Impact of Guided Exploration and Enactive Exploration on Self

Journal of Applied Psychology2001, Vol. 86. No. 6. 1129-1141

Copy]'right 2001 by the American Psychological Association, Inc.0021-9010/01/$5.00 DOI: 10.1037//0021-9010.86.6.1129

Impact of Guided Exploration and Enactive Exploration on Self-RegulatoryMechanisms and Information Acquisition Through Electronic Search

Shelda DebowskiMurdoch University

Robert E. WoodUniversity of New South Wales

Albert BanduraStanford University

Following instruction in basic skills for electronic search, participants who practiced in a guidedexploration mode developed stronger self-efficacy and greater satisfaction than those who practiced in aself-guided exploratory mode. Intrinsic motivation was not affected by exploration mode. On 2 post-training tasks, guided exploration participants produced more effective search strategies, expended lesseffort, made fewer errors, rejected fewer lines of search, and achieved higher performance. Relative lackof support for self-regulatory factors as mediators of exploration mode impacts was attributed to theuninformative feedback from electronic search, which causes most people to remain at a novice level andto require external guidance for development of self-efficacy and skills. Self-guided learning will be moreeffective on structured tasks with more informative feedback and for individuals with greater expertiseon dynamic tasks.

With the accelerated growth of computerized informationsources, electronic search is rapidly becoming critical in the ac-quisition of information for decision making. Increasingly, infor-mation is being stored and presented in electronic rather than inprint form. Some experts estimate that more than half of theworld's information is available only by electronic access (e.g.,Cook, 1995), thus underscoring the need to master electronicinformation technologies to ensure maximum access to neededinformation.

The capability to explore electronic information sources may beincreasingly important in many domains of human functioning, butnovices typically invest as little as 30 min on learning the basiccompetencies of electronic search (Borgman, 1989). This is par-ticularly limiting because task-generated feedback from electronicsearch is not highly informative regarding the effectiveness ofstrategies. For the nonexpert, there is often no way of knowinghow many relevant records there are on the database beingsearched or if the records retrieved are the most relevant. Unless aperson is motivated to continually develop and test alternativestrategies to see how they influence the records retrieved, the basicelectronic search competencies that they develop during theirinitial exposure will not improve with experience. One aim in our

Shelda Debowski, Department of Commerce, Murdoch University,Murdoch, Western Australia, Australia; Robert E. Wood, Australian Grad-uate School of Management, University of New South Wales, Sydney,New South Wales, Australia; Albert Bandura, Department of Psychology,Stanford University.

Correspondence concerning this article should be addressed to Robert E.Wood, Australian Graduate School of Management, University of NewSouth Wales, Sydney, New South Wales 2052, Australia. Electronic mailmay be sent to [email protected].

study was to test the relative effects of introducing guided versusself-directed (enactive) modes of exploration immediately follow-ing the development of basic competencies for electronic inquiry.

The exploration and testing of strategies needed for the devel-opment of search competencies through experience is often under-mined by the fact that many people feel overwhelmed and distrusttheir ability to conduct electronic search activities (Kuhlthau,1991). Our second aim, therefore, was to test the impacts of thedifferent modes of exploration on the self-regulatory processes ofparticipants, and whether the emergent differences in the self-regulatory mechanisms produced subsequent differences in strat-egies and performance of electronic inquiries.

Electronic Search

Although electronic search tasks can vary in the degree ofcomplexity and how quickly they converge to the desired infor-mation set, the structuring of interrogations and search strategies isvery similar across the whole domain. Search strategies are builtby using keywords and other terms and Boolean connectors (and,or, etc.), which determine the pathways that get explored and theinformation retrieved on each pathway. This language has re-mained fairly stable for the past 2 decades and is used in a widerange of electronic search tasks (library databases, Web pages,Internet searches, etc.). Except for minor variations in the proto-cols attached to different commercial products, the basic processesof conducting electronic searches, described in more detail below,generalize across all types of electronic inquiry tasks and differentsearch engines.

The search for information is often a cyclical, exploratory pro-cess. People make inquiries, receive feedback, evaluate the out-comes of their inquiries, and alter their search activities accord-

1129

1130 DEBOWSKI, WOOD, AND BANDURA

ingly. On simple retrieval tasks, search sequences may compriseonly one or two steps. On more complex tasks, the process be-comes more convoluted and demanding, with each inquiry provid-ing important input for constructing further inquiries. Success inobtaining the needed information will be strongly influenced bythe exploration process, including the scope of search activities,the quality of the sources used, and the sequencing of search(Wood, George-Falvy, & Debowski, in press). Effort can also bewasted in redundant, nonproductive inquiries. For many problems,an effective sequence is to search broadly to identify a range ofoptions in the initial steps of the process and then to search theoptions in greater depth (Payne, Bettman, & Johnson, 1993).

Modes of Exploration

Instruction in the use of the technology, the basics of the searchlanguage, and the development of strategies for electronic searchcan be covered in 30 min, but practice in the construction of searchstrategies is needed to consolidate that knowledge. When left totheir own devices, novices may follow many different modes ofexploration during their practice sessions. However, we were in-terested in the potential impacts of interventions that shaped theexploration mode adopted during practice, and with this aim inmind, we focused on two modes of exploration that have beendescribed and applied in different domains of functioning. One isenactive exploration, based on process models of human explora-tion (Greif & Keller, 1990; Condry, 1977) and error management(Frese & Altmann, 1989), which have been shown to influenceintrinsic motivation, learning, and performance. The second isguided exploration, based on the guided mastery model of trainingoutlined in social cognitive theory, which has been shown to behighly effective in transmitting knowledge, competencies, andself-beliefs conducive to cognitive, motivational, and affectiveself-regulation (Bandura, 1986, 1997).

In enactive exploration, individuals are provided with unstruc-tured opportunities to explore and discover effective strategieseither from the outcomes of their trial and error actions or fromexternal sources of advice, such as on-line help functions forelectronic inquiry (Frese & Altmann, 1989; Greif & Keller, 1990;Wood, Kakebeeke, Debowski, & Frese, 2000). Enactive explora-tion is guided by the emerging problems and interests of theindividual and is not predetermined or imposed by a trainer ortraining program. Decisions about what to explore and when toseek guidance are made by individuals, giving them total controland responsibility for their own learning. Self-guided explorationis supplemented with the positive framing of the errors that aretypically encountered in practice as opportunities for learning(Wood, Kakebeeke, et al., 2000). People are encouraged to expectmistakes and to learn from them rather than to fear them or viewthem as evidence of a lack of capability. This reduces the likeli-hood of novices attributing difficulties in practice to a lack of skilland undermining their perceived efficacy.

The self-guided nature of enactive exploration has been identi-fied as producing greater intrinsic motivation in learning endeav-ors than approaches that provide high levels of external guidanceand stress the avoidance of errors (Deci & Ryan, 1980; Dormann& Frese, 1994), but exploratory modes of learning have severalpotential limitations. Without guidance, the mistakes and errorsthat often accompany self-guided exploration on complex tasks

may undermine the achievements needed to develop a robustself-efficacy for the task and create feelings of dissatisfaction withthe lack of progress. Learners may also develop metaphors andanalogies to describe their understanding of electronic inquiryprocesses, which limit the development of more effective strate-gies (Allwood & Kalen, 1993; Gay & Mazur, 1989). An exampleof a commonly used metaphor that is associated with superficialbut wide-ranging electronic search is "surfing the net" (Wood,Atkins, & Tabernero, 2000).

