Problem-based versus lecture-based medical education


Material Information

Problem-based versus lecture-based medical education a meta-analysis of cognitive and noncognitive outcomes
Physical Description:
viii, 281 leaves : ill. ; 29 cm.
Smith, Robert Allan
Publication Date:


Subjects / Keywords:
Teaching and Learning thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Teaching and Learning -- UF   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 2003.
Includes bibliographical references.
Statement of Responsibility:
by Robert Allan Smith.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 030315472
oclc - 52697327
System ID:

This item is only available as the following downloads:

Full Text






This dissertation is dedicated to the memory of Dr. John J. Koran who exhibited
indefatigable enthusiasm for mentoring doctoral student research and was chair of my
supervisory committee until his death.


I am very grateful to my supervisory committee members for providing guidance

and valuable time during my five years of graduate school. Dr. Mary Grace Kantowski,

committee chair, was my exoskeleton throughout the doctoral research and dissertation

construction. I would have imploded without her unwavering support. Dr. Mary Lou

Koran mentored my first educational research proposal and introduced me to differential

psychology. Dr. Ben Nelms inspired in me a deep appreciation for student-centered

learning and a quest to discover appropriate measures of academic achievement. Dr.

Anne Seraphine is a uniquely superior teacher and mentor. She taught me how to get

"intimate with data."


A C K N O W LED G M EN T S ................................................ ........................................... iii

A B ST R A C T ........................... ................................................................................ vi


1 INTRODUCTION ................................................. I

B background and C ontext.......................................................................................... ...
Statem ent of Problem ....................................................................... ..................... 16
Research Questions ...................................... ..................................1 8
D definition of T erm s............................................................................ ................... 19
C ontext o f S tudy .................................................. ........................... .................... 25
Significance of Results .................... ... ....................... 26
L im stations of Study ................................................. ............................................ 28

2 REV IEW O F THE LITERA TUR E ..................................... .............. .................... 30

Theoretical and Scientific Foundations of Problem-Based Medical Education .........30
Features of Problem-Based Learning.............. ..............................................46
Implications Leading to the Present Study ............................... 110

3 METHODOLOGY .............. ....................................... 112

R ationale for a M eta-A analysis .................... .....................................................1... 12
H yp o th eses ...................................................................................... .................... 1 14
Candidate Literature Criteria ........................... .....................................................115
M ethodological D esign ................................................................... .................... 116
Conceptual Summary of Methodology................................................................130
Critique of M ethodology........................................................... .................... 132

4 RESULTS AND ANALYSES............................................................................... 135

D distribution of Effect Sizes............................. ..................................................... 136
Main Effects of Problem-Based Learning ................................ ........................137
Biomedical Science Achievement ................................................................139
C clinical Science A chievem ent ......................................... ............. .................... 161


Problem Solving Achievem ent .................... ....................................................... 175
Self-D directed Learning Skills .......................... ................................................... 183
Attitude Toward Learning................................. .................................................... 196

5 SUMMARY AND CONCLUSIONS ..................................................................212

Questions Answered by this Research.................................................................. 212
A C ritique of the R esearch.............................................................. ..................... 242
Conclusions and Implications for Practice, Research and Theory .........................243



B STUDIES EXCLUDED FROM ANALYSES...... ....................256

LIST O F REFER EN CES................................................................. ..................... 259

BIOGRAPHICAL SKETCH .......................... ........................................................ 281

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


Robert Allan Smith

May 2003

Chairperson: Dr. Mary Grace Kantowski
Major Department: Teaching and Learning

Inconsistent findings comparing the relative efficacy of problem-based (PBL)

versus lecture-based (LBL) learning in medical education motivated this study. In this

meta-analysis of 82 studies involving 45 universities in twelve countries, with 12,979

participants and 121 findings, PBL was associated with a significant positive effect size

for clinical science achievement, problem-solving achievement, self-directed learning

skills, and attitude toward learning. Instructional effects were statistically equivalent for

biomedical science achievement. However, variance analysis revealed a subgroup

exhibiting a small, but significant positive effect size when biomedical science

achievement was compared using criterion-referenced rather than standardized

assessments. t ,,,r. suggest that when achievement and problem solving ability are

examined with assessments contextually framed and requiring the integration and

application of biomedical and clinical science knowledge, medical students and

physicians who undergo PBL significantly outperform counterparts who undergo LBL.

Moreover, given the dominant factual and abstract nature of the content, the lack of

unique superiority of LBL for biomedical science achievement is an important practical

finding. Indeed, differential instruction efficacy for clinical science and problem solving

achievement imply biomedical science knowledge constructed in a PBL environment is

organized for better deployment.

Current findings suggest the success of PBL is not limited to a particular

approach. The advocates of PBL make some substantive claims, using ill-structured

problems, using re-iterative methods, and so on, yet none of these claims were found

uniquely to predict the instructional effect on cognitive outcomes. Instances of

multicolinearity, however, suggest some substantive features may converge efficaciously

as a function of context and desired cognitive outcome. Despite diverse methods of PBL

implementation, problem posing and integration of biomedical and clinical sciences in a

tutorial-group setting are common threads. Since direct instruction also can integrate

biomedical and clinical sciences, further research is required to examine the unique

contribution of small-group learning. Since the ultimate aim of medical education is to

graduate competent clinicians who are expert learners and problem solvers, examining

what distinguishes the structure and function of medical knowledge acquired when

undergoing PBL versus LBL may be more insightful for informing theory and practice

than global contrasts of cognitive and non-cognitive outcomes.


Problem-based learning was introduced as a pedagogical alternative to traditional

lecture-based medical education. Advocates claim problem-based learning

operationalizes principles of modem learning theory and better prepares medical students

and physicians for clinical practice. However, research comparing the relative efficacy of

lecture-based versus problem-based learning has produced inconclusive findings. Present

study used meta-analytic methods to synthesize results of primary research comparing

cognitive and non-cognitive outcomes in medical students and physicians who undergo

lecture-based versus problem-based teaching and learning. Principal aim of current

research was to examine the main effects of instructional approach on five broad

constructs including biomedical science achievement, clinical science achievement,

problem-solving achievement, self-directed learning skills, and attitude toward learning.

A secondary purpose was to identify those outcome measures, methodological and

substantive features that might be responsible for significant variance in instructional

effect. Ultimately, the strengths and weaknesses of each instructional method were

examined to inform practice and theory and serve as a point of departure for future


Background and Context

Abraham Flexner (1910, 1925), the architect of modern discipline-bound, lecture-

based medical education, insisted that understanding normal structure and function were

essential to learning about abnormal structure and function (1910). "The perception of

abnormality or irregularity can only take place against the background of normality or

regularity" (p. 106, 1925). Moreover, Flexner rigorously opposed any clinical training

during the first two years of medical school.

The argument in behalf of mingling clinical and pre-clinical subjects in order to
interest the student is assuredly not convincing. A clinical illustration may
indeed clarify the student's comprehension of a fundamental principle or fact; but,
at least in excess, it is more likely to excite a superficial interest in symptom or
remedy. (p. 112)

Consequently, a hallmark of traditional medical school teaching and learning is

the separation between biomedical science and clinical science education. Preclinical

education is dominated by discipline-oriented, lecture-based instruction with little

intentional instruction in the application of biomedical science to clinical problems (Bok,

1984; Ebert, 1992; Sch6n, 1987; Tosteson, 1994; Swanson & Case, 1997). Faculty

members in the biomedical sciences, although primarily non-physicians, not only specify

what content and skills students must master, they also identify the resources over which

students are to achieve mastery (Distlehorst & Robbs, 1998). Mastery of content is

assessed primarily using examinations incorporating multiple-choice questions (Albanese

& Mitchell, 1993; Distlehorst & Robbs; Vernon & Blake, 1993), and the principal

learning strategy employed by successful students is rote memorization (Lindblom-

Ylinne & Lonka, 1999; Regan-Smith et al., 1994).

During clinical science training, an apprenticeship model guides clinical

education as students undergo clerkships in core medical specialties including internal

medicine, surgery, obstetrics and gynecology, pediatrics, and psychiatry. Elective

clerkships are available to senior medical students. Physicians provide teaching during

clerkships and students are required to learn in less structured environments approaching

those of clinical practice.

The belief that traditional pre-clinical education over-emphasized rote

memorization and failed to prepare medical students adequately for clinical experience

prompted the introduction of problem-based teaching and learning (Barrows & Tamblyn,

1980; Neufeld & Barrows, 1974). This belief was echoed when a panel on General

Professional Education of the Physician published a report entitled Physicians for the

Twenty-first Century (1984) which concluded that "students are led to think that their

education depends upon memorizing as much information as possible" (Panel, p. 1). This

deduction was substantiated by a survey of 453 senior students undergoing traditional

medical education at five universities including the Universities of Florida and New

Mexico, Dartmouth, Wake Forest University (Bowman Gray) and Rush University

(Regan-Smith et al., 1994). Forty-nine percent of the students believed they learned half

or more of the biomedical sciences content by memorizing without understanding.

More importantly, research suggests that much memorized biomedical science

content is irretrievable, or determined to be irrelevant (Miller, 1961; Neame, 1984; Rico,

Galindo, & Marset, 1981). Miller observed poor retention of biomedical science content

in medical students who underwent lecture-based learning and warned that information

presented via teacher-centered methods would "decrease from memory at the same rate

as nonsense syllables" (p. 153). Rico and colleagues observed that students forgot 25%

of the most important concepts one year after completing lecture-based medical

chemistry; 40% after three to five years, and 50% after eight years. In addition,

memorizing information with little or no practical value other than passing examinations

can be boring and possibly even demoralizing for students (Des Marchais & Vu, 1996;

Schmidt, 1983; Sch6n, 1987). Lindblom-Ylinne and Lonka (1999) observed that the

learning environment of a traditional medical school can give misleading cues about how

to study efficiently and obliged many students to study in inadequate ways, driving them

towards externally regulated and superficial learning. Consequently, many students were

rendered incapable of meaningful learning or creative thinking.

Furthermore, proponents of problem-based learning suggest converting from a

highly structured learning environment characterized by rote memorization (i.e., pre-

clinical education) to a more self-directed and independent learning situation (i.e., clinical

clerkships) may be problematic for students (Sch6n, 1983). Indeed, Patel, Evans and

Groen (1989) describe biomedical and clinical science knowledge as "two worlds apart"

(p. 55). Although not mentioned in their report, the two-worlds metaphor employed by

Patel et al. echoes knowledge distinctions expressed by Karl Popper (1972) and is

discussed later.

Magnani (1997) posits that medical students educated in a traditional manner will

encounter difficulty when they convert from biomedical science (unsituated content,

concerned with the attributes of entities such as organs, pharmaceuticals and microbes) to

clinical science (situated material, concerned with attributes and health problems of

people). This view aligns with perspectives expressed by some situational learning

theorists (Brown, Collins, & Duguid, 1989; Greeno, 1998; Lave & Wenger, 1991) and

will be discussed later. In addition, problem-based learning advocates argue students

undergoing lecture-based learning are inadequately prepared when encountering

problems requiring the transfer of their learning to new domains, a skill required to

function effectively in clinical practice (Barrows, 1986, 1990, 1996, 2000; Boud &

Feletti, 1997). This is a consequence Dewey anticipated (1933/1998) when he suggested

a distinction between educating memories versus educating minds:

It is not enough to load the memory with statement of facts and laws and then
hope that later in life by some magic the mind will find a use for them. Even
general principles, when merely memorized, stand on the same level as bare
particular facts. Since they are not used either in understanding actual objects or
events or in giving rise, through what -I.. inpil.. to other conceptual meanings,
they are, to the mind that memorizes them (falsely called learning), mere arbitrary
items of information. (pp. 185-185)

McKegney (1989) rhetorically concurred with these investigators when he wrote

"The experience of studying in medical school has been considered to be analogous to

belonging to an abusive and dysfunctional family" (p. 452). And Barrows (2000) posits,

"This is not only educational malpractice, it is tragically inefficient when you consider

how much energy faculty put into teaching and students put into studying during these

preclinical years to result in such a small yield" (p. xiii). This cognitive predicament was

anticipated by Whitehead (1929) who argued that "inert knowledge" (p. 1) is not just

useless; "it is harmful Corruptio optimi, pessima" (p. 2) (that is, corruption of the best

is worst of all).

Notwithstanding philosophical, anecdotal and evidence-based admonishments by

opponents of traditional medical education, 71% (84/118) of medical schools recognized

by the Association of American Medical Colleges still employ lecture-based, discipline-

oriented preclinical curricula I ,p.,.. Swanson, & Case, 1998). By implication, lecture-

based teaching and learning are considered by the majority of medical schools to be the

appropriate pedagogical method to facilitate physician training.

Arguments in Favor of Traditional Lecture-Based Learning

Reflecting Flexner's (1910, 1925) model, the initial two-years of traditional

medical school curricula in North America incorporates a knowledge reproduction model

in which disciplines in the biomedical sciences are the primary units for the organization

and delivery of content. This content is presented using lecture-based instruction, or

colloquially speaking, students I.. j-,I=-. by being told" (Custers & Boshuizen, 2002, p.

