Introduction

With World Wide Web (commonly known as the Web, a system that is interlinked and accessed via the Internet, contains text, images, videos, and multimedia) becoming accessible to everyone and with its abundant and diverse resources, it offers an adequate context for science students to conduct inquiry so as to promote their knowledge construction and meaningful learning (Butler and Lumpe 2008; Tsai et al. 2012). The act of search can be both a learning experience about the content and about the process for future endeavors. The design of searching tasks can be either unstructured or highly scaffolded. The former relies heavily on the user characteristics in the process, which may be aligned with many classroom implementations of search (e.g., the searching activity of the present study). The latter can be intentional activities that not only promote activation of prior knowledge and development of strategies (Greene et al. 2010), but also foster metacognitive awareness (Azevedo et al. 2004). Previous studies have indicated that students’ online searching strategies (Lin and Tsai 2008) and criteria to evaluate online information (Mason et al. 2010) are guided by their epistemic beliefs. Epistemology, originating from Piaget’s theories of cognitive development and Perry’s studies of students’ intellectual development, exists in a form of beliefs about how a person views the nature of knowledge and the process of knowing (Hofer and Pintrich 1997). Hofer and Pintrich (1997) further suggested four dimensions to better represent the core structure of individuals’ epistemic beliefs, including Certainty of knowledge, Simplicity of knowledge, Source of knowledge, and Justification for knowing. The first two refer to the nature of knowledge, while the last two represent the process of knowing.

A close relationship between epistemic beliefs and metacognition is widely recognized by researchers (Hofer 2004; Mason et al. 2010; Tsai 2004b). Kitchener (1983) proposed a three-level model of cognitive processing to explain how people solve ill-structured problems (e.g., the dispute over the safety concern of electromagnetic waves). Consisting of cognition, metacognition, and epistemic cognition, the first level of the model refers to cognitive processes such as memorizing, reading, and acquiring basic information. A question one may ask at this level is what is an electromagnetic wave? The second level includes metacognitive processing such as monitoring strategies or progress in cognitive tasks of the first level (i.e., cognition). One at this level may prompt how do I search for information related to electromagnetic waves effectively? The third level, epistemic cognition, involves reflection on the certainty of knowledge and the criteria of knowing. Questions one may ask include is the information credible? Or is there any alternative solution? Kitchener further indicated that each level offers a foundation for the next, and the last level, epistemic cognition, may influence monitoring processing or strategies adopted in the tasks of the first two levels. According to Hofer (2004), it is essential for more research to investigate the epistemic processes while students are conducting online scientific inquiry and activating metacognitive awareness during their knowledge construction.

In addition, researchers of epistemology concur that individuals’ epistemic beliefs are related to their experience in disciplinary contexts (Buehl et al. 2002), that is, epistemic beliefs regarding the science domain, for instance, could vary from those regarding history. Thus, in order to investigate students’ strategies and behaviors when searching for online science information, it is necessary for researchers to explore learners’ scientific epistemic beliefs (SEBs). Some research efforts have indicated that students’ epistemic orientations toward science may guide the acquisition of scientific information on the Web (Lin and Tsai 2008; Mason et al. 2010), that is, students who view scientific knowledge as more dynamic in nature tend to employ a more comprehensive evaluation of the online information (Lin and Tsai 2008).

In recent years, there have been a growing number of studies highlighting the utilization of socioscientific issues to promote students’ science learning. These issues refer to social dilemmas that include moral or ethical problems as well as having conceptual or technological connections with science (Sadler 2004). Due to being open-ended and ill-structured in nature, socioscientific issues require one to take multiple perspectives or solutions into consideration when encountering these issues (Sadler and Zeidler 2005). This process not only enables learners to actively practice evaluating, analyzing, and reflecting on information, but also engages them in decision making and justifying claims (Sadler 2004). In addition, educational researchers (Mason and Boscolo 2004; Schommer-Aikins and Hutter 2002; Yang and Tsai 2010) believe that the effects of individuals’ epistemic beliefs become obvious when they are dealing with open-ended and ill-structured problems requiring the application of high-order reasoning and reflective thinking. Similarly, many studies also suggested that the students’ reasoning in socioscientific issues is guided by their epistemic beliefs (Liu et al. 2011; Mason and Boscolo 2004; Schommer-Aikins and Hutter 2002). Since the Internet offers abundant and diverse resources as well as supporting claims of various perspectives, online searching activities are an ideal and potential context for students to explore socioscientific issues (Wu and Tsai 2011).

