Tool use in computer-based learning environments: towards a research framework

Tool use in computer-based learning environments: towards a research framework

Computers in Human Behavior Computers in Human Behavior 22 (2006) 389–411 www.elsevier.com/locate/comphumbeh Tool use in computer-based learning envi...

269KB Sizes 1 Downloads 28 Views

Computers in Human Behavior Computers in Human Behavior 22 (2006) 389–411 www.elsevier.com/locate/comphumbeh

Tool use in computer-based learning environments: towards a research framework Geraldine Clarebout *, Jan Elen Center for Instructional Psychology and Technology, University of Leuven, Vesaliusstraat 2, B-3000 Leuven, Belgium Available online 11 November 2004

Abstract Computer-based learning environments often confront learners with a number of tools, i.e. non-embedded support devices. Such environments assume learners to be good judges of their own learning needs. However, research indicates that students do not always make adequate choices for their learning process. This especially becomes an issue with the use of open learning environments, which are assumed to foster the acquisition of complex problem solving skills. Such open learning environments offer students tools to support their learning. Consequently, it is needed to understand factors that influence tool use and acquire insight in learning effects of tool use. Both issues are addressed in this contribution. A review of the existing literature has been undertaken by performing a search on the Web of Science and the PsycInfo database. Results indicate that there is some evidence for learner, tool and task characteristics to influence tool use. No clear indication was found for a learning effect of tool use. The conclusion proposes a research framework for the systematic study of tools. Ó 2004 Elsevier Ltd. All rights reserved. Keywords: Computer-based learning environments; Tool use; Literature review

*

Corresponding author. Tel.: +32 16 32 5745; fax: +32 16 32 6274. E-mail addresses: [email protected] (G. Clarebout), [email protected] ac.be (J. Elen). 0747-5632/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.chb.2004.09.007

390

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

Computer-based learning environments regularly provide learners with a variety of support devices to foster learning. These support devices can be either embedded or non-embedded. Embedded support devices are totally integrated in the learning environment and cannot but be considered by learners. Examples of such devices are feedback, or the information structure in learning materials. Embedded support devices are devices with which learners are confronted without them having to request or ask for them. By contrast, non-embedded support devices are support devices whose use depends on the learnerÕs initiative. They are added to the environment and it is up to the learners to decide on their use. Non-embedded support devices are also called ‘‘tools’’. A tool could for instance be a button that enables the learner to access additional information. The learners have to take action; they have to click on the button before receiving additional information. In this contribution, the latter kind of support devices, namely tools, are addressed. Given that the use of non-embedded support devices (tools) depends on the learnerÕs action, the integration of tools in learning environments presupposes, by definition, that learners are good judges of their learning needs. Based on their judgments, learners select tools when they need them. Learners control the use of tools. Contrary to the assumptions that both providing learner control and allowing learners to co-construct their learning environment establishes a ‘‘better’’ learning environment, a clear benefit of learner control on learning has not yet been found (see reviews by Friend & Cole, 1990; Goforth, 1994; Large, 1996; Williams, 1996). Most learner control studies report a positive effect on learnerÕs attitude, whereas learning effects seem clearly mediated by various student characteristics. Basically, these reviews conclude that commonly students experience difficulties to make adequate choices for themselves (see also Clark, 1991; Hill & Hannafin, 2001; Land, 2000; Lee & Lehman, 1993; Milheim & Martin, 1991), i.e. choices beneficial for their learning process. In other words, in an instructional context students seem to lack self-monitoring and regulating skills. Applied to tools, it is reasonable to expect that students will have problems to determine when they need help, what kind of help they need and hence, when the use of tools might be beneficial. A recent evaluation study indirectly validated this expectation (Clarebout, Elen, Lowyck, Van den Ende, & Van den Enden, 2004). In this study, students did not use the available tools when working on a diagnostic problem in a computer-based learning environment. Thinking aloud protocols revealed that students thought they would be cheating if they used the tools. In other words, studentsÕ instructional conceptions hampered studentsÕ judgment about the use of these tools. Similar results were published by Marek, Griggs, and Christopher (1999) for adjunct aids in textbooks. StudentsÕ conceptions about adjunct aids influenced the use of these aids, students indicated to be less inclined to use those adjuncts aids promoting a more elaborative study pattern. These and similar studies raise doubt about the assumption underlying the ample use of tools in learning environments: their use cannot be taken for granted. At the same time however, from a constructivist view on learning, the use of open learning environments is advocated to foster the acquisition of complex problem solving skills

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

391

(Jacobson & Spiro, 1995; Jonassen, 1997). These open learning environments confront learners with a complex or ill-structured task to be solved by looking at different perspectives (Spiro, Feltovich, Jacobson, & Coulson, 1991). In open learning environments students are in control and get different tools to their disposal to engage in problem solving tasks. As a consequence, it becomes crucial to gain insight in (variables influencing) tool use. It might even be wondered whether these tools are actually used and if they are used, whether they are used as intended. Similar to research on adjunct aids, it can be hypothesized that different variables will mediate the learning effect of tool use. Elen (1995) provides an overview of different variables mediating the effect of adjunct aids such as the learning task, the nature of the adjunct aid and, whether and when learners are urged to use the adjunct aid. In this contribution, these variables will be addressed with respect to tool use in computer-based learning environments. Additionally, the learning effect of tool use will be addressed. Through means of a literature study, an overview is presented of research on tools in computer-based learning environments. 1 This overview is structured according to questions relating to different variables that might mediate the effect of tool use on learning. First, the methodology will be discussed. Next, results are presented and finally these results are reflected on. The conclusion offers possible solutions and suggestions for further research.

1. Method This literature study started with a search on the Web of Science and in the PsycInfo database. 2 These databases were searched for the last 20 years (from 1982). It can be argued that 20 years is a rather large interval for studies in computer-based learning environment, given the evolutions in this domain. However, this allows to consider also tool use in less complex learning environments and to see whether the complexity of the learning environment plays a role in learnersÕ tool use. Descriptors (see Table 1) specifically relate to the use of tools or synonyms (options, adjunct aids), and to environments in which tools are most likely made available (open learning environments, hypermedia environments). These descriptors were the results of a brainstorm session by two researchers. Additionally, the initial results of this search were presented on two conferences (Clarebout & Elen, 2002a, 2002b). The suggestions raised by the audience were taken into account and entered in a new search. In all searches, the term ‘‘research’’ was added, since the aim of this study was to find

1

This contribution does not deal with the computer as a tool in itself, or more specific as a cognitive tool (see Lajoie, 2000; Lajoie & Derry, 1993; Salomon, 1988). 2 The search was performed June 2003 and updated in September 2004.

