skills. Bawden (2001), in reviewing the new literacy concepts, noted that the concept of information literacy is the most widely used. Doyle (1994, p. 1) defines information literacy as the ability to access, evaluate, and use information drawn from a variety of sources. The information-literate student accesses information efficiently, evaluates information critically, and uses information accurately and creatively (Information literacy standards for student learning, 1998).
Because the Internet is full of partial and sometimes even misleading information, the most important skill when making use of it is critical thinking (Gilster, 1997, p. 87). Ennis (1987, p. 10) crystallizes the concept of critical thinking as reasonable reflective thinking that is focused on what to believe or do. According to Facione (2006), critical thinking skills consist of inference, evaluation, interpretation, analysis, explanation, and self-regulation. In addition to critical thinking skills, a person can have a critical disposition (Ennis, 1987;Facione, 2006). A person with a critical disposition is truth seeking, confident in reasoning, judicious, inquisitive, systematic, analytical, and open-minded (Facione, 2006). Although finding credible and relevant information on the Internet is sometimes challenging, the Internet also creates new opportunities to access alternative viewpoints and to compare argumentation drawn from different sources.
The study by Connor-Greene and Greene (2002) indicates that students need to be taught to evaluate texts critically. In their study 39 university students read an Internet article about the dangers of aspartame and wrote their reactions to that article. More than half of the students accepted the information in the article unquestioningly, a quarter expressed ambivalence, and a fifth of the students did not believe the article. After a 30-minute structured small-group discussion about the article all the groups listed 5-11 problems that appeared in the article. The students mentioned problems such as no research evidence to support allegations, no citation of sources or expert cited in the article, a sensational writing style, and sweeping statements.

Evaluation of Information
Critical thinking skills are important when students evaluate information on the Internet. Judd, Farrow, and Tims (2006) used credibility and relevance as factors in their instrument designed to teach undergraduates to evaluate the information value of Web pages. The dimensions of credibility in their model are expertise (knowledgeable source and factual information) and trustworthiness (unbiased). The dimensions of information relevance are importance and currency.
According to Harris (1997), evaluation of credibility involves evaluation of the authority, quality, style, and accuracy of the text, evaluation of the purpose of the text, and verification of information. In the information science literature the relevance of information has been defined in various ways (Barry & Schamber, 1998;Borlund, 2003;Saracevic, 1996). Relevance can be system-orientated, that is, the connection between a query and a document, or user-orientated, dealing with the user's subjective decision-making. Saracevic (1996, pp. 212-213) divides relevance into five types of which three fall into the user-orientated category. Cognitive relevance refers to how well a document satisfies the user's information need. It includes perceived informativeness, novelty, and information quality. Situational relevance is associated with a task and user's interpretation of it. Finally, motivational relevance concerns the intents and goals of the user as well as the satisfaction experienced by him or her.
In the present study, credibility refers to whether a student is paying attention to distinguishing reliable from unreliable information. Relevance, in turn, refers to whether a student is paying attention to distinguishing essential from non-essential information.

Evaluation Studies
Information evaluation on the Internet has been studied, for example, by surveys (Metzger et al., 2003;Wu & Tsai, 2005). Metzger et al. (2003) found that college students relied extensively on the Internet as a source of information in their academic work. Although students did not consider Web-based information to be as credible as traditional sources of information, they reported verifying information found on the Internet only rarely or occasionally. Surveys, however, do not reveal what kind of evaluation strategies students actually use when they are working on a school assignment and how those strategies are reflected in the quality of material studied.
When Grimes and Boening (2001) interviewed 11 college students about how they evaluated Web sources cited to in their research papers, they found that students evaluated Web sources only superficially, if at all. Owing to inadequate evaluation skills or unwillingness to evaluate material, students used unauthenticated Web resources in their research papers. Whitmire (2004), in turn, studied the relationship of college students' (n = 15) epistemological beliefs (by questionnaire), reflective judgments (interview), and information search (think-aloud) on the Internet. The study indicated that the more advanced epistemological believers were better able to evaluate information sources and authority by using, for example, URL addresses and information of the institutional affiliation of the author. Brem, Russel, and Weems (2001) studied how 81 girls from the 9th, 11th, and 12th grades evaluated scientific argumentation of Websites of three kinds: hoaxes, weaker sincere sites, and stronger sincere sites. They found that evaluation could, in part, be a situated activity. Students used more absolutist reasoning when evaluating weaker sincere sites than stronger sincere sites. By contrast, stronger sincere sites were accepted more easily on the basis of surface markers whereas weaker sincere sites received more attention. At hoax sites students used as little effort as possible to distinguish real from fake.