To profile students’ cognitive and metacognitive strategies during online information searching, Tsai and Tsai (2003) proposed a framework classifying the searching strategies into three domains, they are follows: behavioral, procedural, and metacognitive domains. The behavioral domain described skills required for basic Internet manipulation and navigation, including control and disorientation aspects. The procedural domain indicated content-general searching approaches on the Internet, including trial and error as well as problem-solving aspects. The metacognitive domain concerned with skills involved in higher-order and content-related reflective on the Internet, including purposeful thinking, select main ideas, and evaluation. Based on this framework, Tsai (2009) developed an instrument, namely the Online Information Searching Strategies Inventory (OISSI), to investigate students’ online information searching strategies. Although previous studies have noticed intimate relationship between students’ online information searching strategies and their SEBs (Lin and Tsai 2008; Mason et al. 2010), rarely does the research investigates differences in online searching strategies and behaviors in terms of different levels of SEBs. Thus, underlying Tsai and Tsai’s (2003) framework and using the OISSI instrument (2009), the present study attempted to explore the role that the students’ SEBs played in online searching activities while exploring a socioscientific issue. The findings were expected to provide information for practitioners and researchers to improve students’ abilities and understanding of scientific inquiry. The purpose of this investigation was to investigate the following questions:

  1. 1.

    Is there any significant difference in students’ online information searching strategies between sophisticated and naïve SEB students?

  2. 2.

    Are there different patterns of online information searching behaviors for different SEB students while exploring socioscientific?

Methodology

Participants

This study initially recruited 240 undergraduate and graduate students from four universities in northern Taiwan. These students were studying in colleges of science (30 %), electrical engineering and computer science (16.1 %), liberal arts and social sciences (46.9 %), and management (7 %). In all, 46 % of the students were science-oriented majors, while the remaining 54 % were social science-oriented majors. This percentage corresponds to that of the whole population (43 % in science-oriented majors; 57 % in social science-oriented majors) in Taiwan (Ministry of Education 2012). All the participants were requested to fill out the SEBs survey developed by Conley et al. (2004) (described later). The students with total scores within the top 25 % were defined as sophisticated SEB searchers, while those with full scores within the bottom 25 % were considered as naïve SEB searchers. Among the students in both groups, 42 volunteered to participate in this study and were given an appropriate participation fee as compensation for their time. Finally, 22 (13 females and 9 males) high (sophisticated) and 20 (11 females and 9 males) low (naïve) SEB students were selected in an online searching activity to explore a socioscientific issue.

The SEB survey includes 26 items on a 5-point scale, and its full score is 130. The range of SEB scores in this study was from 75 to 124. In our study sample, the participants expressed relatively high agreement with the SEB questionnaire statements. The mean scores are 117.2 (SD = 3.9) for the high SEBs and 93.7 (SD = 5.8) for the low SEBs. It should be noted that the categorization of the high (sophisticated) and the low (naïve) SEB groups was based on their comparative scores on the questionnaire only within the study sample. The participants (n = 35, 83 %) self-reported that they used the Internet almost every day. The average time for each Internet usage was more than 1 h (n = 39, 93 %). The participants searched for online information at least once a week, and 62 % of the students (n = 26) performed online searches at least three times a week. Thus, it could be supposed that the participants of this study had sufficient experience in performing online-related activities.