392

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

Table 1 Descriptors Descriptors

Web of science (SSI)

PsycInfo

Option(s) use Use of option(s) Tool(s) use Use of tool(s) Open learning environment(s) Electronic learning environment(s) Hypermedia Learner control Instructional intervention(s) Adjunct aid(s) Discovery learning Use of resource(s) Resource(s) use Inquiry-oriented instruction Project-based environment(s) Computer-assisted learning Simulation* Help use Use of help Scaffolds Powerful learning environment(s) Instructional option(s) Instructional explanation(s)

1 12 45 37 1 0 137 28 19 0 15 72 144 1 0 31 127 23 3 13 1 5 8

3 15 141 135 16 1 246 59 86 3 37 68 63 1 2 26 178 11 97 53 6 9 4

research studies involving tool use rather than mere descriptions of tools. If the search yielded too many results (N > 300), ‘‘learning’’ was entered as an additional descriptor (marked with ‘‘*’’ in Table 1). All abstracts were read to find out whether the publications dealt with research (qualitative or quantitative) on tools in computer-based environments at any level of education. If this was the case, the publication was selected. Eventually, only 22 journal articles could be withdrawn. All 22 studies report research results on tool use itself, variables influencing tool use and/or the effect of tool use on learning. No review studies were included, as for instance the review on help seeking behavior by Aleven, Stahl, Schworm, Fischer, and Wallace (2003). Providing the initial numbers of records found, this limited number of selected articles seems very surprising. However, most studies provide a description of tools present in their learning environment studied at hand. The studies do report research results, but these are not related to the tools themselves. Other journal articles use the term ‘‘tools’’, but are actually referring to embedded support devices or to computers themselves serving as a cognitive tool. In order to describe the different articles and to compare them, a classification system of tools was looked for. Jonassen (1999) provides an elaborate categorization system for support devices. This system is part of an instructional design model for so-called ‘‘constructivist learning environments’’, i.e. learning environments that aim at fostering problem solving and conceptual development by confronting learn-

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

393

ers with ill-defined problems. This system provides a category of devices that visualize, organize, automate or supplant thinking skills. Given the comprehensiveness of this framework, it was used for classifying the different tools retrieved in the articles. It should be noted that the Jonassen-system is not the only one. For instance, a similar system is provided by Hannafin, Land, and Oliver (1999). However, their system can be completely integrated in JonassenÕs one. Despite the elaborate system, and probably also due to the origin of the system, one category was added based on the literature reviewed, namely ‘‘elaboration tools’’ (e.g., Carrier, Davidson, & Williams, 1985). It might be argued that elaboration tools are knowledge modeling tools, however, elaboration tools are more directed towards exercising rather than articulating or representing oneÕs knowledge. The following classification system is used: Information resources: provide information students can use to construct their mental models, formulate hypotheses and solve problems. These can be text documents, graphics, video or animations helping students to understand the problem. Access to the World Wide Web is an example of such a tool. Cognitive tools: help students to engage in and facilitate specific kinds of cognitive processing. These are devices to visualize, organize or supplant thinking skills (e.g., visualization tools such as concept maps or simulations). Knowledge modeling tools: help students to reflect on their learning process. These tools provide an environment in which students have to articulate what they know and what the relationships are between different concepts (e.g., semantic network). Performance support tools: facilitate the problem solving process by performing algorithmic tasks for the learners. This allows learners to focus more on higher order cognitive tasks (e.g., calculator, database shells). Information gathering tools: help students in seeking information so that they are not distracted from their primary goal of problem solving. Conversation and collaboration tools: are used in collaborative learning environment to support students in their collaborative knowledge building process (e.g., e-mail, videoconferencing). Elaboration tools: give access to reviews and additional exercises and practices related to the content of the task. For all studies, the nature of the learning task and the tool, the number of subjects, the dependent and independent variables and the results will be mentioned (see Tables 2 and 3).

2. Results This section is structured in line with the research questions. A first section relates to the variables influencing tool use, namely student characteristics, kind of tool, learning task and explicit encouragement. A second section discusses research findings with respect to learning effects of tool use.

394

Authorsa

Kind of tool(s)

Independent variable(s)

Result

Carrier et al. (1985) N = 28/6th graders concept learning

Elaboration

Ability Locus of control

Path analysis: partial b coefficient: .59/No descriptives Path analysis: partial b coefficient: .08/No descriptives

Carrier et al. (1984) N = 44/7th graders concept learning

Elaboration

Learning style (field (in)dependence)

Frequency of option selection (proportion of total)/Definitions: FI: .25; FD: .22/Expository instances: FI: 017; FD: .25/Practice instances: FI: .25; FD: .20/Feedback: FI: .34; FD: .35 ) v2 analysis: only difference for expository instances: v2 = 8.87, p < .05

Carrier et al. (1986) N = 37/6th graders Concept learning

Elaboration

Option type

Amount of times selected: Paraphrased definition: M = 62.8, SD = 37.1/Expository instances: M = 26.1, SD = 25.9/Practice instance: M = 31.5, SD = 30/Analytic Feedback: M = 34.5 SD = 25.9 ) ANOVA: F(3, 105) = 20.13, p < .05 Amount of option use: Encouragement group: M = 38.8/No encouragement group: M = 25.4 )ANOVA: F(1,35) = 4.82, p < .05

Encouragement Chapelle and Mizuno (1989) N = 13/University students concept learning

Elaboration

Prior knowledged

Amount of tool use perminute: High: M = .09/min, SD = .06 Low: M = .14/min, SD = .08 ) T test: t = 1.34; df = 11, n.s. Amount of tool use per sentence High: M = .13/sent, SD = .09 ) Low: M = .28/sent, SD = .14 ) T test: t = 2.27; df = 11, p < .05

Purpose of tool use On-going problem solving:

Extra information High = 0%, low = 7%

High = 78%, low = 48%b Advanced organizer High = 7% Low = 32% Reconfirmation High = 11%, low = 9%

Others

High = 4% Low = 4%

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

Table 2 Factors influencing tool use

Elaboration

Individual/cooperative Leanplus/Fullminus

Amount of total optional screens consulted: Individual group: M = 74%/co-operative group: M = 65% ) ANOVA: F(1,124) = 4.92, p < .05, ES = .35 Amount of total optional screens consulted: Leanplus: 56%/Fullminus: 83% ) ANOVA: F(1, 124) = 51.96, p < .05; ES = .99