170). In effect, the biomedical science disciplines serve as a context for structuring

learning. From this perspective medical knowledge is viewed as a body of established

facts and techniques that are organized ..- ...... II and context free, thus able to be

partitioned and transmitted by experts to novices (Flexner; Hemker, 1998; Papa &

Harasym, 1999; Sch6n, 1983, 1987). Learning is viewed as the successive accumulation

of these facts and techniques primarily by listening, observing, memorizing and

practicing (Schon).

Widespread endorsement of this epistemological stance is evidenced by the

magnitude ofpreclinical education delivered employing a knowledge transmission

paradigm (Sch6n, 1983) that is predominantly lecture-based (Ripkey et al., 1998).

Traditional lecture model assumes superior knowledge of instructor and consequently

situates in their hands responsibility for selecting content to be presented, depth of

content to be covered, degree of interaction between content and practical instances, and

duration of specific content coverage (Dubin & Taveggia, 1968). Traditional biomedical

science lectures are organized in discrete disciplines and sustaining a Flexnerian

philosophy, are designed not to complicate theory with practice. This instructional

approach has the advantage of preserving the fundamentals of each discipline.

Instruction tends to be well structured and linear, leading to expected student behavior

and response (Papa & Harasym, 1999). Lectures can help simplify information,

providing an entry point for complex content (Rosenshine & Stevens, 1986). That is,

lectures facilitate the gradual and logical development of complex or difficult concepts

and theories (Hemker, 1998). Hemker argues:

Having the subject explained by an experienced scientist or clinician is
particularly useful in acquiring the attitude and skills necessary to organize
knowledge, and learning to distinguish between essential and accessory
knowledge. (p. 75)

In this way, the aim is to have students connect the new material to previous material and

to practice until they master the content. Once content mastery occurs, the students can

proceed to experiential learning activities incorporating inquiry and complex problem

solving. Explicit instruction before experiential activities reduces chance for

development of misconceptions during initial learning (Rosenshine & Stevens).

Moreover, instructors can model professional practice by using disciplinary

exemplars and they can convey the intrinsic value of the content through their enthusiasm

(Rosenshine & Stevens, 1986). According to Flexner (1925), a good lecturer is a

"textbook plus a personality" (p. 181). Also supporting this stance, Hirsch (1997)

believes that challenging subject content "can be taught in a lively, demanding way" (p.

43) ir..,L.h lectures. Lectures present minimal risk to students and appeals to those who

prefer to learn by listening or those whose own learning strategies are ineffective

(Hirsch). Accordingly, students develop a comfort level and need only to depend on a

syllabus and instructor for guidance (Barrows, 1996, 2000). Furthermore, instructors

themselves feel more comfortable and perform better in a lecture-based system, and it is

cost effective and less demanding as well (Vernon & Blake, 1993).

Arguments in Favor of Problem-Based Learning

Interestingly, seven years before Flexner's report to the Carnegie Foundation,

William Osler (1903) argued for a more practice-based medical education when he wrote

In what may be called the natural method of teaching the student begins with the
patient, continues with the patient, and ends with his studies with the patient,
using books, and lectures as tools, as a means to an end. take him from the
lecture-room, take him from the amphitheater, put him in the out-patient
department put him in the wards. the best teaching is that taught by the
patient himself. (p. 49-50)

Apparently, for Osler, no medical teaching and learning should be done without a patient

context. R.: ii..ii .- epistemology advanced by Osler, proponents of problem-based

teaching and learning claim that to understand the principles and practices of medicine,

knowledge construction grounded in the integration of biomedical science and clinical

science embedded in the problems reflecting clinical practice is prerequisite (Barrows,

1986, 1990, 1996, 2000; Barrows & Tamblyn, 1980; Boud & Feletti, 1997; Neufeld &

Barrows, 1974). Consequently, problem-based learning experiences are designed to

integrate biomedical and clinical sciences into the context of authentic clinical practice.

Learning experiences are intended to give students an opportunity as well as a

responsibility to establish what they need to know and master the process of inquiry

necessary for resolving health care problems. These learning experiences are tailored

toward achieving educational expectations and objectives (Barrows; Schdn, 1983, 1987).

By design, problem-based instruction is student-centered with tutorial groups that

emphasize collaborative learning (Barrows; Boud & Feletti; Stage, Muller, Kinzie, &

Simmons, 1998). Although an instructor (tutor) does have considerable responsibility in

facilitating inquiry, students are expected to gradually assume responsibility for their own

learning. With appropriate experience and guided practice, it is postulated that students

eventually will gain full independence with the instructor gradually becoming a co-

worker. The emphasis is on active student acquisition of information and skills suitable

to their ability, level of experience, and educational needs. Ideally, students determine

the best methods for learning, find the necessary resources, set the pace, and structure the

acquisition of information within an activity designed to resolve a health care problem

(Barrows). In this manner, problem-based learning focuses on diversity rather than

uniformity. It fosters the acquisition of a rich, integrated and organized medical

knowledge base that is linked to authentic health care problems and encourages

individual and collaborative processes of reasoning (inquiry). This pedagogical approach

appears to be suitable for preparing students for lifelong learning (Barrows; Savery &

Duffy, 1995; Zimmerman & Lebeau, 2000). Moreover, problem-based teaching and

learning operationalizes adult learning theory (Knowles, 1984), the principles of

knowledge activation and encoding specificity (Anderson, 1993, 1996; Ausubel 1960;

Chi, DeLeeuw, Chiu, & LaVancher, 1994; Newell & Simon, 1972), deep-processing

(Bloom, 1984; Craik & Lockhart, 1972; Newble & Clark, 1986), social constructivism

(Cobb, 2002; Prawat, 1993; Savery & Duffy; Vygotsky, 1926/1997), and situated

cognition (Brown et al., 1989; Greeno, 1998; Lave & Wenger, 1991; Schmidt, 1983,

1993; Schmidt et al., 1996; Schmidt, Norman, & Boshuizen, 1990).

Synthesis of Arguments

Using both problem-based and lecture-based teaching and learning methods,

instructors may prepare what they judge to be appropriate learning objectives, learning

resources, choice of pathways, and evaluation materials that reflect their particular

experience and knowledge. Traditional lecture-based instruction begins by exposing

students to disciplinary knowledge, and then assigning problems to promote the use and

mastery of that knowledge. Traditional lecture-based instruction typically employs only

one facet of direct instruction (Rosenshine & Stevens, 1986), that is, didactics begin by

transmitting disciplinary knowledge to students, and then assigning problems to promote

the use and mastery of that knowledge. Each biomedical science discipline (e.g.,

anatomy, biochemistry, pharmacology, and physiology) has a separate block of curricular

time. Dewey (1933/1998) described such isolation of facts from meaning when he wrote,

Their minds are loaded with disconnected piecemeal items that are accepted on
hearsay or authority. and no attempt is made to interpret it in relation to what
it does, how it was caused, and what it stands for. (p. 184)

For Dewey, the rationale for this epistemic stance is that

It assumes that logical quality belongs only to organized knowledge and that the
operations of the mind become logical only through absorption of logically
formulated, ready-made material. In this case, the logical formulations are not the
outcome of any process of thinking that is personally undertaken and carried out;
the formation has been made by another mind and is presented in finished form,
apart from the process by which it was arrived at. Then it is assumed that by
some magic its logical character will be transferred into the minds of pupils. (p.

Students in such a learning environment, were characterized as "spectators" by Dewey

(1916, p. 146). Moreover, mastery learning is based on the assumption that once medical

students exhibit competency, they will remain competent. However, retaining knowledge

acquired during medical school for a lifetime is improbable (Bok, 1984), if not

undesirable given that the half-life of medical education now is estimated to be five-years

(Gorman, Meier, Rawn, & Krummel, 2000). Clearly, the knowledge reproduction model

is inadequate. Thus, information management, rather than information mastery, appears

to be a more appropriate aim of preclinical medical education. Furthermore, knowledge

reproduction models devalue content learned and foster a situation in which learners are

expected to encode knowledge implicitly grounded in theory when the deployment and

value of that knowledge are explicitly attributed to practice (Barrows, 2000; Cobb, 2002;

Perkins & Salomon, 1989; Savery & Duffy, 1995; Sch6n, 1983, 1987).

Because clinical science problems rarely involve isolated disciplinary knowledge,

ex post facto the medical student has to synthesize, integrate, and eventually apply to

practice these isolated blocks of knowledge. Thus, knowledge and skills acquisition

occurs in an environment dominated by instruction, not construction, in a process I

characterize as "retrospective applicability." Consequently, for traditional medical

students, two types of knowledge transfer must occur (Patel, Groen, & Scott, 1988; Patel

et al., 1989; Patel, Groen, & Norman, 1991; Schmidt, DeGrave, de Volder, Moust &

Patel, 1989). Initially, biomedical science knowledge must be transferred to clinical

science knowledge (i.e., the integration of biomedical and clinical science knowledge).

Then, this integrated biomedical and clinical sciences knowledge must be transferred to

representing and solving the health care problems of individual patients.

Rather than assume a mythical transfer (Dewey, 1933/1998) as students convert

from a biomedical science world to a clinical science world (Patel et al., 1989; Popper,

1972; Sch6n, 1983), problem-based teaching and learning assume that there is no

distinction between theory and practice. Accordingly, there is a reciprocal relationship

between biomedical and clinical sciences with one informing the other, all framed within

the context of health care problems (Barrows, 1983, 1986, 1990, 1996, 2000; Barrows &

Tamblyn, 1980; Boud & Feletti, 1997; Neufeld & Barrows, 1974).

Health care problems stimulate a process of inquiry and knowledge construction

in which biomedical knowledge feeds into clinical knowledge (Barrows, 2000; Cobb,

2002; van de Wiel, Boshuizen, & Schmidt, 2000). Problems are designed to provide

students with knowledge that has clinical relevance and therapeutic significance (the

purpose of knowledge) while encouraging them to assume responsibility for their

learning (personally constructed meaning), that is, knowledge as design (Perkins, 1986).

Viewed from this framework, knowledge as information is passive, while knowledge as

design becomes active. Perkins suggests "Knowledge disconnected from context is

nothing short of truth mongering ." (p. 1). Purpose and meaning making should act as

psychological tools (Vygotsky, 1926/1997) for understanding concepts. For Perkins, the

aim of professional education is to teach knowledge as a design for sense making, a

learning process I characterize as "concurrent applicability." Accordingly, in acquiring

knowledge learners need to invoke a cognitive auditing method to ascertain: (a) Is this

fact important?, (b) Is it relevant?, (c) Does it have personal meaning?, and (d) Is

knowledge linked to a purpose? (Perkins, p. 4). From this perspective, "purpose" is the

principal tool to reveal evidential value for triaging and understanding information. In

other words, information calibrated with pragmatism guides meaningful knowledge

construction. Indeed, "when a piece of information is connected to a purpose, it becomes

design-like" (Perkins, p. 4). Extending Deweyan (1933/1998) philosophy, problem-

based teaching and learning is concerned with educating the "mind" versus educating the

"memory." Indeed for Dewey, "learning is learning to think" (p. 78).

Legitimacy of Problem-Based Learning Is Hampered by Equivocal Evidence

Some educators argue that the focus of problem-based instruction on cognitive

processes and student-directed leading results in the acquisition of biomedical science

knowledge they label as "basic science lite" (Swanson & Case, 1997, p. 73). Morgan

(1995) notes that when graduates of the problem-based learning program at McMaster

University go to other medical schools for their clinical training, they have less

biomedical science knowledge initially than other Canadian graduates, although "they

soon catch up" (p.1643). Morgan suggests ". the General Medical Council [the

governing board for medical education in the United Kingdom], with its enthusiasm for

change, would do well to remember Burke's dictum, 'To innovate is not to reform'" (p.


Research comparing knowledge acquisition, problem-solving skills, and self-

directed learning in medical students and physicians who have undergone problem-based

versus lecture-based instruction is contradictory. Three literature reviews (Albanese &

Mitchell, 1993; Berkson, 1993; Colliver, 1999) and one meta-analysis (Vernon & Blake,

1993) comparing the curricular effects of problem-based learning (PBL) versus lecture-

based learning (LBL) have been published. Three of these four reports emphasize

different instructional characteristics and include non-physician training programs

(Albanese & Mitchell; Berkson; Vernon & Blake). Albanese and Mitchell classify and

list quantitative results (unstandardized and unweighted effect sizes) of research

conducted from 1972 to 1992 revealing that PBL students had lower biomedical science

achievement. They conclude that PBL students exhibit an "incomplete understanding of

the basic sciences" (p. 77), but PBL students found learning more "nurturing and

enjoyable" (p. 78). The authors u., ii1 "caution be exercised in making

comprehensive, curriculum-wide conversions to PBL" (p. 52) until further research

establishes guidelines regarding the degree of faculty direction, the necessary cost

effectiveness of implementing PBL, and establishes reasons for biomedical sciences

content weakness in PBL students.

Vernon and Blake (1993), who employed meta-analytic methods to synthesize the

results of comparative instructional research from 1970 to 1992, reported similar

findings. Quantified with effect size (ES), they found PBL medical students had inferior

biomedical science achievement (ES = -0.18), slightly better clinical science achievement

(ES = 0.08) as measured by national licensing examinations, better supervisor rated

clinical performance (ES = 0.28), and expressed greater self-rated satisfaction with their

learning environment (ES = 0.55).

Berkson (1993) employed a traditional, narrative method to review literature from

1983 to 1991 and concluded "the graduate of PBL is not distinguishable from his or her

traditional counterpart. In addition, the experience of PBL can be stressful for students

and faculty. And implementation of PBL may be unrealistically costly" (p. S85).