Instruments

This study used Conley et al.’s (2004) SEB questionnaire to assess students’ SEBs. Based largely on Hofer and Pintrich’s (1997) framework, the instrument focuses on the dimensions of epistemic beliefs regarding science, which include Source (a sample item is, “Everybody has to believe what scientists say.” [scored in reverse]), Certainty (“All questions in science have one right answer.” [scored in reverse]), Development (“Ideas in science sometimes change.”), and Justification (“It is good to try experiments more than once to make sure of your findings.”). As Conley et al.’s study presented the statements of the Source and Certainty dimensions as less advanced SEB perspectives (i.e., assessing the students’ agreement with the authority and certainty of scientific knowledge), these two factors were reversed so that, for each factor, higher scores reflected more advanced SEBs. The questionnaire consists of 26 items rated on a 5-point Likert scale (1 = strongly disagree; 5 = strongly agree).

Conley et al.’s questionnaire was originally designed for fifth graders. However, the instrument has been applied to assess different sample subjects, such as measuring undergraduate students’ SEBs in the studies of Liang et al. (2010) as well as in Liang and Tsai (2010). Due to differences in the targeted participants, a series of processes was conducted to examine its reliability and validity, which included translation of the instrument and reviewing, approving, and verifying with two science education experts. The overall coefficients of internal consistency reliability (Cronbach’s alpha) for the SEB survey are 0.81 and 0.80, reported, respectively, in Liang et al.’s and Liang and Tsai’s studies, showing the reasonable reliability of this instrument. Using undergraduate and graduate students as the participants, the present study adopted Liang et al.’s modified questionnaire. The alpha reliability for the total scale was 0.79, suggesting a sound reliability in assessing the students’ SEBs.

The instrument, OISSI, was administered to assess the participants’ self-perceived online searching strategies. Its theoretical framework was established in Tsai and Tsai’s (2003) study, which used a multiple-case study followed by cross-case comparisons (see Tsai and Tsai 2003 for more detailed information) to profile students’ cognitive strategies for conducting a Web search task into seven aspects (Control, Disorientation, Trial and Error, Problem-Solving, Purposeful Thinking, Selecting Main Ideas, and Evaluation) categorized into three domains (Behavioral, Procedural, and Metacognitive). Based on this framework, Tsai (2009) further developed and validated the OISSI instrument that consists of seven subscales corresponding to the seven search strategy aspects. The OISSI includes a total of 25 items measured by a six-point Likert scale (1 = not like me at all; 6 = very much like me). The Cronbach’s reliability coefficient was 0.91 for the total scale and ranged from 0.64 to 0.88 for the seven subscales (Tsai 2009), which was good enough for testing. In the current study, the reliability was 0.92 for the total scale and ranged from 0.67 to 0.91 for the seven subscales. Also, the reliability coefficients for the Behavioral, Procedural, and Metacognitive domains were 0.87, 0.75, and 0.93, respectively. Table 1 shows the sample items for the seven aspect strategies categorized according to the three domains of the OISSI.

Table 1 Seven aspect strategies categorized into three domains on the OISSI
Full size table

Data Collection Procedure

The targeted participants (42 volunteers in either a high or low SEB groups) were brought into the laboratory individually and were first asked to read two news items regarding a scientific dispute about the risk of electromagnetic waves. One news item holds a position indicating that electromagnetic waves are dangerous, while the other considers them to be safe. The searching task was unstructured and required the students to justify their positions on this issue through seeking online information as well as answering two questions including:

  1. 1.

    After searching the Web for information, which of the two news items that you read before the task do you trust more? Why?

  2. 2.

    Why did you change or why do you insist on your position? What problem(s) may the position contrary to yours have?

The participants entered their responses in the left panel of the browser. No time limitation was imposed on the searching task. Since this task fulfills the characteristics of a socioscientific issue (e.g., open-ended, ill-structured, involving a social dilemma having conceptual or technological connections with science in nature), it can promote one to search online information and trigger epistemic process by actively evaluating, analyzing, reflecting, decision making as well as justifying while conducting this task. Each student’s searching process (including Web sites browsed sequentially, as well as links or buttons clicked) was recorded by the screen-capture software, Camtasia. Upon completing the task, they had to fill out a survey, the OISSI, to assess their self-perceived searching strategies.