For the nonexpert, the potential benefits of enactive explorationmay be further limited by the low fidelity of feedback fromelectronic search tasks. Enactive exploration modes of learninghave been shown to be effective for tasks that provide clearfeedback on the adequacy of responses (Frese et al., 1991; Robert,1987). On electronic search tasks, novice searchers are oftenunable to accurately judge their progress or the adequacy of theirresponses, and they will continue with suboptimal strategies ormake strategic shifts for reasons that are unrelated to the adequacyof existing strategies.

Self-guided exploration without any encouragement to frameerrors positively typifies the natural learning processes of mostnovice electronic searchers (Borgman, 1989) and may partiallyexplain the poor electronic search strategies of nonexperts identi-fied in other descriptions of the task (Brown, 1995; Collantes,1995; Harter & Cheng, 1996; Wildemuth, de Bliek, Friedman, &File, 1995). The manipulation of enactive exploration, describedlater, was intended to be a stronger form of self-guided explorationthan that which is typically used by novices, who may be moreprone to dissatisfaction and to drops in efficacy as a result ofproblems encountered in practice.

In guided exploration, the sequencing of search activities andresponses to problematic situations are predetermined according toa set of principles that have been found to strengthen the self-efficacy beliefs of trainees (Bandura, 1997) and are not left to thepreferences and ad hoc responses of the novice. The principles areprogressive achievement and cognitive modeling, which includesdemonstrations of responses to problems encountered in practiceand the verbalization of the thought processes that guide the choiceof responses (Gist, 1989; Meichenbaum, 1977; Schunk, 1989). Inguided exploration, practice sessions are used to explore the task ina systematic and preplanned manner. Tasks are completed in anassigned order that provides practice on tasks of increasing diffi-culty. The guided enactments of practice tasks ensure that thenovice has a progressive sense of achievement, which strengthensperceived efficacy for the more demanding forms of the task andsatisfaction with progress (Bandura, 1997). Guidance in the se-quencing of practice tasks is supplemented with guidance in thechoice of responses to problems, through cognitive modeling ofresponses from among those taught in the introductory trainingsession. This further strengthens perceived self-efficacy for usingand refining strategies (Latham & Saari, 1979).

Modeling of responses to problems also augments the lessinformative task feedback from electronic search by providingtrainees with a signal regarding the adequacy of their response plusan immediate description of a standard with which they cancompare their own action. The combination of an implicitfeedback-standard comparison and demonstration of a responsefor removing the identified gap facilitates a motivated task focus,a condition under which feedback is most likely to enhance learn-

MODES AND MECHANISMS OF ELECTRONIC SEARCH 1131

ing and performance (Kluger & DeNisi, 1996). The augmentedfeedback received under guided exploration can be contrasted withthat received from enactive exploration, in which the only feed-back is the list of records retrieved by the search statement, whichthe searcher may scan for their relevance to the search task. Tocompare their strategy with the standards covered in the instruc-tional phase of their learning, self-guided explorers must eitherrecall those standards or refer to the available instructional infor-mation. There is no direct feedback from the task or any externalsource regarding the adequacy of the search strategy used or thestandards for judging the outcomes.

Self-Regulatory Mechanisms

Acquiring information through electronic search requires morethan the mechanical application of rules to a database. For a personwith basic operational competencies, the effective use of searchskills is a challenging process, which requires a robust belief inone's capability to master the task, positive affective reactions toprogress achieved, and a sustained interest in the task (Bandura,1997). The impacts of the progressive mastery and modelingcomponents of guided exploration on posttraining performancehave been found to be mediated through perceived self-efficacyand affective self-reactions (Bandura, 1997; Gist, 1989; Gist,Schwoerer, & Rosen, 1989; Williams, 1990). Enactive explorationhas been shown to lead to increased intrinsic motivation (Wood,Kakebeeke, et al., 2000).

Performance of complex activities that have extensive explor-atory requirements is strongly influenced by beliefs of personalefficacy. Perceived efficacy affects problem-solving performance,both directly and by its impact on other self-regulatory determi-nants (Bandura, 1997). Such beliefs influence the choices peoplemake, their goal aspirations, how much effort they mobilize in agiven endeavor, how long they persevere in the face of difficulties,their vulnerability to stress and depression in coping with taxingdemands, and their resilience to adversity. With regard to cognitivefunctioning, people of high efficacy display strategic flexibility,are quicker to discard faulty strategies in search of better ones, areless inclined to prematurely discard good solutions, use good timemanagement, and interpret deficient performance in informativerather than in self-debilitating ways (Bandura, 1997; Bouffard-Bouchard, Parent, & Larivee 1991; Wood & Bandura, 1989).

Affective reactions to the task will also influence the behavioraland cognitive functioning required for effective electronic search.If successful performance is obtainable solely by enhanced effort,then level of dissatisfaction raises performance accomplishments(Bandura & Cervone, 1983,1986). As tasks become more complexand performance is more strongly determined by strategies, self-dissatisfaction may initially spur increased effort but underminesubsequent motivation and strategic thinking (Bandura & Jourden,1991; Cervone, Jiwani, & Wood, 1991). Negative affect can in-terfere with the higher level cognitive processing required forgenerating options, interpreting feedback, and adapting strategiesaccordingly.

Intrinsic motivation can also affect search performance throughseveral different pathways (Bandura, 1997; Deci & Ryan, 1980;Dormann & Frese, 1994). Intrinsic motivation is typically associ-ated with freedom from distracting negative self-evaluations and astrong task focus, both of which can benefit performance on

complex cognitive tasks. Enjoyment of an intrinsically motivatingtask can also lead to persistence and a greater willingness toexplore and experiment with alternative strategies. When tasks areexperienced as boring or frustrating, individuals are more likely tobecome distracted, to spend less energy exploring options, and tolook for opportunities to quit (Deci & Ryan, 1980; Dormann &Frese, 1994).

Figure 1 presents a path diagram of the relationships betweenthe variables mentioned in the preceding discussion. On the basisof the arguments presented, the following specific hypotheses wereadvanced for the impacts of the exploration mode used in practice,following instruction in the basic competencies of search.

Hypothesis 1: Guided exploration will produce stronger self-efficacyand greater satisfaction following practice than enactive exploration.

Hypothesis 2: Enactive exploration will produce greater intrinsicmotivation following practice than guided exploration.

Hypothesis 3: Enactive exploration will produce greater wasted effortin the form of errors, repetition of search lines, and incompletedevelopment of search statements in posttraining tasks than guidedexploration.

Hypothesis 4: Guided exploration will lead to higher quality strategiesand higher performance in posttraining tasks than in enactiveexploration.

Hypothesis 5: The impacts of exploration mode on wasted effort,strategy, and performance in posttraining tasks will be mediatedthrough the intrinsic motivation, self-efficacy, and satisfaction levelsof participants.

MethodThe 2 X 2 experimental design included mode of exploration (guided vs.

enactive) as a between-participants factor and two posttraining search tasks(Task 1 and Task 2) as a within-participant factor. Records of the searchesconducted on the two posttraining tasks provided measures of effort,strategy quality, and performance. Self-regulatory measures were collectedpre- and posttraining and after each of the posttraining search tasks. Duringtraining, all participants received the same instructional segment coveringthe basic competencies of electronic search. A practice segment thatincluded the manipulations of exploration modes followed. All participantswere novices in electronic search.

Participants

The participants were 48 university students undertaking an introductoryeconomics subject. Prior testing established that all participants had little or

Self-Efficacy

StrategyQuality

Performance

Figure 1. Schematic diagram of the hypothesized relationships betweenexploratory mode, self-regulatory reactions, effort, strategy quality, andperformance on electronic search tasks.


no prior experience in the conduct of electronic searches (Internet, etc.),and no participant had ever conducted an electronic search of libraryrecords, which was the task used in this study. The sample included 25male and 23 female students with a mean age of 19 years. They wererandomly assigned to treatment conditions balanced for sex.