Berkson concluded PBL curricula were unlikely to surpass traditional curricula in

imparting problem-solving expertise, acquisition of knowledge, enhancing motivation,

improving self-directed learning, or providing experiences that are more enjoyable.

Colliver (1999), who conducted a narrative review of studies involving only

medical schools with curriculum-wide implementation of PBL from 1992 to 1998, found

eight studies and concluded there "is no convincing evidence that PBL improves

knowledge base or clinical performance" (p. 259).

Others present evidence indicating that PBL compared to LBL students acquire

less content knowledge and exhibit a tendency to develop misconceptions and errors

while applying biomedical science knowledge to clinical problems (Bernstein, Tipping,

Bercovitz, & Skinner, 1995). Moreover, Ripkey and associates (1998) reviewed

interschool differences in United States Medical Licensing Examination (USMLE) Step 1

scores and concluded that curriculum type contributed little (approximately 6% of the

variance) to observed variations in medical students' acquisition of biomedical science

knowledge. Unfortunately, the curricular effects of schools incorporating nontraditional

programs (including problem-based instruction) could not be distinguished because the

sample size was too small.

Since student performance on licensing examinations is considered a benchmark

of curriculum, the degree to which knowledge and problem-solving skills are acquired in

a PBL curriculum has manifested considerable concern and debate (Albanese, 2000;

Colliver, 2000; Norman & Schmidt, 2000). Students in the United States take two of

these examinations (USMLE Step 1 and Step 2) during medical school and, depending on

their school, may be required to pass Step 1 to advance to year three of medical school

and Step 2 to graduate. In addition, the results of USMLE Step 1 and Step 2

examinations significantly affect selection for graduate medical education (e.g.,

residency). Step 3 must be passed before license to practice is granted. Given the high

stakes of these examinations, efforts to improve curriculum often engender concern that

learning instead might inadvertently be impaired.

In summary, only one meta-analytic investigation has compared cognitive and

noncognitive learning outcomes involving health care professionals (Vernon & Blake,

1993). Results of that meta-analysis regarding cognitive outcomes were mixed and

incorporated data from licensing examinations that have changed significantly

(O'Donnell, Obenshain, & Erdmann, 1993; Williams, 1993) over the years. In 1992,

USMLE replaced the National Board of Medical Examiners (NBME) and shifted the

emphasis from measuring factual and abstract biomedical scientific knowledge to

assessing the application of biomedical knowledge (O'Donnell et al.). Moreover,

previous meta-analysis did not statistically distil or analyze effect size variance beyond

its main effects. I h-,,r, r.. there has been no systematic accounting of variance

associated with outcome measures, methodologic features, or substantive features of

problem-based teaching and learning in medical education.

Statement of Problem

Despite cognitive and instructional theory predicting the unique superiority of

problem-based teaching and learning, the research literature is conflicting with regard to

the relative efficacy of problem-based versus lecture-based instruction. In addition,

efficacy is exacerbated by the variety of methods used to implement problem-based

instruction (Barrows, 2000; Blumberg, Michael, & Zeitz, 1990). Consequently, some

medical educators have recommended caution when considering conversion to problem-

based learning (Albanese & Mitchell, 1993; Berkson, 1993; Colliver, 2000; D'Eon, 1999;

Morgan, 1995). Risks cited include inferior biomedical science achievement exhibited

by PBL versus LBL students as evidenced by results of licensing examinations. In

addition, some evidence suggests that compared to their traditional counterparts, PBL

students exhibit greater academic stress and perceive that they acquire less content

knowledge. Cited benefits include better self-directed learning skills and more favorable

attitudes toward learning, although these observations are not uniform across studies.

Consequently, at issue is whether student-centered, problem-based instruction can

facilitate efficient and effective biomedical and clinical sciences knowledge acquisition,

enhance the development of problem solving skills as well as self-directed learning skills,

and stimulate a positive attitude toward learning more than lecture-based pedagogical

methods. Since many medical schools are currently contemplating significant curricular

revision, it is important to examine the efficacy of problem-based instruction (D'Eon,

1999; Harden, Grant, Buckley & Hart, 2000; Kaufinan, 2000; Prideaux, 2000; Rothman,


Medical education research often fails to render clear, generalizable results, due in

part to insufficient sample size and idiosyncratic methodology (Harden et al., 2000). In

addition, important differential effects are not apparent in individual studies (Dubin &

Taveggia, 1968; Lipsey & Wilson, 1993). Research literature involving problem-based

learning is no exception (Aaron et al., 1998; Albanese & Mitchell, 1993; Berkson, 1993;

Colliver, 1999; Vernon & Blake, 1993). Indeed, for Aaron and associates:

Despite a substantial amount of research in a variety of centers of medical
education, it has been very difficult to provide objective evidence that a medical
education system which uses PBL generates graduates who have significant
differences in knowledge, skills, or behavior from those emerging from a
traditional didactic system. (p. 86)

Inconsistencies in primary research findings motivate this study. The research

aim is to quantify the main effects of problem-based versus lecture-based instruction with

regard to biomedical science achievement, clinical science achievement, problem-solving

skills, self-directed learning skills, and positive learning attitudes. A secondary research

objective is to examine the utility of theory guiding problem-based teaching and learning

by analyzing variations in effect size estimates as a function of outcome measures,

research methods, and pedagogical methods. The results should assist us in deciding the

efficacy of problem-based instruction and if theories of cognition can guide teaching and

learning to improve both medical student achievement and the cognitive and

noncognitive processes affecting achievement.

Research Questions

This investigation will analytically compare cognitive and non-cognitive

educational outcomes (dependent variables quantified as standardized weighted effect

size estimates) in medical students and physicians who undergo problem-based versus

lecture-based teaching and learning. The investigation will examine the main effects of

instructional approach on five broad constructs including biomedical science

achievement, clinical science achievement, problem-solving achievement, self-directed

learning skills, and attitude toward learning. Theoretically grounded conditional effects

will be addressed to identify those outcome measures, methodological and substantive

features that may be responsible for significant variance heterogeneity. Outcome features

are concerned with the nature of dependent measures analyzed; methodological features

are concerned with the quality of research design; substantive features are characteristics

of instructional intervention, the fidelity of problem-based instruction, and environment

(e.g., scope of PBL, PBL taxonomy, problem-structure, tutor characteristics, origin of

learning issues, and responsibility for self-directed learning).

The research is guided by the following questions:

1. Is there a positive effect size for biomedical science achievement in students and
physicians who undergo medical education with problem-based compared to
lecture-based teaching and learning?

2. Is variation in biomedical science achievement effect size estimates greater than
sampling error warranting subgroup analyses to test for potential moderation by

a priori defined outcome measures, methodological features, and substantive
features of problem-based teaching and learning?

3. Is there a positive effect size for clinical science knowledge achievement in
students and physicians who undergo medical education with problem-based
compared to lecture-based teaching and learning?

4. Is variation in clinical science achievement effect size estimates greater than
sampling error warranting subgroup analyses to test for potential moderation by a
priori defined outcome measures, methodological features, and substantive
features of problem-based teaching and learning?

5. Is there a positive effect size for problem solving achievement in students and
physicians who undergo medical education with problem-based compared to
lecture-based teaching and ,.. .,,ii, ,. '

6. Is variation in problem-solving achievement effect size estimates greater than
sampling error warranting subgroup analyses to test for potential moderation by a
priori defined outcome measures, methodological features, and substantive
features of problem-based teaching and learning?

7. Is there a positive effect size for self-directed learning in students and physicians
who undergo medical education with problem-based compared to lecture-based
teaching and learning?

8. Is variation in self-directed learning effect size estimates greater than sampling
error warranting subgroup analyses to test for potential moderation by a priori
defined outcome measures, methodological features, and substantive features of
problem-based teaching and learning?

9. Is there a positive effect size for attitude toward learning in students and
physicians who undergo medical education with problem-based compared to
lecture-based teaching and learning?

10 Is variation in attitude toward learning effect size estimates greater than sampling
error warranting subgroup analyses to test for potential moderation by a priori
defined outcome measures, methodological features, and substantive features of
problem-based teaching and learning?

Definition of Terms

Similar instructional methods and outcomes are operationalized differently in the

research literature synthesized in this meta-analysis. Recognizing inherent operational

variability, the following definitions are used throughout this investigation.

Lecture-based learning (LBL). Is defined as a discipline-oriented, teacher-

centered, expository transmission-of-content model. Biomedical science disciplines are

the primary units for organization and content is delivered primarily through lectures,

which may be supplemented with other forms of direct instruction (e.g., demonstration

laboratories, faculty guided skills practice, expert guided seminars) (Rosenshine &

Stevens, 1986). Typically, disciplines include anatomy, behavioral science,

biochemistry, immunology, microbiology, pathology, pharmacology, and physiology.

An assumption of this pedagogical method is that learning results by the accumulation of

information transmitted by the instructor (expert) to the student (novice). LBL is a

knowledge reproduction model with the instructor's expected learning outcomes

expressed in behavioral objectives. Biomedical science instructors not only specify what

content and skills must be mastered, they also identify the resources over which students

are to achieve mastery. Although there may be some performance-based assessment

employing human and computer simulations (e.g., Objective Structured Clinical

Examinations), content mastery is the aim and is assessed primarily by examinations

dominated by multiple-choice questions. Lecture-based instruction may be implemented

as part of a course, an entire course, or an entire curriculum. Curriculum-wide

implementation involves preclinical content (biomedical sciences). This occurs within

the first two years of medical school in the United States and Canada and the first three to

four years in Europe and Australia.

Problem-based learning (PBL). Is defined as a pedagogical method that

assumes learning results from the process of working toward the understanding or

resolution of a health care problem. PBL may be implemented as part of a course, an

entire course, or an entire curriculum. Although preclinical core content and temporal

constraints are similar to lecture-based curricula, curriculum wide implementation of

PBL aims to foster biomedical science knowledge acquisition and understanding while

students negotiate authentic clinical practice problems.

The implementation of PBL varies considerably. Consequently, Barrows (1986)

proposed a taxonomy framed according to obtainable objectives of problem-based

learning including structuring knowledge in clinical contexts (SCC), nurturing the

clinical reasoning process (CRP), the cultivation of efficacious self-directed learning

skills (SDL), and enhancing motivation for learning (MOT). For the purpose of this

research, the fidelity of problem-based instruction is defined according to the following

taxonomy (adapted from Barrows, 1986, pp. 483-484).

1. Lecture-based cases: Students are presented with information in lectures and one
or two patient cases are presented to demonstrate the relevance of the information
provided. No inquiry skills are required.

2. Case-based lectures: Students are given patient case histories as advanced
organizers for content to be presented in a subsequent lecture. Presumably,
activation of prior knowledge is accomplished, but no self-directed learning is
encouraged formally.

3. Case method: Students are given a comprehensive patient case to be studied in
preparation for class discussion. The purpose is to organize and synthesize
information to direct the application of learning. Class discussion amalgamates
student-directed and teacher-directed learning. This technique incorporates
elements of hypothesis generation, information analyses, and clinical reasoning.
However, the patient case information and pattern of discussion questions (often
Socratic style) are synthesized and organized, respectively, by the instructor for
student consumption.

4. Modified case-based method: This approach differs from the case method in that
patient case information is less structured and students may be provided with
numerous alternatives to investigate. However, partial prospective organization
of information obviates a complete and unrestricted inquiry by students.

5. Problem-based learning: Students are presented with a patient's presenting
symptoms. Patient's problems are more ambiguous than those encountered with
case method where the case writer exerts more influence on structuring the issues.
Consequently, students are encouraged to self-direct their inquiries grounded in
their prior knowledge unconstrained by the curriculum, instructor, or peers.

6. Closed-loop or reiterative problem-based learning: This approach extends the
problem-based method in that following initial generation of multiple hypotheses
of what might be happening to the patient; a reflective phase complements the
problem-based format. That is, learning issues are generated collaboratively from
the problem-challenge, and then learning issues are addressed via self-directed
learning. One or two days later, students meet to share the results of self-directed
learning as they re-visit the problem-challenge.

These problem-based learning methods and their associated comparative power

hypothesized to achieve desired educational objectives are summarized in Table 1-1

(adapted from Barrows, p. 483).

Table 1-1. Problem-based learning methods operationalized as a function of relative
power to achieve educational objectives
PBL Method Relative power to achieve educational objectives
Lecture-based cases 1 1 0 1
Case-based lectures 2 2 0 2
Case method 3 3 3 4
Modified case-method 4 3 3 5
Problem-based 4 4 4 5
Reiterative problem-based 5 5 5 5
Note. PBL = problem-based learning; SCC = structuring knowledge in clinical
contexts; CRP = nurturing clinical reasoning process; SDL = cultivation of efficacious
self-directed learning skills; MOT = enhancing motivation for learning. Relative
power to achieve educational objectives is indexed by Barrows (1986) from 0 to 5.

Biomedical science achievement. Is measured as performance on national

medical licensing examinations or locally developed course specific examinations.

Comprehensive biomedical science knowledge in the United States was assessed using

National Board of Medical Examiners i. H'.1L i. Part I until 1992, and thereafter, United

States Medical Licensing Examination (USMLE), Step 1. Step 1 contains over 700

multiple-choice questions that assess student knowledge of important concepts of science

considered basic to the practice of medicine, with special emphasis on biomedical science

knowledge, principles and mechanisms underlying health, disease, and modes of therapy.