Data Analysis

Based on the SEB survey, the participants were divided into high (sophisticated, top 25 %) and low (naïve, bottom 25 %) SEB groups. After the searching tasks, the two types of data collected, including searching strategies (via the OISSI) and searching behaviors (via the screen-capture software), were set as dependent variables and analyzed in terms of the students’ different SEB levels. To make a comparison between the high and low groups, independent t tests (two-tailed) were adopted to examine the differences in searching strategies, whereas sequential analysis was used to compare the searching behaviors of both groups, that is, the main purpose was to examine the role of students’ SEBs in their searching strategies and searching behaviors.

A sequential analysis, according to Bakeman and Gottman (1986), provides a systematic observation for researchers to understand how behaviors are sequenced moment to moment so that they can investigate the dynamic aspects of interactive behaviors. The merit of this approach is that it enables researchers to more accurately examine whether sequential relationships among searching behaviors reach a statistically significant difference. Sequential analysis has been widely utilized in numerous studies, such as investigating behavioral patterns of online collaborative discussion (Hou 2011), analyzing sixth graders’ sequential patterns of using mobile devices in a museum-learning context (Sung et al. 2010), exploring students’ navigation in a hypermedia program (Rezende and de Souza Barros 2008), and examining message response exchanges in online group debates (Jeong 2003).

In this study, the participants’ video records were analyzed through sequential analysis (Bakeman and Gottman 1986) to visualize their online information searching process. In the initial stage of conducting the sequential analysis, the participants’ behaviors during the online searching task were first categorized and then a coding scheme was developed according to the categorization. In the present study, the coding scheme, including seven main behaviors, is presented in Table 2. Based on this scheme, each participant’s video record was coded in chronological order by a researcher who had completed a video analysis training course. For instance, a user has a certain sequence of QRWB during a certain period of time, indicating that he/she first enters a query (Q) and then browses the result page (R). Upon finding a relevant page, the user clicks its hyperlink to read the information on the page (W) and finally adds it as a bookmark (B). After coding, 42 sets of data with 1 057 behaviors were gathered. We calculated each participant’s total transfer frequencies from one behavior to another (this study analyzed Lag 1 sequence) according to lag sequential analysis method (Bakeman and Gottman 1986). Through a series of computations of the sequence transfer matrices (including the frequency transfer matrix, the conditional probability transfer matrix, and the expected value matrix), the computation of adjusted residuals (z-scores) was used to identify those transitional probabilities that were significantly higher or lower than the expected probability (Bakeman and Gottman 1986). Since this study more focuses on exploring the occurrences of sequences reaching the level of significance, we only investigated z-scores greater than +1.96. A significant z-score indicates a significant occurrence of behavioral sequence. This method of analysis has been adopted in previous studies, such as investigating the pairs of students’ interactions in tutoring (Duran and Monereo 2005) and exploring group interaction and critical thinking in online threaded discussions (Jeong 2003).

Table 2 Coding scheme of online searching behaviors
Full size table

Based on Tsai and Tsai’s (2003) framework of classifying online information searching strategies, the results of sequential analysis were further examined to probe whether ones’ searching behaviors (sequences with statistically significant difference) confirmed to their searching strategies.

Results

The participants spent around 20 min on completing the searching task. No statistical difference between the two SEB groups in the time spent on the task was identified. The following sections, respectively, display the results of the comparison regarding the high and low SEB students’ searching strategies and searching behaviors.