Crooks et al. (1998) N = 97/ Undergraduates concept learning

Elaboration

Individual/cooperative Leanplus/Fullminus

Amount of total optional screens consulted: Individual group: M = 75.14, SD = 12.99/Cooperative group: M = 74.92, SD = 12.31 ) ANOVA: Not significant (no statistics) Amount of total optional screens consulted: Leanplus: M = 67.36, SD = 16.11/Fullminus: M = 82.70, SD = 9.19 ) ANOVA: F(1, 96) = 34.36, p < .05

Gra¨sel et al. (2001) N1 = 24/N2 = 12/ University students problem solving

Knowledge

Strategy modeling modeling Instruction

Amount of using additional information corrective: Strategy group: M = 15.88/Without strategy group: M = 9.13 ) Mann–Whitney U test: U = 8.0 p < .05 Amount of using additional information corrective: Instruction group: M = 8.17/Without instruction group: M = 4.83 ) Mann–Whitney U test: U = 8.0, p < .10c

Fischer et al. (2003) N = 11/University students problem solving

Cognitive/ information resources Elaboration

(Prior knowledge) Leanplus/Fullminus

Although prior knowledge was low; information resources hardly used (no statistics provided). Cognitive tool was used by all students (no statistics provided)

Hannafin and Sullivan (1995) N = 133/9th & 10th graders concept learning

Amount of total optional screens consulted: Leanplus: 32%/Fullminus: 78% ) ANOVA: F(1, 132) = 4.13, p < .05 Interaction effect with ability: Fullminus-Low ability: M = 76%, SD = 18.7%/Fullminus-High ability: M = 79%, SD = 20.6% Leanplus-Low ability: M = 19%, SD = 5.1%/Leanplus-High ability: M = 43%, SD = 11.4% ) ANOVA: F(1, 132) = 12.36, p < .05 (continued on next page)

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

Crooks et al. (1996) N = 125/ Undergraduates concept learning

395

396

Table 2 (continued) Kind of tool(s)

Independent variable(s)

Result

Hasselerharm and Leemkuil (1990) N = 110/ Secondary ed. st. concept learning

Cognitive

Learning style (field (in)dependence) Prior achievement

45% used advisement; no significant effect. No statistics provided No significant effect, no statistics provided

Hicken et al. (1992) N = 111/ Undergraduates concept learning

Elaboration

Leanplus/Fullminus

Amount of total optional screens consulted: Leanplus: 32% (SD = 27)/Fullminus: 80% (SD = 25) ) ANOVA: F(1, 92) = 70.80, p < .05; ES = 1.79

Lee and Lehman (1993) N = 162/ Undergraduates concept learning

Information resource

Learning style (active/neutral/ passive) Instructional cues

Selection frequency: Active: M = 0.86, SD = 0.66/Neutral: M = 0.82, SD = 0.72/Passive: M = 0.58, SD = 0.62 ) ANOVA: F(2, 161) = 2.64, n.s. Selection frequency: With cues: M = .94, SD = .58/Without cues: M = .59, SD = .73 ) ANOVA: F(1, 161) = 9.81, p < .05 Interaction effect with learning style: With cues-active: M = .82, SD = .52/With cues-neutral: M = 1.17, SD = 0.58/With cues-passive: M = 0.75, SD = .53/Without cues-active: M = 0.90, SD = 0.78/Without cues-neutral: M = 0.72, SD = 0.67/Without cues-passive: M = 0.72, SD = 0.68 ) ANOVA: F(2, 161) = 5.55, p < .05

Liu and Reed (1994) N = 63/College students concept learning

Performance support/ Information resource/ Elaboration Cognitive

Learning style (field (in)dependence)

Amount of total tool use: FD: M = 16.21/Fmixed: M = 29.28/FI: M: 24.84 ) ANOVAÕs for 5 support tools: n.s., results reported for use of index: F(2, 62) = 2.54, p = .09, no results reported for other 4 tools

Prior knowledge Reading comprehen sion skills

MANOVA: k = 0.88, p < .05: Prior knowledge enhancing effect on use performance support tools F(1, 50) = 6.12, p < .05/testing tools F(1, 50) = 3.99, p < .05/orienting tools: not significant, no F value MANOVA: No significant effects/descriptives, no statistics

Martens et al. (1997)e N = 51/University students concept learning

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

Authorsa

Knowledge modeling/ performance support

(Kind of tool)f

Knowledge modeling tools rarely used (no descriptives provided)/Performance support tool used more frequently

Pedersen and Liu (2002) N = 66/6th graders problem solving

Cognitive tool

Expert support

Number of times notebook was used: Group 1 (modeling): M = 86.5, SD = 21.4/Group 2 (didactic): M = 42.9, SD = 35.3/Group 3 (help): M = 51.4, SD = 24.2 ) F = 14.5; p < .05; ES (g2) = .32

Relan (1995) N = 109/6th graders concept learning

Elaboration

training

Total amount of review: Learner control complete LCC: Comprehensive training: M: 3.2, SD = 2.4/Partial training: M = 2.9, SD = 3.0/No training: M = 3.1, SD = 2.8/Learner control limited (LCL) Comprehensive training: M = 4.2, SD = 3.6/Partial training: M = 1.2, SD = 1.6/No training: M = 2.4, SD = 3.6 ) ANOVA: Not significant within LCC/) ANOVA: Not significant over two groups (no statistics presented)

Renkl (2002) N = 28/Student teachers problem solving

Cognitive

Prior knowledge

Cluster analysis: Four clusters: (1) above average prior knowledge, high far transfer performance, little instructional explanation use/(2) low prior knowledge, good transfer performance, above average use of extended instructional explanations/(3) average prior knowledge, average performance, little use of extensive explanations, frequent use of minimalist explanation/(4) above average prior knowledge, under average transfer performance, little use of instructional explanations

Schnackenberg and Sullivan (2000) N = 99/University juniors concept learning

Elaboration

Ability Leanplus/Fullminus

Amount of optional screens: High: M = 25.04, SD = 15.57/Low: M = 20.15, SD = 15.68 ) ANOVA: F(1, 98) = 3,71, n.s. Amount of total optional screens consulted: Leanplus: 35%/Fullminus: 68% ) ANOVA: F(1, 98) = 30.42, p < .05; ES = 1.08