Presumably Step 1, usually completed late in the second year of medical school, ensures

mastery of the sciences underlying the safe and competent practice of medicine in the

present, as well as the scientific principles required for maintaining competence through

lifelong learning. Locally developed assessments of biomedical knowledge include

multiple choice questions, short answer, and/or essay. When hybrid formats are

employed for assessment, they are referred to as multi-method, multi-modal or mixed


Clinical science achievement. Is measured as performance on national medical

licensing examinations; locally or nationally developed written examinations; objective

structured clinical examinations (OSCE); multi-station, standardized-patient-based

clinical performance examinations (CPX); and supervisor ratings. For instance, the

clinical science knowledge base typically is assessed : -.,. r. ji ,: ji1. ,, -n the subject

examinations of the NBME-II (i.e., NBME-II Shelf-board examinations) administered to

all medical students in the United States during clerkships (e.g., Family Medicine,

Internal Medicine, Obstetrics/Gynecology, Pediatrics, Psychiatry, and Surgery). Clinical

science knowledge and performance skills are quantified using oral examinations (e. g.,

Objective Structured Clinical Examinations [OSCE]) and supervisor ratings. For

example, the OSCE was developed to measure clinical competencies for both medical

students and residents in medical diagnosis, treatment interventions, laboratory

implications, and follow-up care.

Comprehensive clinical achievement in the United States is assessed using the

national licensing examinations, NBME Part II and Part III until 1992, and thereafter,

USMLE Step 2 and Step 3. Step 2 licensing, usually completed during senior year of

medical school, contains over 700 multiple-choice questions which assess medical

knowledge and an understanding of clinical science essential for the provision of patient

care under supervision. It includes an emphasis on health promotion and disease

prevention. This examination ensures that due attention is devoted to the principles of

clinical sciences that reflect the safe and competent practice of medicine. Step 3

licensing, usually completed during the first year after graduation from medical school,

assesses whether physicians can apply medical knowledge and an understanding of

biomedical and clinical science essential for the unsupervised practice of medicine, with

emphasis on patient management in ambulatory settings. This examination provides a

final assessment of physicians who will be assuming independent responsibility for

delivering general medical care.

Similar in orientation to the clinical achievement assessment in USMLE Step 2

and Step 3, licensing in Canada includes two qualifying examinations, Medical Council

of Canada Qualifying Examination (MCCQE) Part I (written in May of the senior year),

and Part II (written after 16 months of post-graduate training). Locally developed written

assessments of clinical knowledge include multiple choice questions, short answer,

and/or essay. When hybrid formats are employed for assessment, they are referred to as

multi-method assessments.

Problem-solving achievement. Is .p.. .i..ll, defined as search strategies,

data-gathering, accuracy of explanation, coherence of explanation, or use of science

concepts linking causes and hypothesis (e.g., diagnostic reasoning and accuracy). The

initial four elements defining problem-solving are considered important because they are

associated with expertise and differential diagnosis strategies and the last element is

important because it reflects a principal aim of problem-based teaching and learning.

Self-directed learning. Is operationally defined as comprising one or more of the

following elements including average time spent in independent study (e.g., medical

texts, journals), medical discussions with peers or mentors outside of formal settings,

visits to the library, use of library or other information resources (e.g., computer

searches), perceptions of self-directed learning and espoused approaches to study.

Approach to self-study emphasized by problem-based instruction includes learning for

understanding (meaning) and flexibility rather than reproducing information (rote


Attitude toward learning. Is operationally defined as self-rated student

perceptions regarding various aspects of the learning environment, including student

perception of teaching, the congruence between stated learning objectives and

assessments, effectiveness of pedagogical methods, preparedness for clinical practice

(e.g., utility of learning), and faculty attitudes. For example, the Medical School

Learning Environment Survey (MSLES Form A and B) is used to tap this construct. The

MSLES consists of seven subscales including nurturance, organization, flexibility,

student-to-student interaction, breadth of interests, emotional climate, and meaningful

learning experience.

Context of Study

A search was conducted to identify primary research, both published and

unpublished, for the period January 1966 through January 2003 comparing the relative

efficacy of problem-based versus lecture-based teaching and learning in medical students

and physicians. This search included perusal of electronic databases, published literature

reviews, bibliographies of primary research, and the examination of journals likely to

publish PBL and LBL research. PBL databases maintained by experts in the field also

were examined to insure the consideration and inclusion of all relevant research

literature. Reports were selected if results were appropriate to compute a standardized

effect-size comparing knowledge achievement, clinical achievement, problem-solving,

self-directed learning, or attitudes toward learning between problem-based and lecture-

based instruction.

Significance of Results

This is the first study to systematically aggregate research in& i,. and analyzes

statistically the variance of effect size estimates to examine potential moderators of

problem-based teaching and learning efficacy in medical student and physician

education. Results of this research are not definitive due to limitations on cause and

effect inference. However, the strength of this meta-analysis compared to primary

research resides in its ability to investigate data accrued over multiple replications using

varied operational definitions; thus constructing aggregate estimates of magnitudes of

effect sizes for theorized relations. The present research facilitated a broader perspective

with increased statistical power for clarifying and understanding the findings of primary

research comparing problem-based and lecture-based instruction. Patterns not evident in

primary research emerged and weight of inferences and conclusions have unique merit of

being current best evidence-based.

The superiority of traditional perspectives of biomedical science knowledge

transfer (e.g., positivism) is not supported by results of this research. The findings of this

research suggest that responsibility for learning can be transferred to students resulting in

biomedical science achievement at least equivalent to, and clinical science achievement

and problem-solving skills superior to traditional lecture-based approaches while

facilitating self-directed learning in a more enjoyable manner. Despite small effect sizes

by conventional standards (Cohen, 1988) for clinical achievement (ES = 0.32) and

problem-solving skills (ES = 0.19), the evidence is consistent with the perspectives

advanced by some situational learning theorists (Brown et al, 1989; Greeno, 1998; Lave

& Wenger, 1991); and the problem-based instructional effect size magnitude is similar to

values reported in other similar domains (Lipsey & Wilson, 1993, 2001). Furthermore, it

is important to appreciate the fact that effect size cannot be accepted literally. This is

because outcome, methodological and substantive variables influence the magnitude of

effect size. Present findings suggest that even in the same medical education research

domain, not all effect sizes are created equally.

Present research revealed varied operational characteristics of problem-based

teaching and learning. The functional details of implementation are missing from much

of the primary research, thus discovering essential elements and perfecting them remain

problems for future process-oriented research. Although, there was little systematic

examination among studies to measure the actual differences between instructional

methods, an important implication of this research is that teaching and learning with

versus without the integration of biomedical and clinical science facilitates superior

clinical science achievement and problem-solving skills in medical students and


Despite the limitations of this research current best evidence derived from a

synthesis of the available research suggests PBL versus LBL affects the efficacious

integration of medical science theory and practice as evidenced by superior clinical

science knowledge and problem-solving achievement while promoting self-directed

learning and positive attitudes toward learning, which are the expressed aims of medical

education (Liaison Committee on Medical Education, 2002).

Limitations of Study

The data employed in this research are only as valid as the primary studies.

Consequently, answers to questions posed by this research are presented to the extent

available in the literature. Ultimately, however, findings and inferences resulting from

this research were influenced primarily by the methods employed to analyze these data

and interpret the results within the framework of principles of information processing,

social constructivism and situated learning. Main effects for clinical science

achievement, problem-solving success, self-directed learning skills and positive attitudes

toward learning are robust and can be generalized with significant confidence, but

analyses of some moderators were limited by small data sets. Implementation details are

missing from much research literature, which limited the sensitivity of analyses, and

heterogeneity within categories of many study features could not be resolved, which

renders impossible unambiguous inferences. Indeed, a significant limitation of the

available comparative research is that it focuses almost exclusively on learning outcomes

rather than the process of learning. This is incompatible with recent attention afforded to


the principles of differential psychology and the social construction of knowledge. Thus,

discovering the essential elements of problem-based teaching and learning and then

perfecting them remains for future research. Expectantly, future research will be guided

by the specific gaps identified and examined in the present research.


This research is designed to examine the differential efficacy of problem-based

versus lecture-based teaching and learning in medical education. Theoretical framework

for this research is grounded primarily by educational philosophy, information-processing

theory, and principles of social constructivism and situated learning. Throughout the

literature review, I will employ this framework to provide insight for the interpretation of

aims, claims, and research evidence concerning the cognitive and non-cognitive efficacy

of problem-based medical education. In particular, I will review the literature exploring

the development of medical problem-solving expertise, and the roles of problems,

students, small-group tutorial, and self-directed learning in problem-based teaching and

learning. A synthesis of 11,di .: .i,..1 implications leading to the present study conclude

the chapter.

Theoretical and Scientific Foundations of Problem-Based Medical Education

Problem posing as an instructional method has been employed for at least two-

millennia (Dewey, 1948). Ancient Greek educators challenged students with problems to

stimulate critical inquiry, problem solving, theory construction, reasoning, and the

acquisition of knowledge. Problem-based teaching and learning is a contemporary

manifestation of this classical Greek instructional method. It was introduced to remedy a

medical school curriculum that over-emphasized rote memorization, artificially isolated

the biomedical sciences from clinical training, and failed adequately to prepare medical

students for the kinds of problems encountered in clinical practice (Neufeld & Barrows,

1974; Barrows & Tamblyn, 1980). As mentioned previously, this anecdotally based

concern was empirically substantiated by a survey of 453 senior students undergoing

traditional medical education at five universities (Regan-Smith et al., 1994). Forty-nine

percent of the students reported they learned half or more of biomedical science by

memorizing without understanding. Rote learning does not provide the student with the

basic knowledge necessary for solving clinical problems (Bordage & Zacks, 1984). In

particular, the inability to solve clinical problems is due both to a lack of organization and

a lack in understanding the relevant information stored in memory (Bordage & Zacks;

Sobral, 1995). The successful resolution of complex medical problems is I .p.ri..i id to

most likely occur when medical information is linked to networks of concepts and their

relationships, then compiled into compact representations called "chunks" (Anderson,

Pirolli, & Farrel, 1988; Chase & Simon, 1973) or schemataa" (Perkins & Salomon, 1989)

that links medical knowledge, clinical procedural knowledge, and production rules


Sobral (1995) confirmed this hypothesis in medical students employing the

validated and reliable Diagnostic Thinking Inventory (Bordage, Grant, & Marsden,

1990). The Inventory measures flexibility in thinking and knowledge structures in

memory (Bordage et al.). Sobral studied the relationship between knowledge

representation and the application of knowledge to problem solving in third-year medical

students undergoing problem-based versus lecture-based learning. He discovered

students undergoing PBL exhibited significantly higher scores in knowledge structures in

memory measure (ES = 0.37, p=.018), but statistically equivalent -1.. ;ih.i in thinking

scores (ES = 0.11, p= .52). Intergroup disparity in functional knowledge structures was

independent of academic achievement, gender, and validated and reliable measures of

learning style, self-confidence, and motivation to learn. Detailed findings and analyses

are presented in Chapter 4. Sobral concluded that the difference in diagnostic expertise

was due to increased organization and understanding of information, an observation that

confirms and extends the findings of other medical (Boshuizen, Schmidt & Wassmer,

1994; Norman & Schmidt, 1992) and nonmedical (Chase & Simon, 1973; Chi,

Feltovitch, & Glasser, 1981) educators.

In a classic study, Chase and Simon (1973) asked expert and novice chess players

to view for 5-seconds to 10-seconds chessboards arranged with authentic games, then to

recall the positions of the 25-chess pieces. Investigators observed expert chess players

could accurately position 90% of the chess-pieces, whereas novices achieved only 20%

accuracy. Accounting for this differential accuracy, the authors concluded that the long-

term memory of chess experts contains many patterns associated with successful chess

maneuvers and strategies (i.e., pattern recognition). However, expert and novice

accuracy were equivalent when trials were repeated with randomly positioned chess-

pieces. Apparently, unless patterns of chess pieces emerge from natural play (i.e.,

typical), advantage of expertise vanishes. That is, in an inauthentic, decontextualized

situation (i.e., atypical), expertise is -i,, iFi.. ..ii diminished, even in an expert's domain.

Implications for education are that although while domain-specific knowledge (e.g.,

biomedical science knowledge) is essential, complex problems can not be solved by

retrieving information and applying procedural rules. Since the resolution of complex

problems depends upon context, domain-specific knowledge must be integrated

appropriately to fit in the problem-space. That is, goodness-of-fit between context-

indexed knowledge and problem-context influences the accuracy of problem solving

performance (Perkins & Salomon, 1989).

In a similar study, Chi and associates (1981) experimentally compared expert

physicists' and physics students' understanding of physics problems by asking them first

to categorize word problems, and then to classify isomorphic problems into similar

categories. Investigators discovered novices focused upon superficial features, whereas

experts articulated problems according to the types of physical principles, theorems, and

laws required to resolve the problem. Importantly, experts' mental representations

appear to be grounded in solution principles (e.g., solved exemplars). Finding suggests

that experts have access to vast procedural knowledge about solutions and their

applicability. This allows experts to represent problems on a level directly related to their

solution or improvement. Compared to experts, novices have less access to information

about effective solution procedures, a view expanded by social constructivists (e.g.,

pragmatic constructivism, Cobb, 2002) and situational learning theorists (e.g.,

concurrently linking concepts and context-particular solutions, Brown et al., 1989;

Perkins, 1986), and is discussed later.