Role of Evaluation in the Search Process
According to Rieh (2000Rieh ( , 2002, people's judgments of information quality and cognitive authority on the Internet are either predictive or evaluative. Predictive judgments are based on a person's expectations of a Web page prior to opening the page. Predictive judgment directs the decision-making and it can be based on knowledge, experience, recommendation, or on the information given in the search results. After opening a new Web page an evaluative judgment is made on the basis of the nature of the information, its source, and the searcher's prior knowledge. In their interview study with seven students, MaKinster, Beghetto, and Plucker (2002, p. 167) described the usefulness of the predictive strategy in an information search task. They found a student who carefully evaluated the content of search results and URL addresses and was able effectively to recognize what Web pages were useful.
Additionally, evaluation directs the information search process. Sutcliffe and Ennis (1998) presented a cyclic model of information search on the Internet. The cycle starts with problem identification, followed by needs articulation, query formulation, query reformulation, and evaluation. Furthermore, the model stresses the user's previous domain and device knowledge on information search activities.
According to Wu and Tsai (2005), Web users' evaluative standards of accuracy and usefulness of information affect their information search strategies in Web-based learning environments. The researchers conducted a survey (610 university students) and analyzed the data with structural equation modeling. Learners who evaluated the accuracy of information by using multiple sources and concentrated on the relevance of the content used elaboration as a search strategy. In the elaboration-strategy a learner tries to integrate information from multiple Websites to meet his or her information need. In contrast, if a learner has adopted a match strategy for searching for information, she or he is trying to find a few relevant Websites only. In the match strategy the learner evaluates the accuracy of information by reference to the authority of the Website and evaluates the usefulness of information according to technical features (such as easy retrieval).

Research Problems
The research problems addressed by this study are as follows: a) How do upper secondary school students allocate their working time among different kinds of actions on the Internet, such as browsing search results, reading texts, and navigating links? b) How do students evaluate the credibility and relevance of information on the Internet when searching for source material for an essay? c) What kinds of texts do students choose to read?

METHOD Participants
Students, 25 in total (14 female and 11 male) from an upper secondary school in Finland, volunteered to participate in the study. The participants were all born during the same calendar year and they were either 16 or 17 years old. A signed consent was sought and received from the parents of the students.

Task
The study was integrated into process writing practice in the mother tongue (Finnish) class. Students were asked to look for source material on the Internet for 40 minutes in order to write an essay on a set topic. The topic of the essay was: Sleeping as a human resource. Sleeping is an issue of daily life, and thus all the students had at least some prior knowledge of the topic. Students were asked to verbalize their thoughts when gathering their source material. They then had 45 minutes to write a first draft of their essay, producing their further drafts at a later date.
When commencing their search for source material on the Internet, the students were allowed to choose one of the three alternative browsers: Internet Explorer, Mozilla Firefox, and Opera. The students were informed that they were free to use all the features of the browser. In each browser the starting page was empty and the students had to decide how to start the search task.
The students were allowed to make notes during the search, but not print the pages. They were, however, asked to choose two documents, which were then printed for them so that they could use them as source material in the second phase of the writing process. The researcher reminded the students choosing the two documents approximately 10 minutes before the search time elapsed.