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

Oliver and Hannafin (2000) N = 12/8th graders problem solving

(continued on next page)

397

398

Kind of tool(s)

Independent variable(s)

Result

Viau and Larive´e (1993) N = 70/College students concept learning

Elaboration

Prior knowledge

Amount of tool use: Regression analysis: Weak: M = 8.3, SD = 6.8, r = .43, p < .05/ Avarege: M = 11.4, SD = 9.1, r = .50, p < .05/Strong: M = 10.8, SD = 6.6, r = .30, p > .05

a

The studies are alphabetically ordered according to the first author. The results presented are taken from Table 6 (p. 39) of this article. However, the authors report different percentages in their text (p. 38), where they state that 71% of the high level students use the tools for on-going problem solving and 67% of the lower level students. c The authors indicate this to be significant. d This was an evaluation study, as such no real independent variables was specified in advance. b

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

Table 2 (continued) Authorsa

Table 3 Learning effect of tool use Kind of tools

Dependent variable

Results

Carrier and Williams (1988) N = 114/6ht graders concept learning

Elaboration

Performance

Pearson correlation of option selection and post-test: r = .28, p < .05; with delayed test: r = .20, p < .05. Low opt. sel.: Mpa = 4.7, SD = 2.8/ Md = 4.8, SD = 3.1/Medium-low opt. sel.: Mp = 5.6, SD = 2.9/ Md = 5.0, SD = 3.4/Medium-high op. sel.: Mp = 8.6; SD = 2.4/ Md = 7.0; SD = 3.7/High opt. sel.: Mp = 6.9, SD = 3.7/Md = 6.5, SD = 3.8 MANOVA on repeated measures: Interaction between treatment and quadratic level of choice: F(1, 86) = 4.28, p < .05; Interaction time by level of option selection: Quadratic trend interaction: F(1, 86) = 4.83, p < .05

Carrier et al. (1985) N = 20/6th graders concept learning

Elaboration

Performance

FisherÕs exact test: High ability high option: post-test = .04; delayed = .11/High-ability low option: post-test = .80, delayed = .80; Low ability high option: post-test = .38; dealyed = .51/Low ability low option: post-test = .11, delayed = .34

Martens et al. (1997) N = 51/University students concept learning

Cognitive

Performance

Interaction between discernability and use of toolsbDiscernability and use of processing tool: (F(1, 42) = 5.66, p < .05)/Discernability and use of testing tool: F(1, 41) = 3.6, p > .06

Morrison et al. (1992) N = 73/6th graders concept learning

Elaboration

Performance Attitude

Correlation between tool use and performance: R = .06; p > .05 Correlation between tool use and attitude: R = .05; p > .05

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

Authors

(continued on next page) 399

400

Table 3 (continued) Kind of tools

Dependent variable

Results

Oliver and Hannafin (2000) N = 12/8th graders problem solving

Knowledge monitoring

Use of higher order skills

Qualitative analysis: No effects found (little use): )no descriptives given

Renkl (2002) N = 48/student teachers problem solving

Cognitive

Performance

One-tailed t test: t(46) = 1.71; p < .05/Post-test: Mc = 42.5, SD = 21.19; ME = 53.71, SD = 23.30; ES = .50/Near transfer: Mc = 54.40, SD = 26.25; ME = 63.93, SD = 28.85; ES = .34/Far transfer: Mc = 35.0, SD = 20.22; ME = 47.32, SD = 22.85

Viau and Larive´e (1993) N = 70/College students concept learning

Elaboration Regression analysis: Frequency glossary consultation: Weak: M = 8.33, SD = 6.77; r = .43; p < .05/Average: M = 11.36, SD = 9.08; r = .50, p < .05/Strong: M = 10.75, SD = 6.63; r = .30, p < .05. Time on glossary: Weak: M = 11.46, SD = 11.29; r = .29, p < .05/Average: M = 16.26, SD = 16.07; r = .39, p < .05/ Strong: M = 12.11, SD = 7.62; r = .54, p < .05

Performance

Multiple regression analysis: 21.6% of variance explained by frequency of glossary consultation/Significant contribution of time (r = .32) and frequency (r = .44) of glossary consultation. No significant contribution for time (r = .08) or frequency (r = .12) of navigation map consultation

a b

Mp the mean score on post-test; Md the mean score on delayed test. These groups consisted also of students receiving a printed version of a textbook.

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

Authors

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

401

2.1. Variables influencing tool use 2.1.1. What student characteristics influence tool use? Research on adjunct aids in textbooks suggests that different student characteristics influence the use of tools (Elen, 1995). In 10 of the retrieved studies ability, prior knowledge, motivation, reading comprehension skills, locus of control and learning style were studied. These 10 studies show that some student characteristics have been considered as influencing variables for tool use. However these studies appear to be inconclusive. The effect of ability seems not stable. Two of the studies (Carrier et al., 1985; Chapelle & Mizuno, 1989) found high ability students to profit more from control over tool use than low ability students. 3 High ability students used the tools more frequently than low ability students. Moreover, Chapelle and Mizuno (1989) showed that high ability use tools differently than low ability students. High ability students use tools as problem solving aids, while low ability students use these tools as advance organizers. In contrast to Chapelle and Mizuno (1989), who indicated that the effect of ability is only related to one specific tool (consultation of facts-tool) and not to the consultation of a grammar-tool or dictionary, Schnackenberg and Sullivan (2000) could not replicate the influence of ability on tool use. Prior knowledge also is a non-stable factor. Martens, Valcke, and Portier (1997) observed a positive effect of prior knowledge on tool use in a computer-based textbook, with more prior knowledge resulting in more tool use. Viau and Larive´e (1993) however, report a curvilinear relation. In their study, average students used the available tool more often than both weak and strong students. At the same time, Renkl (2002) found low prior knowledge students to demand more frequently a tool providing instructional explanations than students with high prior knowledge. Again, these results may be related to the nature of the tool. Martens et al. (1997) and Renkl (2002) used cognitive tools, whereas Viau and Larive´e (1993) used an elaboration tool. Martens et al. (1997) also studied the effect of motivation and reading comprehension skills on tool use. No significant effects were found for these two characteristics. Effects were neither revealed for locus of control. Carrier, Davidson, Williams, and Kalweit (1986), showed no different tool use behavior between students who perceive personal success or failure as a result of their own action (internal locus of control), and students who ascribe success or failure to external factors (external locus of control). Learning style finally, has two meanings in the selected articles. One study denotes activity level as learning style and differentiates between active, neutral and passive learners (Lee & Lehman, 1993). No effects were found. All other studies measure learning style as field (in)dependence (Carrier, Davidson, Higson, & Williams, 1984; Hasselerharm & Leemkuil, 1990; Liu & Reed, 1994). These studies are not

3

The correlations and differences reported are significant on a. 05-level.