Electronic Search Tasks

One form of electronic search that is of growing importance to a widerange of decision making is the interrogation of bibliographic databases.The records stored in bibliographic databases include published journalarticles, unpublished papers, books, technical reports, videos, and movies.Each record within a database is indexed under a range of keywords.Participants were assigned problems containing sets of concepts for whichthey had to find the relevant records from a bibliographic database.

To search the database, participants had to construct search statementsthat contained terms, known as keywords and connectors. The terms usedto define the concepts in a search problem can be obtained from personalknowledge, the database keyword index, the thesaurus, or interim feedbackfrom inquiries. The number of keywords for a concept will depend on thenumber of subconcepts and indicators encompassed by the concept. Asearch that incorporates keywords listed in the database index is morelikely to retrieve relevant records, but personal judgment in the indexingprocess can provide an unpredictable element in the relevance of searchoutcomes. The thesaurus can be used to expand and redefine concepts, andit is therefore of greater value when the concept can be defined in a numberof ways. However, it also requires more cognitive effort to identify key-words by using the thesaurus than the keyword index. More sophisticatedsearchers will review records in the interim feedback for relevant keywordsand then use the thesaurus to link these to other keywords.

The keywords or terms included in a search statement are linked togetherwith Boolean connectors, including the words and and or. The and con-nector narrows the scope of the search to those records that are indexedunder all of the connected terms. The more keywords connected, the morefocused the inquiry becomes and the fewer the records likely to beretrieved. The or connector broadens the focus of the search to retrieve allrecords that are indexed under any of the connected terms. This canproduce a large number of records, depending on the diversity of theincluded concepts.

Each search statement retrieves a certain number of records from thedatabase. These records can be inspected on the computer screen andassessed for their relevance. On the basis of these assessments, searcherscan then modify their inquiry and interrogate the database further toincrease the number of relevant records and reduce the number of irrele-vant ones. The inquiry process can proceed through a number of iterationsuntil the searcher retrieves an appropriate set of information.

Modes of Exploration

The training that included the manipulation of the exploration modeswas conducted in two segments: an instructional segment followed by apractice segment. The format of a brief instruction segment followed bypractice opportunities was similar to that used in other studies of computersoftware training (e.g., Martocchio & Dulebohn, 1994). All aspects of theinstructional segment of the training were the same in both conditions andprovided participants in both conditions with the same information. Themanipulations of enactive exploration and guided exploration modes wereintroduced in the practice segment. The instruction and practice segmentseach took 30 min and were administered to individual participants by afemale librarian who was highly experienced in CD-ROM searching andtraining.

In the common instructional segment, all participants were providedwith an overview of the CD-ROM database, descriptions of the mosteffective search strategies for retrieving documents, and instruction in the

use of the various computer operations and their functions. The search toolswere described, and the instructor demonstrated how to use the QuickReference Guide, the Thesaurus, the Keyword Index, and the On-Line Helpand Guide services for the CD-ROM database. The phases of inquiry in aneffective search strategy were described and demonstrated through anexample search for information regarding the effects of mass media onviolent behavior. The sequential strategy presented to participants wastaken from the supporting documentation for the Silverplatter search ser-vice, which was the source for the CD-ROM databases searched byparticipants in this study. The basics of the strategy presented were thesame as those presented in the search documentation for most commer-cially available databases.

Each of the prescribed eight steps in the strategy were described and thendemonstrated by the instructor. The eight steps were as follows: Step 1:Select a concept; Step 2: Identify several correct terms to describe theconcept, using self-selected keywords, the index, or the thesaurus; Step 3:Link these with or, Step 4: Repeat Steps 1 to 3 for additional concepts untilall are defined; Step 5: Link groups of terms for each concept with and tofind the common records; Step 6: Evaluate the suitability of the recordsobtained; Step 7: Rework any inappropriate concept terms by repeatingSteps 2 and 3; and Step 8: Repeat Steps 5 and 6 and continue until most ofthe records retrieved are related to the assigned topic.

At each step, the instructor demonstrated alternative actions and thenhighlighted the relevant aspects of the outcomes for the participant. Theparticipant was then asked to repeat the actions. For example, at Step 2, theinstructor would say, "I can just think of some words to use as terms for themass media concept, such as ...," which she would write down. Then shewould ask the participant to do the same for the violent behavior concept.Next, she would say, "I can also use the keyword index to find some termsthat I might not have thought about," and she would then demonstrate theuse of the keyword index by locating several terms and writing them down.The participant would do the same for the violent behavior concept.Finally, the instructor would say, "I can also use the thesaurus to identifyrelated words that are not obvious from the keyword index," and she wouldlocate two different synonyms in the thesaurus and write down the relatedterms. The participant would then do the same for the violent behaviorconcept. For Steps 3 to 5, the instructor entered some search statementsinto the computer and drew the participant's attention to the salient aspectsof the output. The participant then repeated the process. All participantswere given a one-page summary of the eight-step strategy process, whichthey were able to refer to at any stage during the practice segment and theperformance tasks.

Following the instructional segment, participants completed a practicesegment under either enactive or guided exploration instructions. Partici-pants in both conditions explored the same five practice topics. They were(a) the effects of alcohol use on academic performance, (b) the effects ofchild care on social skills, (c) issues related to migrants learning anotherlanguage, (d) the link between physical activity and personality develop-ment, and (e) the impact of computer technologies on the design ofbuildings.

Enactive exploration included instructions that covered self-guided ex-ploration, the use of an on-line help facility that provided documentation ofstrategies and responses to problems, and the positive framing of errors(Frese et al., 1991; Wood, Kakebeeke, et al., 2000). Participants were toldthey could complete their inquiries on the five practice topics in any orderor practice on topics of their own devising, as they would do if they werelearning how to conduct searches by themselves. They were encouraged toexplore and to practice, using the competencies they had learned in theinstructional segment. The sequencing of tasks, the steps in the strategyprocess practiced, and the responses to problems were left to the discretionof the individual participant.

Enactive exploration participants had access to an on-line help facilitywith supporting printed information. The help facility was demonstrated sothat participants knew how to access information, and they were reminded


of its availability when they encountered problems. They also had thesummary sheet, which described the eight-step strategy that had beenpresented in the instructional segment. The on-line help facility and strat-egy summary sheet, which are commonly available to people who conductelectronic searches, meant that participants in the enactive explorationcondition had available to them the same information that was provided toguided exploration participants through the modeling of responses. How-ever, the decision to refer to them during practice was at the discretion ofthe individuals in the enactive exploration mode.

To encourage exploration during practice, we gave the enactive explo-ration participants instructions to frame errors and problems positively.These error management instructions were adopted from Frese et al. (1991)and were summarized in two signs that were placed on top of the computerin the participants' line of sight. The first sign stated the following: "If youstrike a problem, regard it as a learning opportunity; I have made an error:Great! There is a way to solve this problem. I can team from this error."The second sign stated the following: "Don't forget to watch the screen;view the screen and see what is changing." These principles were stated bythe experimenter at the start of the practice session and whenever partic-ipants encountered problems.

Before commencing the practice segment, participants were asked ifthey understood that they could work through the tasks in whatever ordersuited them, and they all said yes. All participants also indicated that theyknew how to use the on-line help facility. The participants were asked toread the error management signs, and the instructor reinforced the errormanagement message during the practice session. If participants in theenactive exploration mode could not solve a problem after 3 min, then theinstructor corrected the error and reminded them of the on-line help facilitybut did not explain the solution. Only 2 participants required this assis-tance, each on one occasion. In one case, the person continually omitted thesymbol used to identify lines of search. In the second case, the personcontinually left no spaces between keywords. On all other occasions,participants in the enactive exploration condition corrected errors wellwithin 3 min. The number of practice tasks completed varied according tothe exploration interests of the individual.