Data Collection
The study was conducted in the spring term of 2006. The researcher met each student individually. Each session lasted approximately 1 hour 45 minutes. The session started with a brief questionnaire. Students were asked for background information (age, sex) and information about their use of the Internet (number of hours weekly, familiarity with search engines, self-assessment of their information searching skills). After the students had completed the questionnaire the researcher described the task. The instructions were also given to the students in writing. After reading the instructions, students were asked to confirm whether they had understood the task. They were allowed to ask questions about the process.
The students were instructed to think aloud and report whatever they were thinking while searching for and reading information on the Internet. The students were informed that prompt questions (what is in your mind?; what are you thinking?) would be asked if they remained silent for a longer period. The researcher did not model the thinking aloud, as the aim was to study students' spontaneous thoughts. If students asked what was meant by thinking aloud, they were answered. The sorts of questions asked were "Do I say, that now I'm thinking that I'm going to use Google to find information?" or "Do you mean speaking to myself?" Ericsson and Simon (1993) argue that thinking aloud resembles a natural process common in daily life. After the search task the students were asked how thinking aloud affected their performance.
Web actions as well as students' "think-aloud" were recorded using Easy Video Capture software. Each session was replayed, transcribed, analyzed, and coded.

Think-Aloud Protocols
The "think-alouds" were transcribed and evaluation strategies were separated from the protocols for further analysis. Evaluation strategies were divided into two main categories: evaluation of credibility (see Harris, 1997) and relevance (see Saracevic, 1996). The subcategories of credibility and relevance are presented in Table 1. The examples in Table 1 are taken from the present data. Silent rejections of the Web page were not included in the analysis. The students (n = 24) rejected a page silently 110 times (M = 4.58, SD = 3.80). There was one exception among the students, a girl who silently rejected a page 49 times. The reliability of analysis of think-aloud protocols was examined by having two persons to classify 16% of evaluative comments (the think-aloud protocols of three students) and we found 85.9% agreement in our analyses.

Students' Internet Actions
Students' Internet actions were divided into seven categories: 1) formulating a query or using a URL address; 2) browsing search results; 3) downloading links and back-function; 4) reading texts; 5) skimming and rejecting the Web page; 6) navigating links; and 7) other. The time spent on each of these different actions was measured.

Texts Read by the Students
To begin with, we went through all the texts the students chose to read and formed a category for texts where the credibility of the publisher was somewhat questionable. Texts were classified into this category for the following reasons: information on the authority was not available on the Website; the text was written by an anonymous person, novice or peer group; or the Website was published by an ideological association or organization. The rest of the texts read by the students we divided into six categories according to the publisher of the Website in which the text was included (public association, expert, expert organization or network of experts, Wikipedia, company or trade organization, media, educational institute). The time a student spent reading texts in each category was measured.

Statistical Analysis
We conducted a hierarchical cluster analysis (Aldenderfer & Blashfield, 1984) to identify subgroups of students. The analysis was based on five variables: the number of credibility evaluations, the number of relevance evaluations, the sum of different evaluation subcategories each student used, time spent on reading texts, and the proportion of the total text-reading time spent on reading texts with a questionable publisher credibility. Standardized scores for the five variables were used in the hierarchical cluster analysis employing the between-groups linkage method to identify an appropriate number of clusters. The analysis was based upon the increasing value of the Squared Euclidean Distance between clusters. These are audio clips; the wrong kind of material.

Time Spent on Different Actions on the Internet
It was mentioned for the third time already in these pages, the increase in accident risk, so it must be important.
texts. However, there were considerable inter-individual differences. The five most effective Web-searchers spent more than 75% of their working time on reading texts, whereas the four weakest students spent so long searching for information that the time left to them for reading texts ranged from 30% to 45%. The individual variation can be variously explained. For example, some students had difficulties in formulating effective queries and some in selecting relevant Web pages from their search results. The latter problem can be considered as inability to predict the relevance of Web pages, which in turn has an effect on the number of Web pages first selected and then rejected as irrelevant.