Exploring Students’ Searching Strategies in Terms of Different Levels of SEBs

Table 3 displays the results of comparing the students’ OISSI scores by the sophisticated and naïve SEB groups. As shown, of the three domains in the OISSI, two statistically significant differences were identified in the Behavioral (t = 3.60, p < 0.05) and Metacognitive domains (t = 2.05, p < 0.05). The advanced SEB searchers outperformed the naïve searchers in both domains. This implies that those with sophisticated SEBs were more likely to have superior skills in Internet manipulation or navigation than those with naïve SEBs. Further, they tended to employ more reflective and higher-order cognitive strategies. Regarding the seven aspect strategies in the OISSI, three out of the seven were found to have statistically significant differences, including Control (t = 3.20, p < 0.05), Disorientation (t = 2.65, p < 0.05), and Purposeful Thinking (t = 2.31, p < 0.05). Similarly, the students with sophisticated SEBs outperformed those with naïve SEBs in terms of these three factors. This suggests that the advanced SEB searchers were inclined to express more ease with manipulating the online searching application than the naïve SEB searchers. They were less likely to feel confused and disoriented while searching for online information. In addition, they tended to remind themselves of their purpose during the searching process.

Table 3 Comparisons of the OISSI scores for the sophisticated and naive SEBs
Full size table

Exploring Students’ Behavioral Patterns in Terms of Different Levels of SEBs

Table 4 refers to the adjusted residuals table that offers z-scores information. These values were computed to identify whether the sequential relationships among the advanced SEB students’ searching behaviors reached statistically significant difference. As shown in Table 4, the rows represent initial searching behaviors, while the columns refer to the follow-up behaviors. A value that is greater than positive 1.96 suggests that the continuity of the sequence reaches the level of significance (p < 0.05). According to Table 4, nine significant sequences are identified, including Q (enter a query) → R (browse the result page), W (browse Web information) → B (add a bookmark), B (add a bookmark) → R (browse the result page), R (browse the result page) → Q (enter a query), R (browse the result page) → W (browse Web information), R (browse the result page) → N (click on the “next page” hyperlink to the next page of the results), A (answer the question) → A (answer the question), N (click on the “next page” hyperlink to the next page of the results) → R (browse the result page), and P (click the “previous” button of the browser) → W (browse Web information). To better visualize the connections, these sequences are further illustrated in Fig. 1. As displayed, the sophisticated SEB group demonstrated two bidirectional sequences: Q ↔ R (i.e., the participants tended to browse the result page after keying in the keywords, and they may also consider other keywords to make further searches after browsing the result page.) and R ↔ N (i.e., the participants tended to click on the “next page” hyperlink to read more search results after browsing the result pages.) The former implies a series of recurrent behaviors in which the searchers might try different queries and scan the result pages. The latter suggests that the students tended to explore multiple sources and browse the results. It was reasonable that the more the students refined their queries and assessed the results, the more chances they would have of getting more relevant Web pages to browse later.

Table 4 Adjusted residuals table for sophisticated SEB achievers’ online searching behaviors
Full size table
Fig. 1
figure 1

Sequential patterns of the sophisticated SEB group. Q: enter a query; W: browse Web information; B: add a bookmark; R: browse the result page; A: answer the question; N: click on the “next page” hyperlink to the next page of the results; P: click the “previous” button of the browser

Full size image

In addition, the sequence P (click the “previous” button of the browser) → W (browse Web information) presents the behavior of clicking the previous button on the browser and browsing the Web content. It is possible that the searchers tended to re-browse the information in order to identify nuances between the Web sites. With regard to A (answer the question) → A (answer the question), since there was no sequence connected to code A (answer the question), it is reasonable to assume that students with sophisticated SEBs were likely to answer the questions continuously without the interruption of performing other behaviors. This implies that they might get the whole picture of the searching task before responding to the questions.