Guided exploration included instructions on the order for completing thepractice tasks, reminders of the steps to follow in developing a strategy foreach task, and modeling of responses that followed problems or mistakesin the use of strategies during practice. All guided exploration participantscompleted the five practice topics in the same assigned order, progressingfrom the easiest to the more difficult searches, and they were instructed towork sequentially through the eight steps of the prescribed strategy foreach topic. If the participant did not follow the prescribed strategy or madea mistake, then the instructor would model a response for that step,including a statement of the rationale that guided the selection. For exam-ple, if the participant simply started entering keywords on Step 2 for thefirst topic, then the instructor would say, "Drinking is one of the terms Ihave identified for alcohol use, but I know I will have more success if I findsome other related keywords from the thesaurus." She would then open thethesaurus and point to some relevant keywords. If participants skipped astep, then the instructor would draw their attention to the strategy summarysheet and remind them of the benefits of following the prescribed sequence.The modeled responses by the instructor only demonstrated points that hadpreviously been covered in the common instruction segment, and theyfollowed scripted procedures taken from the on-line help facility that wasavailable to the enactive exploration participants during the practicesegment.

In summary, the sequencing of tasks, the steps in the strategy processpracticed, and the responses to problems were externally directed in theguided exploration mode but were left to the discretion of the individualsearcher in the enactive exploration mode. Participants in the two explo-ration modes received the same information and level of instruction duringthe instructional segment, and no new information was introduced in thepractice segment of the training program. The information on responses to

problems encountered in practice was, however, presented differently inthe two exploration modes. The experimenter modeled this in the guidedexploration condition, whereas participants in the enactive explorationcondition had to either refer to the on-line help facility (which was alsoavailable as a written document on their desk) or rely on recall of infor-mation from the instruction segment.

Postinstruction Performance Assessment

Following the practice session, participants completed inquiries on twodifferent tasks. One inquiry asked for information on the effects of televi-sion watching on teenagers in references published since 1988, and theother one asked about the impacts of technology on teacher satisfactionand performance. Half of the participants in each exploration conditioncompleted the television effects task first and the technology effects tasksecond. The other half of the participants completed the two searches in thereverse order.

Personal Variables

At the outset of the study, background information was collected onparticipants' age; year of study; and previous computing, library, andelectronic search experiences. Background measures also assessed theparticipants' experience in searching library databases and manual indexes.The perceived effectiveness of the instructional programs and intrinsicmotivation were assessed at the end of the practice segments of the trainingprogram. Perceived self-efficacy and self-satisfaction were assessed at fourpoints: at the beginning of the study, at the end of the practice segment, andafter each of the two searches in the postinstruction assessment phase.

Self-efficacy was measured with a 27-item scale, which covered thediverse activities in the four subfunctions of the electronic search process:problem definition, keyword identification, use of connectors and theconstruction of search statements, and participants' beliefs in their ex-pected search capabilities to achieve different levels of performance. Theactivities were obtained through observations of professional CD-ROMoperators, conducting searches and a review of the literature on successfulelectronic search strategies (e.g., Michel, 1994; Quint, 1991). "I canidentify the major requirements of the search from the initial statement ofthe topic" is a sample item. For each item, participants first recordedwhether they could perform the activity described ("yes" or "no") and thenrecorded their confidence in their capabilities on a 10-point scale, with 1representing very low confidence and 10 representing very high confidence.Reliability estimates for each of the four assessments of self-efficacy werehigh (as > .95).

Satisfaction was assessed with six items in which participants rated theirlevel of satisfaction with their search skills and performance. The scaleincluded items such as "At this point in time, how satisfied are you withyour skill in conducting CD-ROM searches?" Participants rated theirsatisfaction on a 10-point scale ranging from 1 (not at all satisfied) to 10(highly satisfied). The reliability estimates for each of the four assessmentswere at acceptable levels (as > .77).

Intrinsic motivation was assessed with seven items, including four itemsfrom the measure developed by Mossholder (1980) and three items fromDaniel and Esser (1980). Mossholder's (1980) items assessed participants'desire to continue working on the task, their level of interest in the activity,their perceived degree of challenge, and their satisfaction with the task. Theitems were anchored with 1 (not at all) and 7 (to a large degree). Danieland Esser's (1980) items were 7-point semantic differential scales with thefollowing anchors: monotonous—exciting, boring—interesting, andstimulating—dull. Principal-components factor analysis revealed that theseven items from the combined scales loaded on a single factor with highinteritem reliability (a = .94).

Perceived training effectiveness was participants' ratings of the modesinstructions. Participants rated whether they found the process easy to


understand, whether too much information was provided, and whether theycould easily recall the procedural strategy steps. They recorded theirjudgments on a 7-point scale, anchored by 1 (totally disagree) and 7(totally agree). The reliability coefficient for the three-item scale wasacceptable (a = .80).

Search Activity and Performance Attainment

Printouts of the search records were used to assess effort, strategyquality, and performance levels.

Total effort in electronic search is reflected in the number of searchstatements developed and how extensively they are used. A search state-ment consists of keywords, other terms, and connectors; and it can belinked to other search statements by using additional connectors. Theidentification of terms and connectors requires significant cognitive effort.Thus, one indicator of effort is the number of terms and connectorsincluded in search statements (Salterio, 1996). Time spent on the task isanother indicator of effort. Participants had a maximum of 30 min to searchfor each of the problems in the testing phase, but they could spend less timeif they wished. An additional effort indicator is the number of differentlines of inquiry produced by the participant and integrated into the search.The computational lines of inquiry generated by the operating system andnot developed directly by the participant were not included in this measureof effort. An aggregate index of total effort was created by summing thestandardized scores for the three components of effort, which had accept-able levels of internal reliability for both the first and second tasks (a, =.83, a2 = .81).

Wasted effort was measured in three ways: the number of repetitive orredundant lines of search generated, the lines of inquiry developed andrejected by the searcher in the process of developing the final searchstatement, and the number of errors. An aggregate measure of wasted effortwas created from the standardized scores of these three components for thetwo tasks («! = .70, a2 = .83). Wasted effort is a component of total effort.Any increase in wasted effort due to rejected lines of search inputs must beassociated with an increase in the number of lines and terms used tocalculate total effort, as both measures are drawn from the same lines ofsearch inputs. However, the reverse is not true. A searcher could input largenumbers of search statements without rejecting any lines of search. Theother components of the two effort measures are not directly related bysource of measurement.

Strategy quality was a composite measure of the depth, breadth, se-quencing, and sources used in the searches for the two tasks. Strategieswere extracted from the printouts of search statements used to explore eachof the assigned tasks. The expert librarian who conducted the training laterscored each of the search strategies on the four properties, as describedbelow. Hard copies of the 96 search records (two search tasks per subject)were given to the librarian 2 weeks after all the training was completed. Noinformation on the searches identified the participant or the condition inwhich the search was completed. Randomization of the order also meantthat any 2 consecutive searches were not necessarily from the sameparticipant. To establish the reliability of the scores, a second librarianscored a 20% random sample of the search printouts. A common scoringchecklist was used to ensure the same method of assessment throughout thescoring processes of both raters. This scoring sheet was tested in a priorpilot study, and it had been compiled from an analysis of 40 earliersearches. The interrater reliability across both tasks for the overall strategyscore was r = .95.

Depth of search was calculated from the number of different keywordterms used to describe each concept that was linked with the or connector.The more keywords used for each concept in the search process, the greaterthe depth of search for that concept. The depth score for each search rangedfrom 0 to 12.