Evaluation Strategies
As Table 3 shows, the evaluation of relevance (M = 16.9) was more frequent than the evaluation of credibility (M = 3.2). There was relatively wide variation among the students in the use of evaluation strategies. The evaluation of credibility ranged from 0-16 comments and the evaluation of relevance ranged from 6-43 comments. For example, six students did not evaluate the credibility of any Web page or text at all. Table 4 shows the frequencies of the subcategories of the credibility evaluations and the number of students who used the strategy in question at least once. Most of the evaluative comments (33%) expressed by the students concerned the publisher or author of the text or the expert interviewed. The second most common strategy was evaluation of the author's argumentation (opinion 10% and arguments 15%); however, this strategy was used by only six students. In fact, one student evaluated the author's argumentation 12 times, thus accounting for most of the comments in this category. Six students evaluated credibility on the 84 / KIILI, LAURINEN AND MARTTUNEN  basis of the style or content of the text. This category accounted for 10% of the credibility evaluations. Relatively few other strategies were used by the students. Predictive evaluation strategies are essential when selecting the most relevant Web pages from several search results. As shown in Table 5, most of the comments directed toward relevance were predictive in nature (31%) or concerned the evaluation of topicality (27%) either on the textual or paragraph level (17%). associations, such as the well-known Finnish child welfare association, accounted for the highest proportion of students' reading time (33.5%). The second mostread type of Website, accounting for 24.1% of reading time, were Websites published by an expert, expert organization or network of experts. This category, which can be regarded as the most reliable source of information for the students' purpose, included research-based Websites such as the Finnish Institute of Occupational Health or National Public Health Institute. The students relied quite heavily on information presented in Wikipedia, 16.5% of total text reading time. Wikipedia was used by 80% of the students.

The Publishers of the Texts Read by Students
The credibility of the publisher was somewhat questionable in nearly 10% of the Websites read by the students. Reading texts from the Websites with the questionable publisher credibility was not equally distributed across all students. Most of the time (76%) spent on reading these texts was accounted for by six students. Table 6 also presents the distribution of the texts students chose as source material, and it shows that 5 of the 50 chosen texts had somewhat questionable publisher credibility.

Evaluation Profiles
The cluster analysis indicated that the five-cluster solution was the most acceptable, as it best highlighted the salient differences between the students. Owing to the small number of students, the results of the cluster analysis must be interpreted with caution. The following five evaluation profiles were emerged (Table 7): versatile evaluators (n = 3), relevance-orientated evaluators (n = 5), limited evaluators (n = 8), disorientated readers (n = 5), and uncritical readers (n = 3). One student did not fit into any of the clusters.
The versatile evaluators used diversified evaluation strategies and their repertoire of strategies also included some cognitively demanding strategies, such as evaluation of argumentation and comparing two or more texts. The versatile evaluators used the time allocated to task effectively. They spent on average 73.7% of their time on reading texts. They mainly read texts from Websites that could be regarded as having rather credible publisher. The following excerpt illustrates the evaluation of a versatile evaluator. The excerpt is a good example of an ongoing evaluation process when the reader finds a very promising Website. The relevance-orientated evaluators concentrated on evaluating relevance. They did not evaluate the credibility of information as frequently as the versatile evaluators and the strategies they used were not as diversified. The students succeeded in spending 69% of the task time reading texts and they also mainly read texts on Websites where the publisher could be regarded as rather credible.