402

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

conclusive: some find no effect at all (Hasselerharm & Leemkuil, 1990); in others field independent learners tend to more frequently use an index tool than field dependent learners, while the mixed group more frequently used a note-taking tool (Liu & Reed, 1994). And, Carrier et al. (1984) report field independent learners to more frequently use an elaboration tool than field dependent learners. Clearly, with respect to the influence of student characteristics on tool use the presented studies do not lead to a firm conclusion. The number of studies is limited and only a restricted number of student characteristics has been investigated so far. However, the results do suggest that the nature of the tool might be important given the interaction effects and the difference in use for different kind of tools. This aspect has attracted specific research attention as will be presented in the next section.

2.1.2. Does the nature of the tools influence tool use? Nine studies report on the influence of the nature of the tool on tool use. In the previous part, three studies already pertain to this issue (Carrier et al., 1986; Chapelle & Mizuno, 1989; Schnackenberg & Sullivan, 2000). Chapelle and Mizuno (1989) revealed that the use of the glossary had a positive effect but the use of a navigation map had no effect. Similarly, Carrier et al. (1986) showed that paraphrase tools were more frequently used than elaboration tools. Schnackenberg and Sullivan (2000) found that a tool to bypass instruction was used less often than a tool to request additional information. The study of Chapelle and Mizuno (1989) confirms that the nature of the tool matters: a glossary was used, a navigation map was not used. Oliver and Hannafin (2000) made a similar conclusion based on a study in which they provided students with different kinds of tools: performance support-, information gathering-, cognitive- and knowledge monitoring tools. Students almost exclusively used the performance support and information gathering tools. Fischer, Troendle, and Mandl (2003) likewise found the use of tools to be related to the kind of tool. In their study, the cognitive tool (a visualization tool) was used, whereas information resources were only seldom used. A large group of studies in this group studies differences between fullminus and leanplus conditions (Crooks, Klein, Jones, & Dwyer, 1996; Crooks, Klein, & Savenye, 1998; Hannafin & Sullivan, 1995; Hicken, Sullivan, & Klein, 1992; Schnackenberg & Sullivan, 2000). Such studies compare students who have access to a tool that allows them to bypass instruction (fullminus) or students who have access to a tool that gives them more instruction (leanplus). The additional instruction consists of reviews, summaries and practice items (elaboration tools). In all these studies, the fullminus group views significantly more instruction than the leanplus group. Fullminus groups only seldom use the tool to bypass instruction and leanplus groups seldom request additional instruction. Crooks et al. (1998) attribute the difference between the leanplus and fullminus group to one elaboration tool, namely the consultation of practice items. No differences were found for any of the other elaboration tools. Carrier et al. (1986) also found a difference in use between different tools in the leanplus groups: paraphrased

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

403

definition, expository instances, practice instances and analytic feedback. Paraphrased definitions were used more often than expository instances. Similarly, practice items were more frequently used than expository instances. Hannafin and Sullivan (1995) did not report specific differences between the effects of various elaboration tools, but they did find an interaction effect between ability and the program version. In the leanplus version, but not in the fullminus version, high ability students selected more (43%) options than low ability students did (19%).

2.1.3. Do learning task and working method influence tool use? None of the studies directly addresses the issue of learner tasks. In order to answer this question, the tasks used in the different studies were looked at and compared to the results with respect to tool use. In 16 studies, subjects had to learn specific concepts. Learning results were measured by a knowledge post-test. For instance, in the studies of Carrier et al. (1984, 1985, 1986) subjects are confronted with a computerbased lesson about four propaganda-techniques used in advertisement. After the lesson, subjects were tested through means of a classification test. Only four studies deal with problem solving tasks (Gra¨sel, Fischer, & Mandl, 2001; Fischer et al., 2003; Oliver & Hannafin, 2000; Pedersen & Liu, 2002). In both groups of studies, results indicate that students tend to use some tools more than others (see previous section) and hence, not all tools are used. It could be expected that in the problem-solving studies, subjects would need more tools, given the more open character of the learning environment. However, the reviewed studies do not confirm this expectation. These studies do not allow to draw firm conclusions of task influence on tool use. Crooks et al. (1996, 1998) addressed the issue of working method. They investigated the influence of individual versus co-operative work on the use of tools. The 1996-study revealed individuals to more frequently use optional elements than cooperative groups. In the 1998-study, however, no differences were found between the two working methods.

2.1.4. Does explicit encouragement of tool use affect tool use? Advice while students are working with an application has a significant positive effect (Carrier et al., 1986; Lee & Lehman, 1993). Students, who receive instructional cues or encouragement to use certain options, use the available tools more compared to students who do not receive these cues or encouragement. However, Lee and Lehman (1993) point to an interaction effect. A positive effect of encouragement seemed to apply only for regularly active learners, not for active or passive learners. Regularly active learners with instructional cues selected more information than learners with the same learning style without instructional cues. Gra¨sel et al. (2001) showed that students who received strategy training made more adequate use of additional information (a glossary), a diagnostic help tool and a database than students who did not receive strategy training. Additionally,

404

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

students in the strategy-training group adapted their problem solving process on the basis of the additional information requested. To complicate matters, Relan (1995) found that trained students used more frequently the elaboration tool before practice than during practice, in contrast to the group who did not receive any training. It can be questioned whether these mixed results are due to the difference in task or the nature of the tools. Finally, Pedersen and Liu (2002) studied the use of a notebook in a problembased learning environment. All students received additional support by an expert on video. In the condition where the expert models his reasoning process by introducing and applying strategies and actually using the tools, the use of the notebook was highest as was the number of relevant notes in that notebook. The other two groups had an expert only providing information on how the tool functions, but not using the tools. In addition, one group received suggestions for specific strategies. These two groups did not differ from one another. Table 2 summarizes the results with respect to variables affecting tool use.