Breadth of search was a measure of the number of concepts covered by

terms included in search statements and correctly linked by using either theand connector or other concept refinement strategies such as field limiters,truncation, and the not connector. Points were allocated for each of theseelements of breadth. The more concepts effectively linked in a search, thegreater the breadth of the search. The scores ranged from 0 to 12 on eachof the assigned topics.

Search sequence was operationalized as the degree to which the searchercombined the different steps into a coherent sequence, proceeding fromexploration of single concepts in depth to a broader search of multipleconcepts. This measure identified the degree to which the adopted searchmatched the procedural strategy outlined earlier, which has been demon-strated to be a highly effective search strategy (Michel, 1994; Quint, 1991).Participants had a description of the optimal sequence to refer to duringpractice. The scores for quality of search sequence on each search problemranged from 0 to 12.

Sources of terms included in the search statement were the fourthcomponent of strategy. Terms could be obtained from four differentsources that varied in the difficulty of their use and the quality of terms.They included the initial search statement; the thesaurus; the keywordindex; and other sources, such as the searchers' own knowledge andretrieved records. Thesaurus keywords are the most accurate in theirdetailed depiction of the topic, but they are also the potentially mostdifficult to use in search statements. The extent of thesaurus usage providedan operationalization of quality of sources.

The standardized scores for depth, breadth, sequence, and sources werecombined to create an aggregate score for search strategy. The internalreliabilities for the strategy composites on the first and second tasks wereacceptable (a^ = .89, «2 = .71).

Performance was the number of relevant records retrieved by the finalsearch statement for each task. A record was categorized as relevant ifterms for concepts from the assigned topic were included in the descriptorsor the abstract. The librarians who scored strategy also scored performance.The search outputs were scored randomly and contained no information toidentify the participant or the condition in which the search was conducted.A 20% random sample of search outputs was scored by the secondlibrarian, yielding an interrater reliability across both test tasks of r — .97.

Results

Before we examined the pattern of relationships between study

variables or tested the study hypotheses, we conducted analyses totest for potential differences in responses that were due to prop-erties of the two performance tasks, the order in which the tasks

were completed, or participant characteristics. All participants inboth exploratory mode conditions completed the same two tasks,

but the order was counterbalanced so that in each condition half ofthe participants completed the television effects task first and theother half completed the technology effects task first. With taskand order of completion coded as dummy variables, there were no

differences for task or order in which the two tasks were completedfor any of the self-regulatory or outcome measures. The scores for

the two performance tasks were standardized and combined for allanalyses of the first (Task 1) and second (Task 2) tasks completedafter the training.

Analyses of variance showed that the groups were comparablein age, level of prior search familiarity, and previous exposure to

CD-ROM instruction. Although all participants were recruited

from the same cohort and were randomly assigned to conditions,the guided exploration group was found to have more years of

academic study than the enactive exploration group, F(l,46) = 9.10, p < .01, T)2 = .02, and reported greater prior expe-


rience with computers, F(l, 46) = 4.79, p < .05, if = .09. Thesetwo personal variables were included as covariates in all reportedanalyses of covariance (ANCOVAs), unless specifically notedotherwise. All other variables were comparable for both groups,including CD-ROM experience, F(l, 45) = 2.93, ns, which wasthe most relevant demographic factor. The groups did not differ inself-efficacy or satisfaction at the outset, Fs(l, 46) < 1.0, ns.

Table 1 presents the means, standard deviations, and correlationmatrix for the study variables. The overall pattern of correlationswas generally consistent with the relationships proposed in Fig-ure 1, although the strength and significance of some relationshipsvaried between Task 1 and Task 2. Some of the correlations inTable 1 were contrary to our expectations and require noting.Intrinsic motivation was not significantly related to explorationmode or to any of the effort, strategy, or performance measures.Total effort correlated highly with wasted effort on both tasks butnot with strategy quality.1 Those who worked harder were morelikely to waste their effort in errors, incomplete searches, andredundancies. Another interesting result was the lack of a relation-ship between Task 1 and Task 2 performance (r = .02). The issuesraised by these results are addressed in the Discussion section.

Effects of Exploration Mode on Self-Regulatory Factors

Hypothesis 1 was supported. ANCOVAs showed that guidedexploration was more effective in developing posttraining self-efficacy, F(l, 45) = 19.15, p < .001, ff = .30, and satisfaction,F(l, 45) = 27.38, p < .001, if = .37, than enactive exploration(Figure 2). The self-efficacy of those who had practiced underguided exploration remained stronger than that of their enactiveexploration counterparts after the first, F(l, 47) = 23.32,p < .001,if = .34, and second performance tasks, F(l, 47) = 19.78, p <.001, if = .30. In a similar manner, the greater satisfaction ofguided exploration participants was still evident after the first, F( 1,47) = 33.75, p < .001, if = .43, and second tasks, F(l,47) = 23.16, p< .001, if = .35.

Hypothesis 2 was not supported. Intrinsic motivation did notdiffer significantly between the two modes of exploration. Thisunexpected result was somewhat contrary to impressions gainedfrom direct observation of participants' behavior during the prac-tice segment. During their practice searches, the enactive explora-tion participants often asked questions of themselves such as "Howdid I do that?" "What do I do now?" "What happens if I do this?"and "Let's see what else I can do." They moved freely betweenpractice problems and appeared to be highly engaged in the pro-cess. Despite an apparent enjoyment of the process, enactiveexploration participants evaluated the experience as less effectivethan guided exploration participants, F(l, 46) = 23.56, p < .001.

Effects of Exploration Mode on Search Effort

Hypothesis 3 was supported. ANCOVAs revealed that the en-active exploration participants wasted more effort across the twotasks, F(l, 45) = 13.55, p < .001, if = .23 (Figure 3). Theyrejected more lines of inquiry, F(l, 45) = 7.83, p < .01, if = .15,and they were more prone to errors, F(l, 45) = 17.93, p < .001,if = .28. An analysis of variance without the two personalcovariates showed that the enactive exploration participants alsoengaged in more redundancies than did their counterparts in

guided exploration, F(l, 45) = 4.55, p < .05, if = .09. However,this result was not significant in the ANCOVA with prior com-puter experience and years of academic study included as covari-ates, F(l, 43) = 3.38, ns, if = .07.

Effects of Exploration Mode on Strategy Qualityand Performance

Hypothesis 4 was supported for both strategy quality and per-formance. Repeated measures ANCOVAs showed that guidedexploration participants demonstrated better quality search strate-gies, F(l, 44) = 46.45, p < .001, if = .51, and retrieved morerelevant records, F(l, 44) = 6.53, p < .05, if = .13, across thetwo tasks than did the enactive exploration participants (Figure 4).The interaction terms for exploration mode by tasks were notsignificant for either strategy quality or performance.

More detailed analyses of the strategy components revealed thatthe guided exploration participants conducted a broader search onthe first task, F(l, 46) = 11.32, p < .001, if = .20, but, withexperience, the enactive exploration group matched them in theirsearch breadth on the second task, F(l, 46) = 0.02, ns. Guidedexploration participants conducted searches of greater depth, F(l,46) = 23.96, p < .001, if = .34, and better sequencing, F(l,46) = 44.34, p < .001, if = .49, across both tasks. They also usedmore informational sources, relying more heavily on the thesaurusfor their keyword selection, whereas their enactive explorationcounterparts made limited use of this important source of key-words, F(l, 46) = 54.28, p < .001, if = .54.