/ KIILI, LAURINEN AND MARTTUNEN
The limited evaluators hardly ever evaluated credibility. They evaluated relevance to a lesser degree than the versatile and relevance-orientated evaluators. They also read slightly more texts on Websites with questionable publisher credibility than the versatile and relevance-orientated evaluators.
The disorientated readers had difficulties in locating relevant information on the Internet, therefore they spent less time on reading texts than the other students. Furthermore, the disorientated readers spent more time on formulating queries (11.1% vs. 5.2%) and on browsing search results (29.1% vs. 14.3%) than the students in the other clusters. Additionally, they spent more time on skimming and then rejecting Web pages (10.4%) than other students (4.9%). One disorientated reader skimmed and rejected the same Web pages repeatedly. Another disorientated reader had difficulties in formulating effective queries. When she did not find relevant Web pages, she started reading her search results and making connective inferences between the example texts listed on the search results. In addition to showing disorientation, this student read her search results uncritically without looking at the context of the sentences at all.
The uncritical readers spent more time on reading texts on Websites where the publisher had questionable credibility (43.9%) compared to the other students. Although the uncritical readers used more evaluation strategies than the limited evaluators and disorientated readers, they were not able to apply them constructively. For example, when the uncritical readers opened an unauthenticated Web page they usually began to read it without paying any attention to the authority.
The student who did not fit into any of the clusters concentrated on the evaluation of credibility; he made 16 comments on credibility (the overall mean was 3.2) and he particularly evaluated the credibility of the author's argumentation (12 out of 16 comments). This student could be regarded as a reader with a critical disposition, which, in turn, differentiated him from the other clusters.