Taken together, these studies suggest that the ability to solve medical problems

can be enhanced best by initially fostering the development of explanatory mental

representations during declarative and procedural knowledge acquisition. Students then

subsequently employ the schemes to recall the information needed to improve or solve a

clinical problem situation. Encapsulating both knowledge acquisition and problem

solving has the advantage of amalgamating into a single operation knowledge

construction and a search-and retrieval strategy (Brown et al., 1989; Chi et al., 1981;

Lesgold et al., 1988; Perkins, 1986).

Extending their initial observations, Chi, DeLeeuw, Chiu, and LaVancher (1994)

revealed significant disparity between good and poor learners in the extensiveness and

quality of problem representation and self-explanations advanced as students negotiated a

problem-space. Their results suggest good learners also exhibit better cognitive

monitoring (metacognition) skills and the ability to assess what they have learned when

challenged with a novel problem. Self-explanations, or private verbal elaborations

related to the problem challenge are viewed by the authors as a mode of studying that can

mediate learning. Moreover, these observations provide evidence for the active, self-

construction of knowledge.

Vygotsky (1926/1997; 1962) described a similar self-mediated learning strategy

as "private talk." In a similar vein, Gal'perin (1969) conceptualized this strategy as

"dialogical thinking." Dialogical thinking initially involves speaking aloud about the task

or concept, followed by covert speaking, that is, "speech minus sound" (p. 263). In the

mind, Gal'perin argues, overt speech is represented by a deep structure, or as he

describes, "the audible image of a word" (p. 263). For Gal'perin, an audible image is

more stable and durable than a visual image. This line of reasoning is similar to the

theory of multiple encoding proposed by Paivio (1986). According to Paivio's Dual

Coding Theory, cognition incorporates representational systems called verbal and visual

systems. Accordingly, when information is coded in both systems, memory is enhanced.

Medical learning from this perspective would not only focus on the acquisition of

biomedical and clinical science knowledge, but also the cognitive representation of this

knowledge. Therefore, efficiency in representing medical knowledge for appropriate re-

deployment in clinical practice is hypothesized to be an essential ingredient in the

development of problem-solving expertise (Boshuizen et al., 1994; Patel et al., 1988;

Schmidt, Norman, & Boshuizen, 1990).

Patel and colleagues (1988) examined problem-solving expertise development in

medical students. They asked students at three separate stages of a biomedical sciences

curriculum--novices, in their first six-months of study, intermediates during their senior

year, and relative experts after approximately two-years of clinical experience-- to solve a

clinical problem and then integrate passages of relevant biomedical science information

into their pathophysiologic explanations of the problem (causal explanations).

Intermediate students were observed to employ more biomedical science information in

their problem-solving explanations than either novice or experts. Explaining the

differential use of biomedical science knowledge, authors presented an intuitively logical,

but previously unsubstantiated assertion that novices lacked adequate biomedical

knowledge, and, unlike intermediate students, experts filtered out inessential biomedical

information in their reasoning process. The implications for medical education are

important, as these findings suggest that there is a necessary threshold for study in the

biomedical sciences if students are to achieve sufficient problem solving expertise.

Boshuizen and associates (1994) conducted an investigation comparing

instructional influence on the integration of biomedical and clinical sciences for problem

solving. They challenged students undergoing either pre-clinical education with

problem-based or lecture-based instructional curriculum with an identical metabolic

problem. Students were asked to solve the problem employing their knowledge of

biomedical science (biochemistry) and clinical science (internal medicine). Students

from problem-based instructional backgrounds initially employed an analytical approach

to exploring biochemical aspects and subsequently linked them to clinical elements of the

problem (i.e., causal reasoning). In contrast, students from lecture-based instructional

backgrounds tended to employ a memory-based approach to ascertain a solution directly,

a common process when learning occurs via rote-memorization. Similar observations in

medical students undergoing traditional lecture-based teaching and learning have been

reported by other investigators (Lindblom-Ylanne & Lonka, 1999, 2002; Miller, 1961;

Neame, 1984; Regan-Smith et al., 1994). Inquiry strategy employed by students

undergoing problem-based teaching and learning resulted in significantly greater problem

solving accuracy. Detailed results and analyses are presented in Chapter 4.

Schmidt and colleagues (1996) conducted a similar study, but with thirty

diagnostic problems epidemiologically representative of Dutch medical practice. Also

included in the study were students attending a medical school employing an integrated

didacticss integrating biomedical and clinical sciences), teacher-directed curriculum.

Compared to students undergoing typical lecture-based instruction, students undergoing

integrated lecture-based and reiterative problem-based instruction exhibited superior

diagnostic accuracy. There were no differences in diagnostic accuracy between students

from integrated and PBL curricula. Detailed results and analyses are presented in

Chapter 4. The authors suggest that the integration of biomedical and clinical sciences

grounded in patient problems is likely an essential feature that enables superior diagnostic

performance independent of whether instruction is teacher- or student-centered. This

observation, if corroborated, has .,I, .i implications for medical education and

provides support for teaching medical concepts as tools (Brown et al., 1989; Perkins,

1986), that is, the integration of theory and practice, or pragmatic constructivism (Cobb,

2002), which is discussed later.

Schmidt and associates (1990) propose that medical students transition from

concept novice to content expert in four-stages. Stage one is characterized by the

formation of elaborate causal networks. These networks account for the causes and

consequences of patient problems using -I..1 ,i i ..hi.. ,I ;.:..:.. ;..,i mechanisms.

Practice fosters the application of declarative knowledge and networks progressively

compile into "high-level, simplified causal models explaining signs and symptoms and

are subsumed under diagnostic label" (Schmidt et al., p. 614). Transition to a second

hypothesized developmental stage occurs when medical students undergo exposure to

patient diagnosis challenges employing a process called "differential diagnosis." For the

novice, "differential diagnosis" requires significant cognitive effort during which students

attempt to link patient symptoms (problem-space cues) to knowledge in a relevant

pathophysiological network. Presumably, continued clinical practice results in the

development of abridgments (the elimination of sub-goals) and the encapsulation of

knowledge (compact representation) corresponding to the process of chunkingg"

described in information processing literature (Anderson, 1990). Anderson introduced

this idea in psychology to explain the learning by doing phenomenon. By reconstructing

representation of medical information, knowledge can be structured and organized into a

form that facilitates faster deployment (Bordage & Zacks, 1984; Glasser, 1984, 1991;

Perkins, 1986). For Anderson, the basic idea behind differential construction of cognitive

representations for deployment efficiency is to remove knowledge that is only employed

in the intermediate steps. Theoretically, reconstructed knowledge streamlines or speeds

up processing which allows problem-solution to be reached directly from problem-space

cues. Thus, as successful differential diagnosis experience increases, medical students

become less likely to activate all possible relevant knowledge to learn diagnosis; rather

only the knowledge relevant for a particular case is activated.

Script theory suggests that in given situations students activate knowledge sets to

understand a situation. For Schmidt and colleagues (I' 'I this ability to distinguish the

particulars of a case without extensive knowledge activation represents a transition from

a network model to an illness-scripts model (Stage 3). Illness-scripts only compile

information required for diagnosis and treatment, e.g., antecedent conditions, patient-

problem description, signs, and symptoms. Scripts contain data about links that unite

items of knowledge (e.g., clinical cues or features) related to an illness. In problem-

situations, these links facilitate decisions regarding the strengths and weaknesses of a

tentative diagnosis.

Stage four of clinical problem-solving expertise development occurs when illness-

scripts remain available in long-term memory to be employed for use subsequently when

similar patient problems present. That is, the memory of specific patients is represented

as "individual entities and not merged into some prototypical form" (p. 617).

Consequently, clinical reasoning is fueled by similarity in presenting cases and the

retrieved memory of previous patientss. The authors suggest that these memories are

stored in episodic memory rendering them readily retrievable.

In summary, illness scripts theory postulates that physicians consistently recall

information in the same sequence independently of how data were presented.

Accordingly, when confronted with a patient problem, a physician searches his repertoire

of illness-scripts for a match (pattern recognition), which further instantiates relevant

illness-script. As discussed earlier, similar cognitive processes were observed in expert

chess players (Chase & Simon, 1973) and physicists (Chi et al., 1981).

An unanswered question is what biomedical science content threshold must be

attained before students can self-explain (Chi et al., 1989, 1994; Gal'perin, 1969;

Vygotsky, 1926/1997; 1962), monitor their learning (Semb & Ellis, 1994; Winne, 1995a,

1995b), and effect integration of declarative knowledge (biomedical sciences) and

procedural knowledge (clinical sciences) to facilitate the development of production rules

(Anderson, 1990) or differential diagnosis skills (Schmidt et al., 1996)? Winne suggests

the quality of self-regulated learning, a key feature of reiterative PBL (although called

self-directed learning by medical educators) may be affected by a student's domain

specific declarative knowledge. If a learner's mental representations for a topic are

limited (i.e., low-domain knowledge), there is less ability to associate novel information

(Schon, 1987). This is true for any form of cognitive learning from recall of declarative

knowledge to application of concepts or production rules (Anderson, 1990; Sch6n). This

question addresses instructional design concerns regarding the magnitude of the

biomedical science foundation necessary before medical students can efficaciously

engage in solving clinical problems (e.g., differential diagnosis) and self-direct their

learning performance. This will be discussed later.

As mentioned previously, Flexner (1910, 1925) addressed the tension between

acquiring knowledge and applying it by insisting a solid foundation in biomedical science

was a prerequisite for subsequent clinical encounters. Positing an extreme position,

Flexner opposed any clinical education during the first two-years of medical school

(1925). Consequently, as was discussed in Chapter 1, traditional medical school

instruction is characterized by the temporal isolation of biomedical and clinical sciences

(Ebert, 1992; Sch6n, 1987). Whereas, as discussed earlier William Osler (1903) argued

for a more practice-based medical education. Reflecting epistemology advanced by

Osler, Neufeld and Barrows (1974) believe imparting biomedical science independent of

the clinical context in which it needs to be applied is inappropriate. Neufeld and Barrows

view clinical problems as contextual focal points for integrating biomedical and clinical

sciences into a coherent and practical learning experience. As a result, problem-based

learning was conceived of as an instructional method employing patient problems as the

context for acquiring knowledge about biomedical and clinical science and transitioning

from content novice to content expert while cultivating metacognitive and problem-

solving skills (Barrows, 1986, 1990, 1996, 2000; Barrows & Tamblyn, 1980; Boud &

Feletti, 1997; Neufeld & Barrows). Problem-based learning emerged from an

educational stance that envisioned learning as central to the educational process, and

learners as active, self-directed participants in that process (Neufeld & Barrows). That is,

problem-based teaching and learning is not solely the addition of problem-solving

activities to an otherwise discipline-centered curricula; problem-based learning is an

educational method applied to a domain of knowledge rather than an isolated

instructional technique (Engel, 1997).

A pedagogical strategy incorporating problem-based learning is aligned with

Glasser's (1991) view that the psychology of instruction should strive to understand the

development of cognitive structures and the processes that characterize the competent

performance of cognitive skills by experts in particular knowledge domains. In concert

with SchOn (1987) and Perkins (1986), Glasser argues that instructional design is best

when guided by a theoretical framework characterizing the acquisition, structure and

organization of knowledge guiding expert practitioners. In contrast, traditional lecture-

based instructional approaches first present discipline-oriented biomedical science, and

then clinical science, during which students presumably learn to apply this knowledge to

health care problems of everyday practice. Assumed apperception may be more apparent

than real (Sch6n). Because research suggests much of biomedical science learned in

traditional manner subsequently is unable to be recalled, or discovered to be irrelevant

(Miller, 1961; Neame, 1984; Schmidt, 1983; Schon).

More than 40-years ago, Miller (1961) observed poor retention of biomedical

science information by his medical students who underwent lecture-based, discipline-

oriented instruction and warned that information presented via teacher-centered methods

would "decrease from memory at the same rate as nonsense syllables" (p. 153). Further

evidence is provided by Levine and Forman (1973) who presented fifty clinically relevant

questions to students starting a neurology clerkship (senior year). Questions were

obtained from a final examination each student had passed in their first year of medical

school. Upon retesting, only 48% of senior medical students received a passing grade.

The authors concluded, "a substantial number of students fail to retain the information

taught in the preclinical years" (p. 869). Such results prompted Neame (1984) to write

"there is little evidence that the existing preclinical courses are essential to high quality

medical practice; mastery of them seems to confer little benefit" (p. 707).

These outcomes may be attributable to the intentional demarcation between

theoretical and clinical sequencing of conventional medical curricula (Sch6n, 1987).

Medical students educated in this manner encounter difficulty when they transition from

theory to practice (Patel et al., 1989). They are inadequately prepared when they

encounter problems requiring the transfer of their learning to new domains, an ability

essential to functioning effectively in practice (Barrows, 1986, 1990, 1996, 2000; Schon).

These findings have important implications for medical education in general and clinical

practice competence in particular; to manage the relentless medical knowledge explosion

efficiently, students and physicians must have an extensive awareness of sciences basic to

medicine (Barrows, 1983; Tosteson, 1994).