2.2. Learning effects of tool use Whereas in the previous section variables influencing tool use were discussed, this section addresses the effect of tool use on learning. Six out of 21 studies on tool use report learning effects. Three of these studies deal with the effect of elaboration tools on learning (Carrier et al., 1985; Carrier & Williams, 1988; Morrison, Ross, & Baldwin, 1992; Viau & Larive´e, 1993); two other pertain to cognitive tools (Martens et al., 1997; Renkl, 2002) and, one to the influence of knowledge monitoring tools (Oliver & Hannafin, 2000). These studies provide mixed results on the effect of tool use on performance. Viau and Larive´e (1993) showed that the use of an elaboration tool (glossary) explains 21.6% of the variance in performance results. They did not find this effect for the use of a navigation tool, which suggests that an elaboration tool may have more influence on performance than a processing tool. Carrier and Williams (1988) indicated a moderate effect of tool use on performance. Moreover, this effect was mediated by ability. High ability students benefit more when using the tools than low ability student. However, they did have some statistical problems since only few lower ability students actually used many tools. Morrison et al. (1992) investigated the difference between a learner control and a program control group. The learner control group performed worse than the program control group. This group did not often use the elaboration tool present. This lead the researchers to calculate the correlation between tool use and post-test score for the learner control group. They could not find any significant correlation. By using a higher order skill test, Oliver and Hannafin (2000) were unable to reveal an effect of the use of performance tools on higher order learning. It has to be noted that this is the only study where no knowledge test was used. Martens et al. (1997) compared two groups, one in which students had access to cognitive tools (discernability group) and one group where these tools were totally

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

405

integrated in the environment after students activated them (non-discernability group). They report an interaction effect between the groups and the studentsÕ use of the support devices. Students who seldom used the elaboration and cognitive tools scored higher on a post-test than students who frequently used these tools in the discernability group. The opposite was found for the non-discernability group. Renkl (2002) compared a group who did not get access to instructional explanations with a group that did get access to these explanations when clicking on a button. He found that the learners in the experimental group were significantly more successful in the post-test than the participants in the control group. A further analysis showed this effect to be due to the scores on the far transfer test rather than on the near transfer test. In this study, four clusters of users could be identified. A first cluster are students with high prior knowledge and a high gain on the transfer test although they rarely relied on the instructional explanations. A second cluster of students had low prior knowledge, but had an overall good transfer performance and they made often use of the extensive instructional explanations. A third cluster are students with average prior knowledge and an average score on the transfer test. They used the minimalistic explanations very frequently, but not the extensive ones. In a last cluster those students could be classified with also high prior knowledge, but only average transfer performance. They rarely sued the instructional explanations, although the results on the post-test shows that they might have benefited from this use. These results do not give a clear picture of the learning effects of tool use, but they do give some indication that positive effects of tools cannot be taken for granted. This seems to be related to the tools themselves, the way students use them, and specific student characteristics. Of course, in order to find an effect of tool use on learning, students have to use the tools (adequately). Clearly, this is one of the methodological problems of studying the effect of tool use. For instance, Oliver and Hannafin (2000) report that while cognitive tools and knowledge monitoring tools were provided, students only seldom used them. Moreover, if students used these tools, they did not use them as intended. For example, knowledge monitoring tools were developed and integrated to promote higher order thinking processes, to organize information or to justify ideas. However, students used these tools to make lists of web pages. This could also explain the lack of positive effects of tool use on higher order reasoning. Table 3 summarizes an overview of the different studies reporting on learning effects of tool use.

3. Discussion Up to now the use of tools has attracted only minimal research attention. The search revealed 17 studies addressing factors influencing tool use, and five studies dealing with learning effects of tool use.

406

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

Although the analysis does not reveal a clear and convincing picture, some preliminary conclusions can be drawn: tool use seems to be influenced by (1) student characteristics, (2) the kind of tool, and (3) additional advice. Looking at the learning effect of tool use it is striking that only a small number of studies deal with the influence of tool use on performance. Oliver and Hannafin (2000) aimed at studying this effect, but could not draw any conclusion since students hardly used the tools. If the effects of tool use on performance are studied, only learning results are looked at, the learning process itself and the effects of using these tools on the learning process are addressed in one study only. This might be related to the kind of task students are confronted with in these studies, namely a concept acquisition task (see Tables 2 and 3). After working in this environment a knowledge test is administered to measure whether students know these concepts. These findings also raise the question whether there was a need for students to make use of the different tools in order to complete the task. Moreover, the reported studies often lack a theoretical basis on which the inclusion of tools in the learning environment and the measurement of particular (influencing) variables can be justified. Performing a thorough task analysis before studying tool use might solve this problem. Providing that open learning environments foster the acquisition of complex problem solving or higher order reasoning skills (Jacobson & Spiro, 1995; Jonassen, 1997), the research analysis, therefore, calls for a systematic research program; as insight in tool use becomes of particular interest with the use of open learning environments. In open learning environments, the learners construct their own knowledge in interaction with the environment. In other words, there is a high level of learner control and learners have to regulate their own learning (Hannafin, 1995). This also includes making adequate choices towards tools to be used. In designing such environments, it is important to understand the process of tool use. This also means that not only the amount of tool use should be considered, but also what students actually do when using tools (e.g. Gra¨sel et al., 2001). However, in order to do so, some requirements have to be met: (1) Students should first use tools before the adequacy of the actual use can be studied. Studies indicate that this is not always the case. For instance, some of the studies in which fullminus and leanplus groups are studied found either no effect between these groups on performance (e.g. Hannafin & Sullivan, 1995; Crooks et al., 1998) or they found the fullminus group to outperform the leanplus group (e.g. Crooks et al., 1996; Schnackenberg & Sullivan, 2000). These findings are explained by referring to the number of tools used. The leanplus group almost did not use the available elaboration tool. In a similar effort, Morrison et al. (1992) revealed that the learner control group had reviewed only a limited number of items. From the possible total of 12 review items, 42.3% reviewed three items, 40% four or five items and only one person reviewed them all (N = 73). There seems to be evidence that merely providing tools to students does not result in the use of these tools (e.g. Oliver & Hannafin, 2000; Fischer et al., 2003). However, studying the adequacy of tool use also implies some requirements of the methodology used. While most studies use log files to track tool use, only the number of

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

407

use or kind of tool is looked at. The analysis of the log files on the level of when these tools are used was not found. In three studies some other valuable methods were used that allow gaining insight in the adequacy of tool used. Renkl (2002) used thinking aloud procedures that allow to link tool use with participantsÕ cognitive processes. Similarly, Oliver and Hannafin (2000), and Fischer et al. (2003) made use of observations to gain insight in tool use. (2) A second requirement to study the adequacy of tool use relates to the relation between the tool and the task. The studies investigate tool use, but the adequacy of tool use for learning or the need for students to consult these tools when solving the task is never questioned. An evaluation of the tools themselves, in view of the task is required to establish the need for using these tools. Apparently, this cannot be taken for granted. Morrison et al. (1992), for instance, did not find any correlation between tool use and scores on a post-test. Similar results were found by Martens et al. (1997). However, with respect to the latter study it should be noted that tool use was measured through a questionnaire that was administered after students worked with the environment. It might be that students did not adequately report on their actual tool use.