DISCUSSION
The results indicated, consistently with previous studies (Grimes & Boening, 2001;Metzger et al., 2003), that most of the students only seldom evaluated the credibility of the obtained information (M = 3.2) when using the Internet as a source of information in an authentic learning task. With respect to credibility, evaluation of the publisher or author or expert interviewed was the most frequently used strategy. When students are trying to locate useful Web pages from a huge amount of available information, the evaluation of relevance is obviously more necessary than the evaluation of credibility. Thus, the students evaluated relevance (M = 16.9) more frequently than credibility. When evaluating relevance, predictive evaluation and evaluation of the topicality of the text either on the textual or paragraph level were the most frequently used strategies. In this study cognitively demanding strategies, such as the evaluation of argumentation or comparing texts at the inter-textual level, were only seldom used.
It is reasonable to assume that predictive evaluation strategies are associated with effective access to relevant information (see MaKinster et al., 2003). In this study 27% of all the evaluative strategies were predictive in nature. However, they were not able to explain how effectively students were able to use the task time reading texts. One explanation for this could be that for some students predictive evaluation strategies are difficult to verbalize. More information is needed about the interrelationship between predictive evaluation strategies and effective use of the Internet. However, it was evident that the ability of some students to predict the relevance of Web pages was inadequate. These students did not consider the context of the search terms when browsing search results and might open similar irrelevant pages repeatedly. They were not able to reflect on their behavior and change their search strategies, or modify their ineffective queries. Whitmire (2004) found that the weakest students tended to choose the first hits from the search results because they assumed that search engines order the results according to their relevance and credibility.
Although the students did not frequently evaluate the credibility of information, most of the students succeeded in locating and reading texts which were posted by a rather credible publisher or author. The most frequently selected sources of information were public associations, expert organizations or networks of experts, as well as Wikipedia. The reading of texts where the credibility of the publisher was in doubt tended to be done by a small group of students.
Reading on the Internet is a complex process in which defining the need for information, searching for relevant information, and deploying evaluation strategies are interwoven. Students with ineffective search strategies may be compelled to use lower standards of evaluation. In this study some students changed their standards of evaluation during the task because they were unable to access better material.
Furthermore, the results of this study showed considerable individual variation between students when they were using the Internet as a source for an essay. Students differed in how much and how versatile evaluation strategies they used were, how effectively they managed to allocate the task time to reading and how much time they spent on reading texts where the credibility of the publisher was questionable. The study found five evaluation profiles: versatile evaluators, relevance-orientated evaluators, limited evaluators, disorientated readers, and uncritical readers. Additionally, one student, who did not fit into any of the profiles, could be regarded as a reader with a disposition toward critical thinking. He paid active attention to the author's argumentation; to cite Ennis (1987), he was "focusing on what to believe." Evaluation profiles may have a profound effect what reading material is selected on the Internet; this in turn, has an influence on the quality of learning.
The present study raises some points of interest that could usefully be addressed in future research. Because of the need for critical thinking skills when using the Internet as a source of information, it would be important to learn more about the relationship between different evaluation profiles and learning. In this study, versatile evaluators and persons able to read with a critical disposition were in the minority. It would be interesting to know how differently the same profiles would be distributed among university students and adults and whether, in particular, having a critical disposition would be more representative.
The think-aloud method is regarded as an effective method for gaining an access to on-line processing (Ericsson & Simon, 1993;Pressley & Afflerbach, 1995). In spite of its advantages, there are also some methodological concerns. In the first place, there are individual differences in students' abilities to think aloud (Pressley & Afflerbach, 1995, p. 12). For example, some students had to be prompted more than others. The timing of prompts (was a student reading a text or browsing search results at the time) might also have an effect on students' thinking-aloud. Moreover, some students might not be willing to think aloud. For example, one student, who was constantly concerned about the time remaining for the task, said: "Well, it is a little bit difficult to talk simultaneously; it takes all one's effort." Secondly, the think-aloud procedure might also encourage students to process material more actively and therefore thinking aloud might improve students' performance. In this study the students were asked afterwards about how thinking aloud affected their performance. One of the students answered: "When I was thinking aloud, somehow I was thinking deeper. When I am only browsing pages I am not thinking too much." In the main, however, the students felt that thinking aloud either did not affect their performance or that it impaired their performance. A few students reported that it was stress caused by the research situation which impaired their performance rather than thinking aloud per se.
Thirdly, some types of information are probably more representative than others in think-aloud protocols (Pressley & Afflerbach, 1995, pp. 9-10). In this study the students were not instructed to concentrate on evaluation and therefore some evaluative thoughts they might otherwise have had may have been excluded from examination. For example, the students evaluated some Web pages as irrelevant very quickly, in one or two seconds. This was especially the case when it was obvious that the Web page did not relate to the task at all. Thus, the reasons for these kinds of fast decisions were not spelled out. In the future, evaluation strategies could be studied by asking students to give reasons for their decision to read or to discard a Web page. It is, however, possible that these kinds of instructions might direct students' thinking. It should perhaps be emphasized in defense of this study that the aim was to examine students' spontaneous attention to evaluation.
In spite of its methodological limitations, the useful outcome of this study was the development of a coding scheme for the evaluation of Web sources, which can be taken advantage of in future research or teaching students evaluation skills.
The study also provided descriptive information for teachers about the problems students encounter when using the Internet as an information source.
Traditionally students have read text books and articles selected by the teacher and thus, the appropriateness of the contents has been checked in advance. When using the Internet in schoolwork, there is a risk that some students may not be able to use the Internet effectively and might read and extract irrelevant or even misleading information. On the other hand, teachers might not be aware of the problems students encounter, because students are confident about their skills in using the Internet (Macklin, 2001, p. 306;Pierce, 1998). In this study as well, the disorientated and uncritical readers might have had an erroneous impression of their skills. They reported in the pre-questionnaire that their information searching skills are quite good, although in fact they had poor literacy skills.
This study showed that using Internet as an information source is a demanding task and students need to be scaffolded to be able to evaluate information critically. The use of individual or collaborative scripts could be one means for supporting students. Collaborative scripts have previously been applied, among other uses, to support reading comprehension (see Kollar, Fischer, & Hesse, 2006). Evaluation strategies can be considered epistemic scripts, as according to Weinberger, Ertl, Fisher, and Mandl (2005), epistemic scripts can guide students' attention toward critical aspects of the task. Epistemic scripts are the kind of task strategy which could also be applied to individual learning (ibid). By means of epistemic scripts students could be guided to pay attention, for example, to the authority of Websites and the purposes of the author. Stadtler and Bromme (2007) found that evaluative prompts received during Internet reading helped students to judge information critically. However, more research is needed to discover whether prompting has long-term effects on students' performance. Students rely frequently on the Internet in their schoolwork and therefore should be taught to be responsible in the use of information; in Salomon's (1998, p. 10) words: It is school's mission, particularly when colored by constructive hues, to enable students to think and to acquire skills of intelligently handling information, not to drown them in it.