The results of the quantitative analyses for the components ofwasted effort and strategy quality were supported by qualitativeassessments of the search records. The librarian who scored theparticipants' search strategies was instructed to review the searchrecords a second time and to identify an example of the mostcommon search behaviors from each search record. This reviewresulted in two examples for each participant. A review of theseexamples revealed that the enactive exploration participants oftenused the same terms repetitively, experimenting with differentlinks and combinations, but not adding new terms that increasedthe depth of search for concepts. The other common enactiveexploration strategy was to use different terms, only to discardthem in favor of new terms, rather than to develop a coherent lineof inquiry by adding the terms together in a single search state-ment. By this approach, they used many terms, but there was littlestrategy refinement or integration despite high effort. Guided ex-ploration participants drew terms from the thesaurus and the indexand, as shown in the results of the quantitative analyses, moreeffectively integrated these terms in their search statements thantheir enactive exploration counterparts.

Mediating Role of Self-Regulatory Factors

Hypothesis 5 received very limited support. The predicted me-diation roles of self-efficacy, satisfaction, and intrinsic motivationfor differences in wasted effort, strategy quality, and performance

1 With rejected lines of work excluded from the wasted error measure,the correlations between total effort and wasted effort for Tasks 1 and 2,respectively, were rl = .48, p < .01, and r2 = .59, p < .01.


Table 1Alpha Coefficients, Means, Standard Deviations, and Correlation Matrix for First and Second Posttraining Performance Tasks

Variable 1

1. Exploration mode2. Age3. Years of study4. Prior computer experience

.27

.16-.39**

5. Perceived training effectiveness .48**6. Self-Efficacy 17. Self-Efficacy 28. Self-Efficacy 39. Self-Efficacy 4

10. Satisfaction 11 1 . Satisfaction 212. Satisfaction 313. Satisfaction 414. Intrinsic motivation15. Total effort Task 116. Wasted effort Task 117. Strategy quality Task 118. Performance Task 119. Total effort Task 220. Wasted effort Task 221. Strategy quality Task 222. Performance Task 2

TotalMsSDs

Enactive explorationMsSDs

Guided explorationMsSDs

Guided vs. enactive

.02

.55**

.58**

.55**

.14

.61**

.65**

.59**

.18-.26-.38**

.74**

.28t-.30*-.52**

.60**

.32*

0.50

0.00

1.00

.36*

.15-.04

.11-.14-.12-.14

.07-.04-.01-.06-.03-.08-.02

.12-.19-.25-.13

.03

.22

19.401.22

19.211.06

19.581.350.37

.29*

.19

.12

.28

.22

.15

.30*

.36*29*.31*.04

-.13-.03

.35*

.03-.21-.13

.17

.21

2.040.54

1.830.56

2.250.440.95**

.09

.28

.43**

.30*

.38**

.48**

.43**

.23

.42**

.18

.14

.08

.49**-.14-.02-.23

.27

.39*

8.572.03

7.961.68

9.222.191.26*

—.18.64**.60**.58**.17.70**.65**.61**.41**

-.01-.25

.37**

.17-.07-.05

.28

.10

5.641.21

5.061.26

6.210.831.15**

(.95).33*.20.28.45**.35*.15.34*

-.16.20.09.01

-.24.09

-.06-.04

.32*

3.161.46

3.121.44

3.191.510.07

(.97).86**.81**.26.82**.64**.71**.39**.13

-.12.52**.14

-.06-.22

.40**

.32*

5.962.02

4.861.75

7.051.672.19**

(.98).92**.21.70*.82**.71**.42**

-.03-.29*

.62**

.26

.03-.19

.54**

.35*

6.182.10

4.971.68

7.391.792.42**

Note. Alpha coefficients are shown in parentheses on the diagonal. Exploration mode was dummy coded so that enactive exploration = 0 and guidedexploration = 1. For guided vs. enactive, Fs are from 1 X 2 analysis of variance for each study variable.tp<.06. *p<.05. **/>< .01.

by participants in the two exploration modes were tested sepa-rately, using the three-step hierarchical regression procedure rec-ommended by Baron and Kenny (1986). The results of theseanalyses, which are shown in Table 2, revealed no significantmediation effects on Task 1 and partial support for the mediationof performance differences on Task 2. The significant effect forexploration mode on Task 2 performance (ft = .31, p < .05) wasreduced to nonsignificance (/3 = .14, ns) with the introduction ofself-efficacy, satisfaction, and intrinsic motivation into the regres-sion equation. However, the reduction in the standardized beta forexploration mode did not reach significance (t = 1.0, ns).

The regression analyses for the mediation hypothesis did revealthat the effects of exploration mode on wasted effort and strategyquality were independent of participants' motivational reactions tothe tasks. With self-efficacy, satisfaction, and intrinsic motivationincluded as predictors in the regression models, exploration mode(dummy coded as enactive exploration = 0 and guided explora-tion = 1) was a significant negative predictor of wasted effort(0i = -.53,p< .01;/32 = -.53, p< .01) and a positive predictorof strategy quality (j3j = .59, p < .01; /32 = .38, p < .05), on bothTask 1 and Task 2.

Discussion

As we hypothesized, practice under guided exploration pro-duced higher levels of perceived self-efficacy, satisfaction, strat-egy quality, and performance and lower levels of wasted effort onelectronic search tasks than self-guided, enactive exploration. Fol-lowing instruction in the basic competencies, participants who hadthe benefit of guided exploration during practice worked smarterand more efficiently and exhibited less wasteful effort on laterperformance tasks. They spent less time on the tasks, developedbetter strategies, and reduced the amount of effort they needed toacquire the required information for assigned topics. These differ-ences are especially noteworthy because participants in the enac-tive exploration condition were instructed in the effective strategysteps at the outset and had a description of these steps to refer tothroughout the performance of the search tasks. This is a consid-erable aid if the person chooses to use it. The enhanced levels ofself-efficacy and satisfaction produced by guided exploration inpractice remained when participants conducted their searches forthe performance tasks without any guidance.

An important feature of the electronic search task used in thisstudy is that the natural task feedback provides little direct guid-


10 11 12 13 14 15 16 17 18 19 20 21 22

(.98).25.69**.74**.79**.40**

-.01-.26

.57**

.15

.02-.17

.52**

.33*

6.702.10

5.562.01

7.831.492.27**

(.77).37*.28.37*

-.01.06.09.32*

-.01.18.10.23.43**

3.271.18

3.120.97

3.441.380.32

(.94).75**.79**.38**.10

-.07.61**.14.06

-.13.49**.34*

6.141.71

5.101.48

7.171.252.07**

(.95).77**.41**

-.17-.34*

.63**-.32*

.05-.16

.57*

.34*

6.541.77

5.381.47

7.681.232.30**

(.96).41**

-.09-.26

.56**

.13-.01-.18

.52**

.35*

7.011.77

5.951.74

8.021.092.07**

(.94).18.05.12

-.09.07.01.13.02

4.890.97

4.720.81

5.071.100.35

.67**-.05-.05

.55**

.36*-.11-.05

53.9222.53

59.7124.25

48.1219.4811.59

-.31*-.14

.44**

.53**-.22-.28

24.6026.80

32.9624.22

16.2527.1316.71**

.36*-.03-.28

.67**

.50**

3.472.03

1.970.81

5.001.763.03**

.07-.06

.37*

.02

2.632.61

1.922.14

3.332.871.41f

.61**

.08-.13

50.4725.89

58.3932.61

42.8814.1715.51*

—-.26-.19

25.1737.98

42.3948.26

8.678.76

33.72*

.35*

3.651.57

2.711.10

4.591.761.88**

—

3.237.01

1.041.55

5.429.384.38*

ance for the novice in regard to the efficacy of strategies. Althoughthey can inspect the records they retrieve with each inquiry andjudge their relevance for the problem being explored, the novicesoften lack the knowledge needed to categorize records and have noway of knowing whether there are other relevant records that theyhave not retrieved. The lack of both clear standards and directivetask feedback contributes to unsystematic variations in perfor-mance by novices, which can weaken the relationship betweenperformance scores on different tasks. The nature of the taskfeedback will also weaken the relationship between self-regulatoryreactions and task performance, as discussed below.