4. Conclusion This contribution has reported on different studies addressing tool use. Some indication is provided of variables influencing learning and of the effect of tool use on learning. Although only a limited number of studies were retrieved, a resemblance with factors influencing the use of adjunct aids in texts was found. When advocating the use of open learning environments, the issue of tool use becomes more apparent. Learners are in control and decide autonomously on the use of these tools. Reviews on learner control and some of the studies reviewed, reveal that students hardly apply monitoring and regulation skills for their learning process or for making adequate choices with respect to their learning. The limited number of studies addressing tool use in computer-based learning environments indicates that more research is needed to identify the different learner, tool and task characteristics that affect tool use. For instance, a student characteristic that was not considered, but that might be of relevance to tool use are studentsÕ conceptions about these tools. Winne (1985) already indicated that the functionality students ascribe to certain elements in a learning environment influence whether and how they will use these elements. This was already illustrated in the introduction. Similar, interviews by Brush and Saye (2001) clearly show that students do not use support devices as intended because students were unsure how to use them. This also illustrates that the problem not only relates to tools but to embedded support devices as well. Greene and Land (2000) reveal that even when support devices are integrated in the environment, students tend to use them inadequately. They provided questions to students to encourage deep-level processing. Instead of using these questions as a tool to aid cognition, students responded with a superficial activity rather than with underlying cognitive

408

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

processes. This may explain why the provision of additional advice results in positive effects. Further research might identify how the optimal use of tools can be fostered. Elaborating on the issue of providing additional advice or training can do this. But also task, tool and student characteristics should be studied, such as studentsÕ conceptions or metacognitive skills. Different authors acknowledge the importance of metacognitive skills for making adequate decisions (de Jong & van Joolingen, 1998; Hill & Hannafin, 2001; Kinzie & Berdel, 1990; Land, 2000; Lee & Lee, 1991). Hence, it can be expected that metacognitive skills are related to the sensible use of tools. Moreover, most studies in this review deal with elaboration tools, and only a limited number deal with other kind of tools, such as cognitive or knowledge modeling tools. In order to be able to draw conclusions for the design of open learning environments, more research is needed in which students are confronted with problem solving tasks, rather than concept acquisition tasks. In fact, the same suggestions can be made for studying tool use, as Mayer (1979) made for the study of advance organizers: ‘‘Future theories should attempt to specify exactly what are the ‘‘subsuming concepts’’ in the advance organizer, how they are related to the instructional information, and how the learning outcome of an advance organizer subject differs from the cognitive structure acquired by someone who learns without an advance organizer’’ (Mayer, 1979, p. 163). This citation points also to the necessity of theories underlying the implementation of tools in a learning environment. In Fig. 1 an attempt is made to give an outline for further research with respect to the use of tools. In this figure different aspects that can be involved in further research are presented.

Task characteristics Learning process Tool characteristics

Student characteristics

quantity of tool use

quality of tool use

Additional cues

Fig. 1. Research outline.

learning results

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

409

References Aleven, V., Stahl, E., Schworm, S., Fischer, F., & Wallace, R. (2003). Help seeking and help design in interactive learning environments. Review of Educational Research, 73, 277–320. Brush, T., & Saye, J. (2001). The use of embedded scaffolds with hypermedia-supported student-centered learning. Journal of Educational Multimedia and Hypermedia, 10, 333–356. Carrier, C., Davidson, G., Higson, V., & Williams, M. (1984). Selection of options by field independent and dependent children in a computer-based concept lesson. Journal of computer-based instruction, 11, 49–54. Carrier, C., Davidson, G., & Williams, M. (1985). The selection of instructional options in a computerbased co-ordinate concept lesson. Educational Communication and Technology Journal, 33, 199–212. Carrier, C., Davidson, G., Williams, M., & Kalweit (1986). Instructional options and encouragement effects in a micro-computer concept lesson. Journal of Educational Research, 79, 222–229. Carrier, C., & Williams, M. (1988). A test of one-learner control strategy with students of differing levels of task persistence. American Educational Research Journal, 25, 286–306. Chapelle, C., & Mizuno, S. (1989). StudentsÕ strategies with learner-controlled CALL. Calico Journal, 7(2), 25–47. Clarebout, G., & Elen, J. (2002a). September. The use of tools in learning environments: a literature review. Paper presented at the New educational benefits of ICT in Higher Education-conference, Rotterdam, the Netherlands. Clarebout, G., & Elen, J. (2002b). November. The use of tools in computer-based learning environments: a literature review. Paper presented at the AECT conference, Dallas, USA. Clarebout, G., Elen, J., Lowyck, J., Van den Ende, J., & Van den Enden, E. (2004). Evaluation of an open learning environment: The contribution of evaluation to the ADDIE process. In A. Armstrong (Ed.), Structional design in the real world (pp. 119–135). Hershey, PA: Idea Group Publishing. Clark, R. E. (1991). When teaching kills learning: Research on mathematantics. In H. Mandl, E. De Corte, N. Bennett, & H. F. Friedrich (Eds.), European research in an international context. Learning and instruction (Vol. 2, pp. 1–22). Oxford: Pergamon Press. Crooks, S. M., Klein, J. D., Jones, E. E., & Dwyer, H. (1996). Effects of cooperative learning and learnercontrol modes in computer-based instruction. Journal of research on computing in education, 29, 109–123. Crooks, S. M., Klein, J. D., & Savenye, W. C. (1998). Effects of cooperative and individual learning during learner-controlled computer-based instruction. Journal of Experimental Education, 66, 223–244. de Jong, T., & van Joolingen, W. (1998). Scientific discovery learning with computer simulations of conceptual domains. Review of Educational Research, 68, 179–201. Elen, J. (1995). Blocks on the road to instructional design prescriptions. Leuven: University Press. Fischer, F., Troendle, P., & Mandl, H. (2003). Using the internet to improve university education: problem-oriented web-based learning with MUNICS. Interactive Learning Environments, 11(3), 193–214. Friend, L. C., & Cole, C. L. (1990). Learner control in computer-based instruction: a current literature review. Educational Technology, 30(11), 47–49. Goforth, D. (1994). Learner control = decision making + information: a model and meta-analysis. Journal of Educational Computing Research, 11, 1–26. Gra¨sel, C., Fischer, F., & Mandl, H. (2001). The use of additional information in problem-oriented learning environments. Learning Environment Research, 3, 287–325. Greene, B. A., & Land, S. M. (2000). A qualitative analysis of scaffolding use in a resource-based learning environment involving the world wide web. Journal of Educational Computing Research, 23, 151–179. Hannafin, M. (1995). Open-ended learning environments: Foundations, assumptions, and implications for automated design. In R. D. Tennyson & A. E. Barron (Eds.), Automating instructional design: Computer-based development and delivery tools (pp. 101–129). Berlin: Springer. Hannafin, M., Land, S., & Oliver, K. (1999). Open learning environments: Foundation, methods and models. In Instructional design theories and models. A new paradigm of instructional theory (Vol. II, pp. 115–140). Mahwah, NJ: Lawrence Erlbaum.