Regarding the naïve SEB group, eight values with statistical significance are identified in Table 5, consisting of Q (enter a query) → R (browse the result page), W (browse Web information) → B (add a bookmark), B (add a bookmark) → R (browse the result page), R (browse the result page) → W (browse Web information), A (answer the question) → Q (enter a query), A (answer the question) → A (answer the question), N (click on the “next page” hyperlink to the next page of the results) → R (browse the result page), and N (click on the “next page” hyperlink to the next page of the results) → N (click on the “next page” hyperlink to the next page of the results). These sequences are further illustrated in Fig. 2. As shown, the unidirectional sequence Q → R represents that the participants browsed the result page (R) after entering a query (Q). This indicates that the students tended not to try different queries. In addition, the sequence N (click on the “next page” hyperlink to the next page of the results) → N (click on the “next page” hyperlink to the next page of the results) shows that the students simply kept clicking the “next page” hyperlink of the result pages without pausing to read the information they had found. This might suggest that the participants felt impatient or disoriented to a certain degree so that they performed the searching task carelessly. Further, while answering the questions, the sequence A (answer the question) → Q (enter a query) shows that the naive SEB students tended not to answer the questions continuously, that is, they were likely to seek relevant information for each question separately, rather than judging and analyzing all the information until they could confidently answer all of the questions.

Table 5 Adjusted residuals table for naïve SEB achievers’ online searching behaviors
Full size table
Fig. 2
figure 2

Sequential patterns of the naïve SEB group. Q: enter a query; W: browse Web information; B: add a bookmark; R: browse the result page; A: answer the question; N: click on the “next page” hyperlink to the next page of the results

Full size image

In summary, the students with high SEBs tended to try various queries (i.e., Q ↔ R), whereas those with low SEBs rarely did. Second, after entering their queries, the high SEB students were inclined to explore the result pages frequently (i.e., R ↔ N), implying the behavior of looking through multiple sources, while the low SEB students were less likely to browse the result pages, or browsed them carelessly (i.e., N ↔ N and N → R). Third, when surfing the Web content, the high SEB students tended to constantly compare the information among the Web sites by clicking the previous button of the browser (i.e., P → W), while the low SEBs did not display this behavior. Finally, there is some evidence that the low SEB students seemed to answer the questions separately and incomprehensively (i.e., A ↔ Q). In general, the behavior analysis supported that the high SEB students tended to display more recurrent searching behaviors and more in-depth exploration of multiple sources, implying the usage of better metacognitive acts.

Discussion and Implications

Conducting online information searching tasks has become a very common learning activity when the Internet is integrated into science classrooms (Kim et al. 2007; Songer et al. 2002). However, students are challenged with effectively and critically retrieving, evaluating, selecting, judging, and integrating information gathered from the Internet (Tsai 2004a; Walraven et al. 2008). According to the previous studies, students’ online searching strategies (Lin and Tsai 2008) and behaviors (Mason et al. 2010) are guided by their epistemic beliefs, which may vary in different disciplinary contexts (Buehl et al. 2002). This study explored the role of students’ SEBs in their online search strategies and behaviors. In addition, limited research has been conducted to utilize sequential analyses to visualize the searchers’ online information-seeking behaviors or to embed socioscientific issues into the searching tasks. Thus, in addition to using a survey, this study also adopted sequential analysis as a way to examine the searching behaviors and probed how the students’ SEBs might guide their reasoning in online searching activities that explore a socioscientific issue.

The results of this study show that, in comparison with students with naïve SEBs, those with more sophisticated SEBs perceived themselves as applying more metacognitive strategies in their online searching, that is, the more advanced SEB searchers perceived themselves as using more high-order search strategies, such as self-reflections and self-monitoring on the goals and process of searching. This finding responds to the results of Muis’ study (2007) that a person’s epistemic beliefs can transform into epistemic standards that foster a metacognitive monitoring process. In fact, our searching behavior analysis also indicated that the high SEB students had more recurrent searching behaviors and more in-depth exploration of multiple sources, suggesting the usage of better metacognitive acts. Thus, it is likely that during the online information searching activities, those with advanced SEBs might believe in complex and tentative knowledge, which guided them to contemplate metacognitive ideas such as “how can I find supportive evidence effectively and efficiently.” It was this process that ultimately drove them to purposefully filter the online information. This finding is in line with the previous studies (Lin and Tsai 2008; Tu et al. 2008), suggesting that the students of more advanced epistemic beliefs might adopt more sophisticated evaluative standards and strategies to judge online information.