Contrary to our prediction, enactive exploration did not producegreater intrinsic motivation for electronic inquiry. Encouragingpeople to explore and to make errors can lead to heightenedinterest, persistence, and better performance on tasks where self-guided exploration quickly leads to discoveries that can be used toimprove performance (Dormann & Frese, 1994). However, self-guided exploration by novices is less effective when feedback doesnot facilitate learning or positive motivation. Moreover, intrinsicmotivation did not contribute to the prediction of strategy qualityor performance. Intrinsically motivated interest in a task will notnecessarily be translated into superior performance on tasks where

working harder and longer is no guarantee of success (Bandura &Jourden, 1991; Wood, Kakebeeke, et al., 2000).

The very limited support for the hypothesized mediating func-tions of the self-regulatory reactions may also be due to the natureof task feedback from electronic search. When standards are un-clear and feedback does not provide a clear indication of progress,self-regulatory reactions are less reliably related to future perfor-mance (Bandura, 1997; Cervone et al., 1991). Even if novices setgoals based on the records retrieved, they may be using totallydifferent standards from those used by experts to categorizerecords as relevant. In the absence of interpretable and reliableevidence regarding their progress on the task, participants' efficacyjudgments and their feelings of satisfaction were based on self-assessments developed during the instruction and practice seg-ments and were not related to their actual achievements on theperformance tasks. The mean levels of self-efficacy and satisfac-tion increased between the pre- and posttraining measures and thenremained relatively stable after each of the performance tasks(Figure 2). The two measures were also strongly autocorrelatedacross the posttraining assessments (Table 1). Thus, experience onthe two tasks did not strongly influence participants' self-regulatory reactions. As individuals became more adept at inter-


Self-Efficacy Strength

• Guided Exploration

Enactive Exploration

Satisfaction Level

1 2 3Assessment Phase

2 3Assessment Phase

Figure 2. Self-efficacy and satisfaction for guided exploration and enactive exploration groups at pretraining,posttraining, Posttask 1 and Posttask 2 assessment phases.

preting feedback and deriving the connection between their searchstrategies and records retrieved, self-regulatory reactions shouldbecome more closely aligned to strategy quality and performance.

The results of the mediation analyses did show that the searchpatterns used by the participants during practice were carried overto their searches for the two performance tasks, independently ofany motivational effects due to self-regulatory reactions to thesearch task. The low fidelity of task feedback from electronic

search will often mean that inexperienced searchers will persistwith the behaviors that they develop during training and practice.For novices who practice under self-guided exploration, this ismore likely to include suboptimal strategies and inefficientbehaviors.

The results of this study do not rule out the possibility thatself-guided exploration will produce stronger motivation, bettertask understanding, and more effective strategies under different

12

10

8

6

4

2

0

Lines of RedundantSearch Statements

* *Guided Exploration

• •Enactive Exploration

.••*

>..*'

,•*

*^^^

"̂"̂

1 1

14

12

10

8

6

4

2

0

Lines of IncompleteSearch Statements

^*

^^^^

1 1

14

12

10

8

6

4

2

0

Number of Errors

•

*

«^^^""" ""*

J. _ .. 1

Taskl Task 2 Task 1 Task 2Task 1 Tasl

Figure 3. Effects of guided exploration and enactive exploration on level of wasted effort as represented bylines of redundant search statements, lines of incomplete search statements, and number of errors committed forPosttraining Tasks 1 and 2.


Strategy Quality Rating Relevant Records Retrieved

—* Guided Exploration

• • Enactive Exploration

Task 1 Task 2 Taskl Task 2

Figure 4. Effects of guided exploration and enactive exploration on strategy quality ratings and numbers ofrelevant records retrieved for Posttraining Tasks 1 and 2.

circumstances. Enactive exploration may be better suited to tasksthat include more informative feedback and to other forms ofelectronic inquiry, such as "surfing the net" to identify a particularWeb site, which requires minimal competency in search proce-dures and allows superficial processing of interim feedback(Wood, Atkins, & Tabernero, 2000). For complex, ill-structured

tasks that provide low-fidelity feedback, the evidence suggests thatguided mastery training plus extended guided exploration duringpractice is needed to build initial competencies before the benefitsof self-guided exploration will be realized. For example, Simonand Werner (1996) found that staff who learned how to use a newsoftware program under a self-guided exploratory mode had still

Table 2Standardized Betas and Adjusted R2s for the Three-Step Mediated Regression Analyses WithSelf-Regulatory Factors as Mediators of the Impacts of Exploration Mode on (a) Wasted Effort,(b) Strategy Quality, and (c) Performance on Posttraining Tasks 1 and 2

Task 1 Task 2

Analysis step

1

23

23

23

Dependent variable

Self-efficacySelf-satisfactionIntrinsic motivationWasted effortWasted effort

Strategy qualityStrategy quality

PerformancePerformance

Independentvariable

Exploration modeExploration modeExploration modeExploration modeExploration modeSelf-efficacySatisfactionExploration modeExploration modeSelf-efficacySatisfactionExploration modeExploration modeSelf-efficacySelf-satisfaction

ft

.55**

.61**

.18-.38**-.53**

.34-.10

.74**

.59**

.00-.36

,27t.28.08.04

AdjustedS2

.13**

.13*

.54**

.56**

.07t

.04

0

.58**

.65**

.19-.52**-.53**

.21-.16

.60**

.38**

.14-.21

.31*

.14

.18

.10

Adjustedfi2

.25**

.22**

.35**

.38**

.09*

.08

Note. Intrinsic motivation was not related to exploration mode or outcomes, either singly or in combinationwith the other self-regulatory variables. Therefore, it was not included in the subsequent steps of the mediationanalysis.tp<.06. *p<.05. **p<.01.


not matched the comprehension or skill levels of those taught bymodeling after 1 month of on-the-job experience with the program.

Electronic inquiry is a major element in present-day life, and itis likely to play an even bigger role in the acquisition of informa-tion with further advances in electronic technologies. Most peoplewill not receive extended training in electronic search, and expe-rience may prove a poor teacher because of the lack of directionprovided by task feedback. Support for self-improvement by peo-ple with basic electronic search skills is an important issue thatrequires further research. One area for future research is the impactof feedback interventions on the self-regulatory mechanisms andskill development of novice electronic searchers. Although thebenefits of some feedback interventions have been found to beproblematic (Kluger & DeNisi, 1996), those that enhance per-ceived efficacy and provide a functional task focus will lead toimproved understanding and performance on complex tasks (Ban-dura, 1997; Martocchio & Webster, 1992). For example, Martoc-chio and Dulebohn (1994) found that augmentation of task feed-back with information on factors under the control of the traineeled to greater understanding and better performance of a complexcomputer software program than supplemental feedback on factorsoutside the trainees' control. Interventions to augment task feed-back from electronic search could include similar types of diag-nostics as well as feedback on compliance with strategy steps orredundancies and other forms of wasted error.

References

Allwood, C, & Kalen, T. (1993). Using a patient administrative system: Aperformance evaluation after end-user training. Computers in HumanBehavior, 9, 137-156.

Bandura, A. (1986). Social foundations of thought and action: A socialcognitive theory. Englewood Cliffs, NJ: Prentice Hall.

Bandura, A. (1997). Self-efficacy: The exercise of control. New York:Freeman.

Bandura, A., & Cervone, D. (1983). Self-evaluative and self-efficacymechanisms governing the motivational effects of goal systems. Journalof Personality and Social Psychology, 45, 1017-1028.

Bandura, A., & Cervone, D. (1986). Differential engagement of self-reactive influences in cognitive motivation. Organizational Behaviorand Human Decision Processes, 38, 92-113.

