9 How Much and What Type of Guidance is Optimal for Learning ...

More documents

Recommendations

Info

How Much and What Type of Guidance is Optimal? 1650123456789012345678901234567890123456Guidance in Poorly Defined or Ill- Structured Domains of KnowledgeSome instructional researchers and developers argue that accurate and completeguidance is not possible for many tasks required in modern curricula that comefrom ill- structured domains of knowledge (Jonassen, 1997; Spiro, Feltovich,Jacobson, & Coulson, 1992). Poorly structured problems are those that “possessmultiple solutions, solution paths, fewer parameters which are less manipulable,and contain uncertainty about which concepts, rules, and principles are necessaryfor the solution or how they are organized and which solution is best”(Jonassen, 1997, p. 65). Examples include medical diagnosis, historical analysis,leadership or organizational management and counseling psychology. It is alsoargued that the research tasks used to demonstrate the benefits of proceduralinstruction, such as Anderson’s ACT- R theory, tends to be drawn from morestructured domains such as mathematics, and therefore may not generalize topoorly structured domains.Describing a domain as “ill structured” most often means that either domainexperts do not agree or that there are no solutions to some problems. Nearly allproblems contain “multiple solution paths,” many of which achieve an acceptableresolution to a problem. In this case, the best option is to teach the most direct andsimple solution path to novices. In general, when experts fail to consistently solvecomplex problems we can hardly expect students to discover solutions duringinstruction. In the case where students are expected to invent a solution, the preferableinstructional approach is to provide expert- based procedures for inventingsolutions to problems in the domain. In this case the focus of the instruction shiftsfrom students discovering solutions to students learning a procedure for discoveringsolutions. The important issue for those designing problem- solving instructionis whether there are experts in a knowledge domain who consistently succeed atsolving problems or performing tasks in that domain. Expert solutions to problemscan be captured using cognitive task analysis and taught to novices.Cognitive Task Analysis for Capturing ExpertiseResearch on the use of cognitive task analysis (CTA) to capture and identify theautomated knowledge used by experts has grown in recent years (Clark & Estes,1996; Schraagen, Chipman, & Shalin, 2000; Clark et al., 2007). As evidence of theinstructional value of using CTA to identify automated and unconscious expertknowledge, Lee (2004) performed a meta- analytic study of the effectiveness ofCTA- based training and performance- improvement studies in a variety of organizationsand focused on different types of tasks. She reported an overall medianpercentage of post- training performance gain effect size of 1.72 (an averageincrease of 44% on outcome performance measures) for CTA- based instructionwhen compared to more traditional instructional design using behavioral taskanalysis. Most of the outcome measures reviewed emphasized application oflearning rather than recall or recognition tasks.Velmahos et al. (2004) studied the expertise of emergency medicine expertswho teach in a medical school. In a controlled study, half of a randomly assigned
166 R. E. Clarkgroup of 24 medical students were taught a routine emergency procedure in atraditional modeling and practice strategy by expert emergency physicians whoteach. The established teaching strategy employed is called “see one – do one–teach one.” The student first watches a procedure performed by an expert whoexplains it in a “think aloud” fashion, and then practices the procedure whilegetting feedback from the same expert. Finally the student teaches anotherstudent to perform the procedure while the expert observes. While this instructionalmethod has served medicine for decades, recent concerns about medicalmistakes have refocused interest on the way medical students are trained andhave encouraged attempts to close the gaps identified in the way complex proceduresare taught (Starfield, 2000). The “ see- do-teach” students’ post- trainingperformance was compared with the other half of the medical students who weretrained with information gathered in a “cognitive task analysis” or CTA (Clark &Estes, 1996; Schraagen et al., 2000; Clark et al., 2007) on the same emergencyprocedure. The CTA interview is designed to expose automated decisions madeby experts and make them available for training. The CTA- trained students wererequired to use the procedures they saw demonstrated. The emergency medicineexperts who were interviewed with CTA also served as the instructors for thesee- do-teach condition. All students received both memory and performancetests. It was clear from the analysis that the information provided to thesee- do-teach students contained significant omissions and errors.After training, whenever the medical students performed the routines withpatients in the following year, they were observed and evaluated with checklistsby judges who were unfamiliar with the instructional method they had experienced.The experimental group who received training based on cognitive taskanalysis outperformed the expert- taught control group on all analytical (diagnostic)and many performance items by over 50% during the year followingtraining. Velmahos (personal communication) also reported that the traditionallytrained doctors caused three serious medical emergencies applying themedical protocol with patients (average for new physicians) and those with CTAtraining made no life- threatening mistakes.Research SuggestionsDisputes about variations in guidance can best be determined by reference toevidence from “randomized, controlled tests of competing instructional procedures[where] altering one [relevant] variable at a time is an essential feature of aproperly controlled experiment” (Sweller, Kirschner, & Clark, 2007, p. 115). Inaddition, the hypothesized operation of the strategies selected for examinationshould be drawn from an evidence- based view of human cognitive architecture.A balanced review of existing studies will indicate that most of the disputes aboutguidance may stem from different strategies for designing guidance experiments.Design issues appear to be the root of the disagreements with Kirschner, Sweller,and Clark’s (2006) argument about the “failure” of constructivism, discovery andproblem- based learning. Rejoinders to this review (for example, Hmelo- Silver etal., 2007; Schmidt et al., 2007) pointed to evidence from studies where lower or1234567891111111111222222222233333333334444444
Page 1 and 2: 9 How Much and What Type ofGuidance
Page 3 and 4: 160 R. E. Clarkadvanced solutions t
Page 5 and 6: 162 R. E. Clark1 Guidance Must Prov
Page 7: 164 R. E. ClarkExperts Are Largely
Page 11 and 12: 168 R. E. Clarkdesigners and teache
Page 13 and 14: 170 R. E. Clarkare asked to determi
Page 15 and 16: 172 R. E. Clarkcontext, certainly a
Page 17 and 18: 174 R. E. ClarkHer results suggest
Page 19 and 20: 176 R. E. Clarkpossible in emergent
Page 21 and 22: 178 R. E. ClarkAssembling procedura
Page 23 and 24: 180 R. E. ClarkBrown, J. S., Collin
Page 25 and 26: 182 R. E. Clarkintelligence in educ

9 How Much and What Type of Guidance is Optimal for Learning ...

Create successful ePaper yourself

Delete template?

Save as template?