Barrows (1983, 2000) posits that when biomedical science content is coupled with

a genuine health care problem, students will integrate knowledge into meaningful

representations. This epistemological stance is nicely characterized by Glasser (1991)

who wrote "Instruction should have the mission of making itself unnecessary; learners

should become mindful architects of their own knowledge" (p. 131). Indeed, the

philosophy of PBL is represented in an ancient Chinese proverb (Smith, 1984, cited by

Bickley, Donner, Walker, & Tift, 1990).

If you tell me, I will forget.
If you show me, I might remember.
But if you involve me, I will learn.

Di.. i..i... the etymology of the word curriculum, Walton and Mathews (1989)

emphasized a curriculum is more than a matrix of topics, just as a house is more than an

assemblage of bricks and mortar. For Walton and Mathews, the "bonding" and

"structure" are essential features of curricular construction. Theoretically, a medical

curriculum that promotes contextual learning and self-directed learning and fosters

motivation while focusing on the health care needs of society provides substantial

bonding and structure. Problem-based instruction appears to support this approach more

adequately than discipline-oriented, teacher-centered, syllabus-driven, lecture-based

instruction. Problem-based instruction aims to (a) structure knowledge for use in clinical

contexts, (b) develop effective clinical reasoning process, (c) develop effective self-

directed learning skills, and (d) increase a positive attitude toward learning (Barrows,

1986). Reverberating principles of adult learning theory (Knowles, 1975), Barrows

suggested the nature of health care problems and the locus of control of learning effect

the probability of realizing the aims of PBL. The nature of health care problems can be a

complete case, a vignette, or a comprehensive problem simulation. Locus of learning can

be teacher-centered, student-centered, or a hybrid of the two. Consequently, there are

varied instructional approaches to problem-based teaching and learning.

As discussed in Chapter 1, Barrows (1986) developed a taxonomy to distinguish

operational variants of problem-based learning. For Barrows, only reiterative PBL

achieves all four aims, thus this pedagogical classification is "as much a taxonomy of

teaching-learning methods, within which problem-based learning fits, as it is of problem-

based learning itself' (p. 38). Reiterative PBL views construction of a framework of

knowledge and ways of thinking about clinical problems as essential features of effective

medical education. Moreover, quintessential elements of the reiterative PBL framework

is that it includes all antecedents of health and expresses the Aristotelian aphorism that

"medicine begins in philosophy, and philosophy ends in medicine" (cited in Tosteson,

1994, p. 9). From this perspective, PBL appears to be both a pedagogical method and a

philosophy of education at once (Maudsley, 1999). A reiterative PBL curriculum is

structured to foster a process designed to establish and sustain an environment of free-

inquiry and co-learning which includes a tutored group of students (Barrows, 1990).

Free-inquiry and collaboration provides an opportunity for the kind of academic

pluralism espoused by Dewey (1933/1998). This process emerges in a particular

sequence, for example, the Maastricht University (The Netherlands) "seven jump"

(Schmidt, 1983), which enables learners to identify their needs in understanding a

problem and, once these are identified, to pursue their goals usually independently, and

finally regroup to synthesize their findings. According to Schmidt, the process should

incorporate the following seven-steps (p. 12).

1. Tutor clarifies terms and concepts not readily comprehensible;
2. Tutorial group defines problems; agree which phenomena need explanation;
3. Tutorial group analyzes problem via brainstorming based on prior knowledge and
4. Tutorial group proposes possible explanations and develops working hypotheses;
5. Tutorial group generates and priorities learning issues;
6. Students research learning issues via self-directed learning;
7. Students synthesize explanations in tutorial group, and collaboratively apply
newly acquired information to the problem.

Savery and Duffy (1995) contribute additional insight with a comprehensive

model for reiterative PBL implementation incorporating principles of social

constructivism and ways to apply those principles to learning and understanding. They

characterize the philosophical stance of social constructivism with three propositions.

First, understanding is intertwined with one's interactions with the environment. Second,

cognitive conflict or an indeterminate situation is a stimulus for learning and influences

the organization and nature of what is learned. Third, knowledge and warranted

assertions emerge via social interaction and self-assessment of understanding. For

Savery and Duffy, PBL is a pedagogical approach that most ideally reflects and employs

these propositions. Grounded in this framework, Savery and Duffy propose eight

instructional principles to guide the implementation of PBL (pp. 32-34).

1. Anchor all learning activities to a larger problem.
2. Support learner in developing ownership for the overall problem or task.
3. Design an authentic task.
4. Design task and learning environment to reflect the complexity of the
environment the learner should be able to function in at end of learning.
5. Give learner ownership of process used to generate solution.
6. Design learner environment to support and challenge learner's thinking.
7. Encourage testing ideas against alternative views and alternate contexts.
8. Provide opportunity for, and scaffold reflection on both content learned and
learning process.

What then are the active ingredients of problem-based teaching and learning?

Shulman (1986) suggests "the curriculum and its associated materials are the material

medical of pedagogy, the pharmacopeia from which the teacher draws those tools of

teaching that present or exemplify particular content and remediate or evaluate the

adequacy of student accomplishments" (p. 10). Barrows (1986), Schmidt (1983), and

Savery and Duffy (1995) provide the material medical of a pedagogical format whereby

medical students actively acquire information and are encouraged to reason and generate

hypotheses that can subsequently be confirmed or refuted. During this process, students

identify learning issues to be studied during subsequent periods of self-directed learning,

and this acquired knowledge is shared and collectively applied to analyze authentic

problems. This is a learning process characterized by free inquiry, collaboration and self-

reliance, the pharmacopeia from which students draw tools for developing expertise and

professional practice competence.

Features of Problem-Based Learning

Lave and Wenger (1991) differentiate between a teaching curriculum and learning

curriculum. "A teaching curriculum provides structured resources and meaning for the

learners which are moderated by the instructor." Whereas, a learning curriculum

incorporates situated opportunities, for instance, varied exemplars from everyday practice

"viewed from the perspective of the learners" (Lave & Wenger, p. 97). Engendering this

stance, authentic problem-centered learning builds a cognitive repertoire of exemplars

(e.g., schemata or illness scripts) from which knowledge and skill may be appropriately

retrieved and applied to the solution of practice problems (Norman & Schmidt, 1992).

Echoing Dewey ( 1-" I "- i, this stance is contrary to traditional, or objectivist theories

of teaching and learning which value abstract knowledge over practice, thus

dichotomizing theory and practice (Biggs, 1996; Schon, 1983, 1987).

Problem-based instruction aims to focus the learning process, not the content. In

this manner, problem-based instruction transfers control of the learning process from

teacher to student, which is conforms with concepts of "learning curriculum" (Lave &

Wenger, 1991) and "thinking curriculum" (Resnick & Resnick, 1992, p. 37). As

suggested by Barrows (1986, 2000), Schmidt (1983), and Savery and Duffy (1995),

realizing the aims of problem-based teaching and learning requires a multidimensional

instructional strategy that incorporates roles of problems, students, small-group tutorials,

self-directed learning, and assessments.

Role of Problems

A design feature of problem-based teaching and learning is to construct problems

that are effective in fostering an adaptation to clinical problem space and cultivating

expertise. In this respect, problems serve as the initial stimulus and framework for

learning biomedical and clinical science content. Thus, problems "become the 'cognitive

scaffolding' for information rather than the scientific discipline itself' (Bickley, 1993, p.

545). By this means, problems are not constrained by disciplines and designed to

integrate biomedical and clinical sciences. From an educational perspective, "problem

posing" is a pedagogical tool employed as a challenge leading to learning which emerges

while working on the problem. A problem provides a context that theoretically reinforces

factual learning and the application of concepts. Problems can be employed as orienting

activities to stimulate curiosity, illustrate principles, and as advanced organizers to

stimulate prior knowledge (Ausubel, 1960). Consequently, a fundamental assumption of

problem-based instruction is that it renders learning more relevant, and thus more

transferable, by engaging students in authentic practice problems (Schmidt, 1993). For

Schmidt, at least four-aims of student learning are served by employing problems

including (a) problems trigger the activation of whatever knowledge is available; (b)

authentic problems situate activated knowledge and novel information in a professional

practice context; (c) problems trigger cognitive conflict and group discussion helps

learners to elaborate their knowledge structures; (d) problems engage students in such a

meaningful manner as to inspire epistemic curiosity (the desire to know).

Problem structure. Problems may be either well-structured or ill-structured

(Newell & Simon, 1972). Well-structured problems are those with a single, agreed upon

solution. Both initial state and appropriate end-point (goal state) are clear and

unambiguous. Ill-structured problems are those without an agreed upon solution and

there are likely to be differential means of solving the problem. However, well-

structured and ill-structured problems should not be viewed as a dichotomy; rather they

anchor two ends of a problem-space continuum (Reitman, 1965). Indeed, as Simon

(1973) noted, many authentic problems are presented as ill-defined problems, but become

well-structured in the hands of the problem solver. Simon emphasizes information

provided by the problem-solver is responsible for converting an ill-structured problem to

a well-structured problem. Simon also suggested differential content knowledge to

discriminate strong from weak problem-solving performance, but cautioned that problem-

solving expertise is not grounded entirely in mastery of passively recalled content.

Knowledge must be retrieved and structured to be functional for clinical problem solving

(Bordage & Zacks, 1984; Glasser, 1984, 1991; Perkins, 1986; Sternberg, 1994); because

medical problems often necessitate acquisition, analysis, and synthesis of supplemental

information (Elstein, Shulman, & Spafka, 1978).

The initial step in problem solving is to encode given elements of problem-space

(Newell & Simon, 1972). Encoding involves identifying most salient features of a

problem space, storing these features in working memory and retrieving relevant

information from long-term memory. A problem-space consists of the external world as

perceived, internal representations of these perceptions, and resulting interactions

(Greeno, 1998). From a situational learning perspective, a problem-space has intrinsic

features, or affordances as advocated by Gibson (1977). For Gibson, affordances entailed

what environment offers to the problem-solver, both good (potentials) and bad

(constraints) qualities. Thus, affordances are not invented by the problem-solver;

affordances are perceived and populate objective sense of situation. Gibson's perspective

represents an extension of Dewey's (1933/1998) notion that a problem situation provides

"unique doubtfulness" (p. 108), that is, doubt lies not in the mind, but in the situation.

Further, "The nature of the problem fixes the end of'iu..i, and the end controls the

process of thinking" (p. 15).

Problem function. An ability to provide a robust, high-fidelity representation of

clinical problem-space is instrumental to effective problem-based instruction settings

(Savery & Duffy, 1995). In particular, the extent to which physical environment and

social context are salient to problem situation will govern student perceptions of the

authenticity and relevance of contextual elements (Barrows, 1986). The selection of

problems is guided by curricular objectives (Barrows, 1996; Boud & Feletti, 1997). The

student's need-to-know is stimulated by the problem, and an important gauge of knowing

is the ability to apply knowledge to the problem (Norman, 1988). By instructional

design, the student is responsible for recognizing the need to know; acquiring the

necessary knowledge and skills, and applying this back to the problem in a reiterative

process (Barrows, 2000; Savery & Duffy). Students identify learning issues posed by

specific problems to help develop understanding about underlying concepts and

principles. Thus, knowledge acquisition and understanding are constructed via working

on a problem ("concurrent applicability") rather than in the traditional approach in which

acquired knowledge is a prerequisite to working on a problem ("retrospective


From this vantage, problem interpretation is a function of perception and learning

to perceive situational affordances that are instrumental to developing problem solving

expertise. Dewey (1933/1998) and Gibson (1977) further imply that interpretation does

not begin where perception ends; rather efficacious perception adapts and emerges with

action in the problem-space. Extending these positions, Schdn (1987) envisions experts

as practitioners who efficaciously link perception, interpretation and action (e.g.,

knowing-in-action) to resolve clinical problems. The capacity to facilitate an optimal

solution or to improve a situation, varies with the problem-solver's ability to inquire and

work creatively within the knowledge and skill demands of a problem and the

affordances (cues) of problem space (Greeno, 1998).

Importantly, a health care problem may cue an enormous range of associated

knowledge, much of which will be irrelevant to achieving a solution or improvement.

Cue acquisition is the process of collecting data in clinical problem solving (Elstein et al.,

1978). For medical students and physicians, research suggests accuracy of cue

acquisition and interpretation is associated with accuracy of diagnosis (Elstein et al.).

I hu.. effective cue acquisition and interpretation can transform an ill-structured problem

into a well-structured problem (Simon, 1973). Lovett and Anderson (1996) argue that the

appropriate use of problem-space cues and eventual problem solving success are

influenced by an individual's ability to integrate processing current information with

processing experienced-based information. Instrumental to the integration of

performance and learning is a conceptual coupling of immediate information and

successful experience in a manner quantifiable and meaningful (Lovett & Anderson).

This position, which reflects the essence of situated learning principles, is discussed later.