Bandura, A., & Jourden, F. T. (1991). Self-regulatory mechanisms gov-erning the impact of social comparison on complex decision-making.Journal of Personality and Social Psychology, 60, 941-951.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variabledistinction in social psychological research: Conceptual, strategic, andstatistical considerations. Journal of Personality and Social Psychology,51, 1173-1182.

Borgman, C. L. (1989). All users of information systems are not createdequal: An exploration into individual differences. Information Process-ing and Management, 25, 237-252.

Bouffard-Bouchard, T., Parent, S., & Larivee, S. (1991). Influence ofself-efficacy on self-regulation and performance among junior and se-nior high-school age students. International Journal of BehaviouralDevelopment, 14, 153-164.

Brown, M. E. (1995). By any other name: Accounting for failure in thenaming of subject categories. Library and Information Science Re-search, 17, 347-385.

Cervone, D., Jiwani, N., & Wood, R. (1991). Goal setting and the differ-ential influence of self-regulatory processes on complex decision-making performance. Journal of Personality and Social Psychology, 61,257-266.

Collantes, L. Y. (1995). Degree of agreement in naming objects andconcepts for information retrieval. Journal of the American Society forInformation Science, 46, 116-132.

Condry, J. (1977). Enemies of exploration: Self-initiated versus other-initiated learning. Journal of Personality and Social Psychology, 35,459-477.

Cook, B. (1995). Managing for effective delivery of information servicesin the 1990s: The integration of computing, educational and libraryservices. In The virtual information experience: Proceedings of theseventh Australasian information online and on disc conference andexhibition (pp. 1-19). Sydney, New South Wales, Australia: AustralianLibrary and Information Science Association, Information ScienceSection.

Daniel, T. L., & Esser, J. K. (1980). Intrinsic motivation as influenced byrewards, task interest, and task structure. Journal of Applied Psychology,65, 566-573.

Deci, E. L., & Ryan, R. M. (1980). The empirical exploration of intrinsicmotivation processes. In L. Berkowitz (Ed.), Advances in experimentalsocial psychology (Vol. 13, pp. 39 - 80). New York: Academic Press.

Dormann, T., & Frese, M. (1994). Error training: Replication and thefunction of exploratory behavior. International Journal of Human andComputer Interaction, 6, 365-372.

Frese, M., & Altmann, A. (1989). The treatment of errors in learning andtraining. In L. Bainbridge & S. A. R. Quintanilla (Eds.), Developingskills with new technology (pp. 65-86). Chichester, England: Wiley.

Frese, M., Brodbeck, F., Heinboke, T., Mooser, C., Schleiffenbaum, E., &Thiemann, P. (1991). Errors in training computer skills: On the positivefunction of errors. Human-Computer Interaction, 6, 77-93.

Gay, G., & Mazur, J. (1989). Conceptualising a hypermedia design forlanguage learning. Journal of Research on Computing in Education, 21,119-126.

Gist, M. E. (1989). The influence of training method on self-efficacy andidea generation among managers. Personnel Psychology, 42, 787-805.

Gist, M. E., Schwoerer, C., & Rosen, B. (1989). Effects of alternativetraining methods on self-efficacy and performance in computer softwaretraining. Journal of Applied Psychology, 74, 884-891.

Greif, S., & Keller, H. (1990). Innovation and the design of work andlearning environments. The concept of exploration in human-computerexploration. In M. A. West & J. L. Fair (Eds.), Innovation and creativityat work (pp. 231-249). Chichester, England: Wiley.

Hatter, S. P., & Cheng, Y.-R. (1996). Colinked descriptors: Improvingvocabulary selection for end-user searching. Journal of the AmericanSociety for Information Science, 47, 311-325.

Kluger, A. N., & DeNisi, A. (1996). Effects of feedback intervention onperformance: A historical review, a meta-analysis and preliminary feed-back intervention theory. Psychological Bulletin, 119, 254-284.

Kuhlthau, C. C. (1991). Inside the search process: Information seekingfrom the user's perspective. Journal of the American Society for Infor-mation Science, 42, 361-371.

Latham, G. P., & Saari, L. M. (1979). Application of social-learning theoryto training supervisors through behavioral modeling. Journal of AppliedPsychology, 64, 239-246.

Martocchio, J. J., & Dulebohn, J. (1994). Performance feedback effects intraining: The role of perceived controllability. Personnel Psychology,47, 358-373.

Martocchio, J. J., & Webster, J. (1992). Effects of feedback and cognitiveplayfulness on performance in microcomputer software training. Per-sonnel Psychology, 45, 553-578.

Meichenbaum, D. H. (1977). Cognitive behavior modification: An integra-tive approach. New York: Plenum.

Michel, D. A. (1994). What is used during cognitive processing in infor-mation retrieval and library searching? Eleven sources of search infor-mation. Journal of the American Society for Information Science, 45,498-514.


Mossholder, K. W. (1980). Effects of externally mediated goal setting onintrinsic motivation: A laboratory experiment. Journal of Applied Psy-chology, 65, 202-210.

Payne, J. W., Bettman, J. R., & Johnson, E. J. (1993). The adaptivedecision maker. Cambridge, England: Cambridge University Press.

Quint, B. (1991, July). Inside a searcher's mind: The seven stages of anonline search—Pt. 2. Online, 28-34.

Robert, J. (1987). Learning a computer system by unassisted exploration.In J. Bullinger & B. Shackel (Eds.), Human-computer interaction—Interact '87 (pp. 651-656). New York: Elsevier Science.

Salterio, S. (1996). Decision support and information search in complexenvironment: Evidence from Archi data in auditing. Human Factors, 38,495-505.

Schunk, D. H. (1989). Self-efficacy and cognitive skill learning. In C.Ames & R. Ames (Eds.), Research on motivation in education: Vol. 3.Goals and cognition (pp. 13-44). San Diego, CA: Academic Press.

Simon, S. J., & Werner, J. M. (1996). Computer training through behaviormodeling, self-paced and instructional approaches. Journal of AppliedPsychology, 81, 648-659.

Wildemuth, B. M., de Bliek, R., Friedman, C. P., & File, D. D. (1995).Medical students' personal knowledge, searching proficiency, and data-base use in problem solving. Journal of the American Society forInformation Science, 46, 590-607.

Williams, S. L. (1990). Guided mastery treatment of agoraphobia: Beyondstimulus exposure. In M. Hersen, R. M. Eisler, & P. M. Miller (Eds.),Progress in behavior modification (Vol. 26, pp. 89-121). NewburyPark, CA: Sage.

Wood, R. E., Atkins, P. B., & Tabernero, C. (2000). Self-efficacy andstrategy on complex tasks. Applied Psychology: An International Re-view, 49, 430-446.

Wood, R. E., & Bandura, A. (1989). Impact of conceptions of ability onself-regulatory mechanisms and complex decision making. Journal ofPersonality and Social Psychology, 56, 407-415.

Wood, R. E., George-Falvy, J., & Debowski, S. (in press). Motivation andinformation search on complex tasks. In M. Erez, U. Klienbeck, & H.Thierry (Eds.), Work motivation in the context of a globalizing economy.Hillsdale, NJ: Erlbaum.

Wood, R. E., Kakebeeke, B. M., Debowski, S., & Frese, M. (2000). Theimpact of enactive exploration on intrinsic motivation, strategy andperformance in electronic search. Applied Psychology: An InternationalReview, 49, 263-283.

Received March 20, 2000Revision received January 20, 2001

Accepted January 21, 2001

Impact of Guided Exploration and Enactive Exploration on Self

Documents

Transcript of Impact of Guided Exploration and Enactive Exploration on Self