410

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

Hannafin, R. D., & Sullivan, H. J. (1995). Learner control in full and lean CAI-programs. Educational Technology, Research and Development, 43, 19–30. Hasselerharm, E., & Leemkuil, H. (1990). The relation between instructional control strategies and performance and attitudes in computer-based instruction. In J. M. Pieters, P. R. Simons, & L. De Leeuw (Eds.), Research on computer-based instruction (pp. 67–80). Amsterdam: Swets & Zeitlinger. Hicken, S., Sullivan, H., & Klein, J. (1992). Learner control modes and incentive variations in computerdelivered instruction. Educational Technology, Research and Development, 40, 15–26. Hill, J. R., & Hannafin, M. J. (2001). Teaching and learning in digital environments: the resurgence of resource-based learning. Educational Technology, Research and Development, 49, 37–52. Jacobson, M. J., & Spiro, R. J. (1995). Hypertext learning environments, cognitive flexibility and the transfer of complex knowledge. Journal of Educational Computing Research, 12, 301–333. Jonassen, D. H. (1997). Instructional design models for well-structured and ill-structured problem-solving learning outcomes. Educational Technology, Research and Development, 45, 65–91. Jonassen, D. H. (1999). Designing constructivist learning environments. In C. M. Reigeluth (Ed.), Instructional-design theories and models. A new paradigm of Instructional Theory (Vol. II, pp. 215–239). Mahwah, NJ: Lawrence Erlbaum. Kinzie, M. B., & Berdel, R. L. (1990). Design and use of hypermedia systems. Educational Technology, Research and Development, 38(3), 61–68. Lajoie, S. P. (Ed.). (2000). Computers as cognitive tools. No more walls (Vol. II). Mahwah, NJ: Lawrence Erlbaum. Lajoie, S. P., & Derry, S. J. (Eds.). (1993). Computers as cognitive tools. Hillsdale, NJ: Lawrence Erlbaum. Land, S. M. (2000). Cognitive requirements for learning with open-learning environments. Educational Technology, Research and Development, 48(3), 61–78. Large, A. (1996). Hypertext instructional programs and learner control: a research review. Education for Information, 14, 95–108. Lee, S., & Lee, Y. H. (1991). Effects of learner-control versus program-control strategies on computeraided learning of chemistry problems: for acquisition or review?. Journal of Educational Psychology, 83, 491–498. Lee, Y. B., & Lehman, J. D. (1993). Instructional cueing in hypermedia: a study with active and passive learners. Journal of Educational Multimedia and Hypermedia, 2, 25–37. Liu, M., & Reed, W. M. (1994). The relationship between the learning strategies and learning styles in a hypermedia environment. Computers in Human Behavior, 10, 419–434. Marek, P., Griggs, R. A., & Christopher, A. N. (1999). Pedagogical aids in textbooks: Do college studentsÕ perceptions justify their prevalence?. Teaching of Psychology, 26(1), 11–19. Martens, R. L., Valcke, M. M., & Portier, S. J. (1997). Interactive learning environments to support independent learning: the impact of discernability of embedded support devices. Computers in Education, 28, 185–197. Milheim, W. D., & Martin, B. L. (1991). Theoretical bases for the use of learner control: three different perspectives. Journal of computer-based instruction, 18, 99–105. Morrison, G. R., Ross, S. M., & Baldwin, W. (1992). Learner control of context and instructional support in learning elementary school mathematics. Educational Technology, Research and Development, 40(1), 5–13. Oliver, K. M., & Hannafin, M. J. (2000). Student management of web-based hypermedia resources during open-ended problem solving. The Journal of Educational Research, 94, 75–92. Pedersen, S., & Liu, M. (2002). The effects of modeling expert cognitive strategies during problem-based learning. Journal of Educational computing research, 26, 353–380. Relan, A. (1995). Promoting better choices: effects of strategy training on achievement and choice behavior in learning controlled computer-based instruction. Journal of Educational Computing Research, 13, 129–149. Renkl, A. (2002). Worked-out examples: instructional explanations support learning by self-explanations. Learning and Instruction, 12, 529–556. Salomon, G. (1988). AI in reverse: computer tools that turn cognitive. Journal of Educational Computing Research, 4, 123–139.

G. Clarebout, J. Elen / Computers in Human Behavior 22 (2006) 389–411

411

Schnackenberg, H. L., & Sullivan, H. J. (2000). Learner control over full and lean computer-based instruction under differing ability levels. Educational Technology, Research and Development, 48(2), 19–35. Spiro, R. J., Feltovich, P. J., Jacobson, M. J., & Coulson, R. L. (1991). Knowledge representation, content specification and the development of skill in situation-specific knowledge assembly: some constructivist issues as they relate to cognitive flexibility. Educational Technology, 31(9), 22–25. Viau, R., & Larive´e, J. (1993). Learning tools with hypertext: an experiment. Computers & Education, 20, 11–16. Williams, M. D. (1996). Learner-control and instructional technology. In D. H. Jonassen (Ed.), Handbook of research for educational communications and technology (pp. 957–983). New York: Macmillan Library. Winne, P. H. (1985). Steps towards cognitive achievements. Journal of Elementary School Journal, 85, 673–693. Geraldine Clarebout is research assistant at the Center for Instructional Psychology and Technology at the University of Leuven. Jan Elen is professor at the Center for Instructional Psychology and Technology at the University of Leuven.