In a similar vein, Sternberg (1994) suggests problem-solving expertise involves

the trajectory through memory encoded with strong context specific cues to limit the

retrieval of relevant information for a successful solution (i.e., encoding specificity)

(Lovett & Anderson, 1996). Sternberg's position is derived from evidence that indicates

successful retrieval is contingent upon the context in which that knowledge was

constructed (Godden & Baddeley, 1975). Context dependence is demonstrated when

superior recall occurs when the retrieval situation corresponds to the situation of initial

learning. The potency of context-dependent learning was revealed in a classic study by

Godden and Baddeley who investigated recall in Royal Navy divers. One group

memorized twenty words on shore, while another group memorized the same words

while submerged. Words memorized on shore were recalled better on shore and words

learned submerged were better recalled underwater. Thus, memory for words was

significantly better in the context of original learning. Indeed, encoding specificity as an

element of information-processing theory (Lovett & Anderson) resembles context

emphasis in a situational learning perspective (Albanese, 2000). This is discussed later.

Learning from this perspective is not only associated with the acquisition of knowledge,

but also its representation. Therefore, as indicated earlier, efficiency in representing

knowledge for re-deployment is viewed as a primary determinant of problem-solving

expertise. So, the flexible application of knowledge in problem solving requires that

knowledge elaboration .._.1, 'ji .....i'1 .. Iuall,

Clearly, knowledge in a specific context is necessary for initial learning, but not

sufficient for understanding. The implications for PBL are that these dimensions must be

addressed individually to create a learning environment conducive to knowledge

construction, relevant retrieval, and appropriate action in practice (Schdn, 1983, 1987).

Guided by the principle of prior knowledge activation, advocated by both information

processing theorists (Anderson, 1990; Ausubel, 1960) and social constructivists (Savery

& Duffy, 1995), instruction can be tailored to foster linkage (networking) and integration

of existing knowledge with the processing of new information.

A core property of problem-based teaching and learning is its representational

flexibility, a quality that facilitates inferential use of knowledge in novel contexts

(Barrows, 2000). Theoretically, perceiving and confronting authentic health care

problems will shape a medical student's clinical reasoning and problem solving skills.

When students are challenged with such a complicated, conceptual process, perception

and action are learned together as a unit (Dewey, 1929a, 1948, 1933/1998).

Extrapolating this insight to medical education, learner, activity and environment

mutually constitute the other. For Dewey, experience is the initiating phase of thought

for a student, as students need an empirical situation to engage their interests and

generate action. Thus, thought is intertwined with action. Dewey wrote ". old

experience is used to suggest aims and methods for developing a new and improved

experience. Consequently experience becomes in so far constructively self-regulative"

(1948, p. 94). Indeed, Dewey's stance of knowledge-in-use is grounded in his conception

of educative experience and education as growth (Soltis, 1990). Clarifying Dewey's

position, Soltis wrote:

Previous experience that enters into present experience to inform it, organize it,
transform it, and reconstruct it is not just useful knowledge in the technical sense
of knowing how to do something, It is useful in the richer and broader sense of
being able to use one's past experience to orient oneself in a new situation, to
interpret its manifold dimensions, to analyze its components, to guess at or
anticipate its future, and to bring one's purpose to bear on the ongoing interaction
of self and situation. (p. 311)

Inherent in Dewey's (1916, 1948) educational philosophy is the activation of

relevant prior knowledge as a prelude to constructing ("intellectual growth") new

knowledge. For Schon (1987), Deweyan inquiry is similar to the concept of designing,

"not the activities of the 'designing professions' such as architecture, landscape

architecture, and industrial design, but the more inclusive processes of making things

(including the representations to be built) under conditions of complexity and

uncertainty" (p. 30). Examples are physician's construction of a diagnosis or a teacher's

construction of a lesson plan (Sch6n).

Also, consistent with Dewey's (1933/1998) advice, situated learning aims to teach

knowledge and skills employing contextual experiences that reflect how knowledge will

be useful in practice (Lave & Wenger, 1991). If development of a student's ability to

think is a principal goal of education, then students must learn in situations where

thinking has an opportunity to occur (Dewey). Specifically, Dewey advocated situations

designed to foster reflective inquiry. For Dewey, reflective inquiry mediated experience.

A learning environment fostering reflective inquiry would include a genuine situation of

experience and a genuine problem developed within a situation as a stimulus for thinking.

The student must possess cognitive resources to make observations (perceive

affordances), to work the problem (problem-solving strategies), and be responsible for

suggesting solutions (propositions) while testing those propositions to reveal their value

and discover their validity (Dewey). Dewey's conception of inquiry resembles scientific

method and integrates theory (abstract representations) and practice (Soltis, 1990). That

is, reflective inquiry, oriented vis-a-vis established theory, develops propositions and

tests them within the context of a problem. For Dewey, "To grasp a meaning, to

understand, to identify a thing in a situation in which it is important, are thus equivalent

terms" (p. 117). Thus, performance in some situations is limited by a tendency to

recognize problems and attune to affordances. From this perspective with empirical

support (Godden & Baddeley, 1975), the power of knowledge is situation specific and

influences one's mode of participation in a problem-space. Further, situational

empowerment can be viewed as integrating power potential (competence) and actual

power (performance) (Bandura, 1986, 1997). This is discussed later.

Influenced by Deweyan tradition, an assumption of problem-based instruction is

that clinical perception is an active rather than a passive process, and results from

attention, prior experience, and a conscious search for information (Barrows, 2000;

Barrows & Tamblyn, 1980). The initial perception of a problem situation serves as a

stimulus for solution ideas, thus this represents a crucial phase in the quality of clinical

reasoning (Barrows & Tamblyn). A challenge for medical students is to actively attend

to and interpret situational affordances. That is, to exhibit situational intelligence and be

sensitive to evidence (e.g., patient signs and symptoms) particularly when evaluating and

revising solution possibilities. Such a process of thinking considers options and reasons

(intelligent search) before rendering judgment or commencing action (intelligent

hesitation), because some impulsive tendencies or reactions "need constant regulation" to

obviate premature closure (Dewey, 1933/1998, p. 23). Moreover, such a disposition

fosters prospective awareness, which can be viewed as Promethean (wise before events).

"To be genuinely thoughtful, we must be willing to sustain and protract the state of doubt

which is the stimulus through inquiry, so as not to accept an idea or make positive

assertion of a belief until justifying reasons have been found" (Dewey, p. 16).

According to the principles of situated learning, instructional methods that

employ repeated exposure to problems increase the representational richness of problems,

which increases the learner's ability to identify and interpret affordances, and understand

problems (Barrows, 1986, 2000; Greeno, 1998; Resnick, 1989, 1991) as they transition

from concept novice to concept expert. Iterations are important because students will

acquire some understanding from original learning situations, but will miss much of the

detail until later exposures (Resnick). Thus, each iteration cycle provides new insight,

and increases opportunities for reflection and understanding (meaningful abstractions).

For Regan-Smith and associates (1994), abstractions allow medical knowledge to be

remembered and retrieved. Further, abstraction of contextual learning is thought to be

important vis-a-vis knowledge transfer (Dewey, 1933/1998; Regan-Smith et al.; Salomon

& Perkins, 1988).

Salomon and Perkins (1988) have characterized transfer as "low-road transfer"

and "high road transfer." Low-road transfer involves automated (no requirement for

reflection) transfer of frequently practiced skills, while high-road transfer requires the

abstraction of knowledge and intentional translation to accommodate the demands of a

novel situation. According to Salomon and Perkins, high-road transfer is fostered best

when instructional conditions encourage students to engage in "mindful abstractions," for

example, intentional abstraction of a principle after successfully applying the principle in

varied situations. Thus, students acquire knowledge at once linked to situations of

deployment and stripped of context. In other words, learning is from the particular to the

general rather than vice versa. From this view, learning is an inductive process used to

construct theories of use, or induction into practice (Sch6n, 1987).

Although considered a pragmatist, Dewey (1933/1998) viewed abstract thinking

as an essential element of a complete act of thought. For example Dewey wrote, "When

thinking is used as a means to some end, good or value beyond itself, it is concrete; when

it is employed simply as a means to more thinking, it is abstract" (p. 223). Apparently,

for Dewey, knowledge constructed by reflective inquiry is potentially certain. Thus,

ideas can be abstracted from particulars of experience and become relatively independent

of their context through a process of inductive reasoning. In further support of Dewey's

position advocating purposeful or pragmatic abstraction, Salomon and Perkins (1988)

have called reflection "high road to transfer."

Theoretical support of this position is emerging from what previously were

diametrically opposed cognitive (information-processing theory) and situated learning

perspectives (Anderson, Greeno, Reder, & Simon, 2000; Cobb & Bowers, 1999).

Whereas traditional cognitive theorists focus on the individual as a unit of analysis,

situated learning perspectives focus on interactive systems that incorporate individuals as

participants, interacting with each other, as well as with the physical characteristics of the

learning environment or problem-space (e.g., materials and representational systems)

(Cobb & Bowers). The convergence of these two perspectives appears sine qua non to

move beyond the dichotomy in medical education between knowing and doing and

understanding the successful integration of biomedical science and clinical practice,

which represents the utility of this notion of dual acquisition of knowledge. Presumably,

cognitive flexibility and problem-solving skills transfer will be facilitated by designing

performance challenges that require similar declarative and procedural knowledge in

varied problem-situations (Anderson et al.; Cobb, 2002; Cobb & Bowers; Perkins &

Salomon, 1989).

Problem efficacy. Problem effectiveness is operationalized as the degree of

congruence between student generated learning issues and instructional objectives

(Mpofu, Das, Murdich, & Lanphear, 1997). Research outcomes reveal varying

magnitudes of correspondence between instructional objectives and student-generated

learning issues prompted by posed health care problem (Dolmans, Gijselaers, Schmidt, &

van der Meer, 1993; Mpofu et al., 1997; Wendelberger, Simpson, & Biernat, 1996). In a

randomized study of second-year medical students undergoing a six-week course in

normal pregnancy, parturition, and child development in a PBL curriculum, Dolmans and

colleagues observed problem effectiveness varied from 28% to 100%, with an average

effectiveness of 64%. Interestingly, 6.2% of learning issues generated were unexpected,

and 47% of those issues were rated by faculty to be either moderately relevant or relevant

to objectives. Extending this line of research, Wendelberger et al. examined the

correlation between student generated learning issues and PBL curriculum objectives and

observed better than 90% agreement. Similarly, Mpofu et al. observed 55 to 100% (mean

91.4%) congruence between tutorial group generated learning issues and instructional

objectives. Quantifying problem effectiveness affords opportunities to identify problems

that inadequately stimulate learning issues.

The importance of problem quality was quantified further by van Berkel and

Schmidt (2000) who surveyed 1350 health sciences students enrolled in 120 courses over

six years. Data were collected regarding prior knowledge of students, quality of

problems posed, tutorial group effectiveness, tutorial group attendance, self-study time,

interest in subject content, and academic achievement. Using intercorrelations and path

analysis, investigators revealed problem quality was influenced more by prior knowledge

(.61) than tutor performance (.39). The effectiveness of a tutorial group was influenced

more by problem quality (.45) than tutor performance (.37). Tutorial group attendance

was effected directly by tutorial-group functioning (.48) and inversely by amount of prior

knowledge (-.31) or quality of problems (-.19). Tutorial group attendance was greater in

students who needed to know about the content related to the problem or who perceived

the quality of the problem to be wanting. Moreover, an inverse relationship between

group attendance and self-study time suggests that students compensate between time

allocated to collaboration and individual study. Intrinsic interest in content was

influenced more by problem quality (.68) than tutorial group effectiveness (.32).

Resultant achievement was influenced by group attendance (.36), quality of problem (.32)

and self-study time (.14). These results extend and confirm previous work quantifying

important inter-relationships and the multivariate nature of problem-based learning

(Schmidt & Gijselaers, 1990; van Til, van der Vleuten, & van Berkel, 1997) and are

discussed later.

Summary. Congruent with the stance of some social constructivists (Savery &

Duffy, 1995) and situated learning theorists (Brown et al., 1989; Greeno, 1998; Lave &

Wenger, 1991; Perkins, 1986), interactions between students and multiple features of

complex learning environment created by problem-oriented inquiry co-influences student

activity, motivation and achievement. In addition, these observations align well with

adult learning theory (Knowles, 1975, 1984) that posits adults are motivated by learning

perceived as focused on relevant problems (e.g., face validity), pj, ;1 i..r, i active

evolvement, designed so that students have responsibility for their learning, immediate

applicability to practice (e.g., concurrent applicability), and based on cycles of action and

reflection (e.g., reiterative or recursive cognitive processing). Problem-based instruction

capitalizes on these values, which promote learner-centered and problem-oriented

approaches to learning. The aim is to develop physicians equipped with adult learning

skills that enable them to adapt to varied and uncertain practice situations (Dewey,

1933/1998; Schon, 1987).

Role of Students

"Curriculum reform, if it is to have any chance of success, must begin with the

students. (Nelms, 1993, p. 94). In accordance with this view, students undergoing

problem-based instruction are required to assume responsibility for managing their

learning. Unfortunately, most medical students have spent their previous academic years

assuming the teacher was the principal disseminator of knowledge (Norman & Schmidt,

1992). This type of educational process can be characterized as one in which students are

spoon-fed (Dolmans, Schmidt, & Gjiselears, 1995). Research suggests this lack of

independence, reliance on the expertise of the instructor, and rote memorization of facts,

may cause many new PBL students to experience significant academic stress and anxiety

of varied etiology (Bernstein et al., 1995; Dyke, Jamrozik, & Plant, 2001; Gijselaers,

1996; Lieberman, Stroup-Benham, Peel, & Camp, 1997; Moore, Block, Style, &

Mitchell, 1994; Moore-West, Harrington, Mennin, Kaufman, & Skipper, 1989; Moore-

West & O'Donnell, 1985; Norman & Schmidt, 1992).