In addition, this study also found that those students who held advanced SEBs were less likely to perceive confused and disoriented while searching for online information. Similar to the aforementioned description, the evaluative standards and strategies the sophisticated SEB searchers held might steer them purposefully and metacognitively to seek online resources of concern to them, which might relieve their sense of disorientation.

Through examining the students’ video records, the results from the sequential analysis found that those students who held more advanced SEBs provide some evidence of metacognitive searching behaviors such as refining queries, filtering the results, and comparing relevant information among Web sites. For instance, the high SEB students tended to try different queries and evaluate multiple sources. These behaviors, according to Kitchener’s (1983) model of cognitive processing, represent how students with sophisticated SEBs metacognitively monitor searching strategies or processes to seek relevant information. However, the naïve SEB searchers tended not to try different queries and browsed the search result pages superficially. These findings are resonant with the previous studies (Kitchener 1983; Mason et al. 2010), indicating the important role that individuals’ epistemic beliefs play in guiding their metacognitive and cognitive processing. In addition to searching behaviors, the present study also found that when replying to the questions, the students with sophisticated SEBs were inclined to get the whole picture of the information they had found before responding. Conversely, the naïve SEB searchers tended to answer the questions separately, rather than integrating the information until they could confidently answer both questions. This finding is in line with the prior studies (Muis 2007; Tsai 2000), suggesting that students who have more advanced SEBs tend to adopt more meaningful learning strategies such as elaboration and integration of the target information.

In addition, future researchers should also be aware of students’ engagement in searching tasks. For example, if they do not take the task seriously, they are not likely to reflect typical knowledge acquisition strategies. Therefore, embedding searching activities into their subject learning can be a beneficial way to enhance students’ engagement. Furthermore, it is possible that learners’ characteristics (e.g., intelligence quotient or prior knowledge of the searching tasks) may have an impact on students’ searching strategies and behaviors. Thus, it is suggested that future studies pay attention to learners’ characteristics while implementing similar research.

The current study found that the students’ epistemic orientations toward science play an important role in their online searching strategies and behaviors regarding science information. It is essential for future researchers and educators to take the role of SEBs into consideration when designing online searching activities. For instance, some possible suggestions that may metacognitively engage students in searching activities include offering a pop-up window that guides students’ reflection or evaluation on the information, or providing adaptive scaffolding. Take Puntambekar and Stylianou’s (2005) study for example; they designed meta-navigation prompts according to the searchers’ navigation to support their online information searching activities, which had a positive impact on the implementation. Future studies, in addition to assessing searching performance, can also investigate the influence of implementing searching assistance on improving the participants’ SEBs. The present study utilized a socioscientific issue to engage the searchers in the online searching tasks. Thus, for the researchers of socioscientific issues, how the students conduct argumentation and develop informal reasoning may arouse their interest. In addition, one may question whether using a questionnaire to examine the participants’ SEBs in the present study might risk not accurately representing the participants’ SEBs, although this approach has been utilized in many studies (Liang et al. 2010; Tu et al. 2008). Therefore, future studies are encouraged to utilize a more complicated method, such as the think-aloud approach (Mason et al. 2009) to probe the participants’ SEBs.

In conclusion, promoting scientific inquiry through online information searching activities has become a growing trend in schools. Implications drawn from this study include that future research can integrate online information searching tasks into inquiry-based science curricula so as to probe the students’ searching strategies and behaviors. Similarly, one can also investigate the role the searchers’ SEBs play in their searching strategies, behaviors, and outcomes. Further, how the students’ searching outcomes map onto their searching behaviors and how the participants’ perceived behaviors map onto their observed behaviors are also worth investigation.