Cognitive dissonance, Research results are mixed regarding how students cope

with cognitive conflict in problem-based learning environments. Moore-West and

O'Donnell (1985) conducted a quasi-experimental, longitudinal study comparing learning

environment anxiety in students undergoing PBL versus LBL. They employed a

questionnaire with established validity and reliability to measure different dimensions of

anxiety at entrance to medical school and again two-years later. Despite equivalent

anxiety scores at entrance to medical school and controlling for Medical College

Admission Test scores, gender and ethnicity, mean anxiety level changed little in PBL

students (M = 16.41, SD = 8.41), but the level was significantly greater in LBL students

(M = 27.57, SD =15.61, t = 4.32, p < .0001). In a follow-up study of similar design,

Moore-West et al. (1989) confirmed and extended earlier results and concluded a

curriculum incorporating PBL versus LBL manifests less academic stress associated with

mastering an enormous knowledge and skills base while more effectively preparing

students to manage stress situations similar to those encountered by practicing physicians.

Additional support for the differential effect of PBL and LBL on student

perception of academic stress is provided by Lieberman et al. (1997). They conducted a

quasi-experimental study (pre-post test design) employing valid and reliable equivalent

forms of the Medical School Learning Environment Survey (Feletti & Clark 1981a,

198 Ib) and revealed significantly less stress and _... ii, l, ,...r, positive attitude

toward learning in similarly matched first-year PBL versus LBL students undergoing

simultaneous parallel curricula in the same institution.

Academically more mature medical students were studied by Bernstein and

colleagues (1995) who reported second year students undergoing transition from LBL to

PBL initially expressed concern that PBL would result in I ,... .e.. -c gaps, reinforce the

wrong information, and make inefficient use of valuable time" (p. 245). In fact, after the

first PBL sessions (5-weeks), these students exhibited a shift in perceptions from 38% to

52% agreeing PBL was more relevant and effective than LBL (chi squared = 6.5, p<


Further insights were provided by Moore et al. (1994) who used the Cognitive

Behavior Survey (Mitchell, 1994) to obtain perceptions in second year students who had

been assigned randomly to either PBL or LBL at Harvard Medical School. The

Cognitive Behavior Survey assesses medical student learning in preclinical years

incorporating 115 Likert-scale, open-ended, fill-in, and rank-order items with subscales

including learning behavior, nature of learning, and epistemological stance. Chronbach's

alpha for the subscales was .82, .79, and .76, respectively. Problem-based learning

students particularly were anxious regarding "what and how much to study" (p. 988),

which adversely effected the dynamics of small group tutorial sessions, a key component

of the PBL process. Approaching graduation, these students again were surveyed and

asked to select key words that described their preclinical experience. "Stressful,

engaging, and difficult" and "nonrelevant, passive, and boring" were descriptors cited by

PBL and LBL students, respectively, statistically more often (p_< .05). Although

instructionally stressful for students, authors believed PBL experiences presented

challenges comparable to those faced in clinical practice, thus contributing to the

authenticity of the learning situation. Moreover, tolerance of ambiguity was significantly

greater in PBL versus LBL students (p = 04). In addition, students undergoing PBL

scored higher on a subscale assessing reflection (p < .001) and lower on a subscale

assessing memorization (p = .01).

Dyke and associates (2001) in a randomized trial found PBL students initially

expressed anxiety regarding self-directed learning and engaging in responsible learning

activities with peers (detailed results and analysis presented in Chapter 4). However, as

students gained experience with PBL, anxiety waned, and on average, nearly vanished by

the end of the course. Moreover, PBL students' positive perception of their tutorial-

group's commitment to learning increased progressively through the course. Students,

however, frequently noted that problem-based instruction was a very time-consuming

style of learning.

Active learning. For cognitive constructivists, "cognitive disequilibrium"

(Piaget, 1969) or "cognitive dissonance" (Savery & Duffy, 1995) is required to motivate

self-orchestrated, active learning. Coherent with these beliefs, problem-based instruction

is designed to create what Dewey (1933/1998) called a "problematic situation" (p. 102)

and invites students to engage problems in a process mirroring reflective inquiry (Savery

& Duffy). For Dewey, reflective inquiry is a scientific method in which problems are

identified, problems are studied via active engagement, and conclusions (warranted

assertions) are reached, as problems are resolved tentatively. That is, problem solutions

are tentative until better information becomes available. Dewey (1929b) was committed

to employing science as a method for solving problems and learning.

Command of scientific methods and systemized subject matter liberates
individuals; it enables them to see new problems, devise new procedures, and, in
general, makes for diversification rather than for set uniformity. But at the same
time these diversifications have a cumulative effect in an advance shared by all
workers in the field. (pp. 12-13)

In addition, reflective inquiry allows practitioners to know possible problems and

their solutions in advance to anticipate and influence their outcomes. Dewey

(1933/1998) described this habit as "pre-reflective" (p. 106). Prospective awareness, or a

Promethean disposition is more likely to cultivate a practitioner who interprets transactss)

situations rather than reacts to situations (Dewey). Instruction embracing this philosophy

would be designed to facilitate methods of cognition, not didactics, because methods of

cognition can be employed to improve situations involving uncertain outcomes, limited

guidance, limited contextual knowledge, and the need to consider multiple perspectives

(Dewey, 1916; Spiro, Feltovitch, Jacobson, & Coulson, 1992). Indeed, according to Karl

Pearson, "The true aim of the teacher should be to impart an appreciation of method

rather than a knowledge of facts" (cited in Pickering, 1956, p. 116). Such a pedagogy

strongly parallels scientific method and reflective inquiry (Dewey, 1933/1998) and

should increase situational competence, particularly in indeterminate clinical problem-

spaces (Sch6n, 1987), which is a feature of PBL (Barrows, 1986). Dewey (1916) may

have described this method best:

What is meant by calling a method intellectual? Take the case of a physician. No
mode of behavior more imperiously demands knowledge of established modes of
diagnosis and treatment than does his. But after all, cases are like, not identical.
To be used intelligently, existing practices, however authorized they may be, have
to be adapted to the exigencies of particular cases. Accordingly, recognized
procedures indicate to the physician what inquiries to set on foot for himself, what
measures to try. They are standpoints from which to carry on investigations; they
economize a survey of the features of the particular case by suggesting the things
to be especially looked into. The physician's own personal attitudes, his own
ways (individual methods) of dealing with the situation in which he is concerned,
are not subordinated to the general principles of procedure, but are facilitated and
directed by the latter. (p. 201-201)

Reflective inquiry appears to be a suitable methodological framework to guide a

student's role (operational conduct) in a problem-based learning environment. Reflective

inquiry is a strategy for knowing that incorporates a process for analyzing, comparing,

and contrasting concepts and their implications for action, or practice. Reflective inquiry

is both a means for adapting and a tool to generate "warranted assertions." Thus,

Dewey's theory of inquiry involves thought intertwined with action to complete an act of

thought. That is, reflective experience, which continues from doubt and

indeterminateness to resolution of doubt, to generation of new doubt (Dewey, 1948,

1933/1998). This method of thinking involves a reiterative process of clarification,

testing, and refinement.

For Dewey (1933/1998), uncertainty is not confined to the mind, but involves

situation because, if doubt were only in the mind, cognitive processes alone would

resolve uncertainty. Reflective inquiry commences with situations that are problematic

and obstruct meaningful action. The inquirer frames the problem and formulates a

strategy of action. Consequently, the inquirer is immersed in, and in transaction with, a

problematic situation, which is a perspective held by situational learning theorists (Brown

et al., 1989; Greeno, 1998). Personal actions change elements of the situation. That is,

actions can open or close opportunities, rendering subsequent action more or less

plausible, respectfully. Therefore, the consequences of action can be either resources or

obstacles in the acted in and upon situation (Sch6n, 1983). Consequence (end) of action

(means) becomes the resource (means) or an obstacle in changed situation. Quality or

satisfactoriness of "end" determines its potential as a subsequent "means" in facilitating a

repertoire of new possibilities. For Dewey, an "end" not envisioned as a "means" in

facilitating new possibilities was ostensibly a cul-de-sac, or obviated a complete act of


Reflective inquiry is not just a useful tool, but also an organizing principle for

processing information and constructing and re-constructing knowledge (i.e.,

operationalized .....i.c a As indicated earlier, Salomon and Perkins (1988) claim

that reflection is essential for high-road to transfer. Indeed, the evidence suggests

successful learners and problem-solvers employ "deep processing" (Coles, 1985, 1990,

1997; Craik & Lockhart, 1972; Newble & Entwistle, 1986; van Til et al., 1997), a method

approximating reflective inquiry. When students employ deep cognitive processing to

attain understanding, their behavior is characterized by the logical, systematic

examination of evidence and interaction of evidence (Newble & Entwistle). Deep

processing involves critical thinking rather than immediate, uncritical assent to certainty,

which forces premature closure and entrance into an intellectual cul-de-sac (Dewey,

1933/1998). Students employing deep cognitive processing scrutinize contradictory

information, attempting to understand plausible alternative explanations, elaborating

interrelationships between evidence, competing explanations, and considering the fullest

range of evidence (Newble & Entwistle).

Student activity, van Til and others (1997) further clarified the roles of students

engaged in problem-based learning. They distinguished two dimensions of student

behavior characterized as (a) a function of learning style, using the levels of processing

framework (Craik & Lockhart, 1972), and (b) activity in students, as active or passive

participants during tutorial-group meetings. Combining the anchors of these two

dimensions--activity of PBL behavior and style of PBL behavior (surface cognitive

processing or deep cognitive processing)--four behavior combinations were generated

including deep-active, surface-active, deep-passive, and surface-passive. The authors

observed in second (n = 46), third (n = 30), and fourth (n = 57) year medical students

differential achievement on a written progress test as a function of activity and style

dimensions of PBL behavior. Inter-group scores regressed on year of training dimension

were statistically similar, thus the data were pooled for summary. Highest achievement

scores were obtained by students exhibiting deep-active behavior (M = 62.6, SD = 18.0)

and lowest scores were observed in students with surface-passive ',..J. i..r ',1 =52.1, SD

= 15.5) (Effect size = 0.63). However, regression analysis of PBL behavior dimensions

revealed that only the activity dimension was .ILI|"i.-,. 1 .. associated with achievement

(R2 = .28, p < .05). The differential effect of each dimension was most pronounced in

second-year and least in fourth-year students. These results imply that student

achievement is influenced significantly by activity patterns during PBL. Moreover,

formatively tracking these dimensions may inform educators about instructional effects

on the learning process. Implications of this research are important because they support

the argument that learning outcomes research is fircri,,,, ...I .: I.. unless

associated strategies for learning are revealed (Dubin & Taveggia, 1968).

Interestingly, the methodology and results reported by van Til et al. (1997)

illustrate that investigating instructional intervention solely with achievement outcomes

unmasks only part of the learning process (Dubin & Taveggia, 1968). Further, ten Cate

(2001) suggests:

In educational outcome research, a research design with an educational method as
an independent variable and a curricular test as the independent variable, however
elegantly it is conducted and however pristine its methodological aspects, is
usually too simple. Actually, the test operates as a dependent variable and
independent variable at the same time; the test is part of the curriculum. (pp. 85-

These results complement research indicating that the principal cause of learning is

student activity, and not instructional approach (Dubin & Taveggia; ten Cate; van Til et


Active behavior and deep cognitive processing (reflection) appear appropriate

strategies to enhance a student's ability to transact efficiently and successfully their role

in problem-based teaching and learning. Moreover, deep processing also is an

operational feature of metacognition (Barrows, 2000). Flavel (1987) distinguished three

elements of metacognitive knowledge including knowledge about self, knowledge about

task, and strategic knowledge. Thus, reflection appears a fundamental feature of

metacognition, and a necessary condition for achieving intentional regulation of learning.

As previously suggested, reflective inquiry (Dewey, 1933/1998) is a method of

metacognition, that is, thinking about thinking. Thus, such a disposition seems tailored

for adoption by students to facilitate the development of strategies for effectively

executing their role in PBL. Moreover, from a cognitive and instructional theory

perspective, an effective inquiry method and comprehension monitoring are desired

learning processes (Savery & Duffy, 1995) and play an important role for medical

students in coming to understand professional practice (Schon, 1987).

Dewey considered reflective inquiry instrumental to obviating dichotomy between

learning and doing and providing a method to enable intelligent action. "The value of

any cognitive conclusion depends upon the method by which it is reached, so the

perfecting of method, the perfecting of intelligence, is the thing of supreme value"

(Dewey, 1929a, p. 200). An aim of medical education is to enable medical students and

physicians to analyze information, to improve their problem-solving skills, and to reflect

their learning process (Liaison Committee on Medical Education, 2002).

Summary. While self-directed and collaborative learning responsibilities can

precipitate anxiety, principles of cognitive psychology suggest anxiety can promote

greater effort and enhance performance, both collectively and individually B.i,,dr j,

1986, 1997). Moreover, if adopted, Dewey's method of reflective inquiry can equip

students with a psychological tool to search efficiently and effectively for goodness-of-fit

between problem-representation and contextual data. For Federman (1999):

Implicit in PBL and the tutorial process is an awesome respect for the beginning
student. The recognition that one is contributing to the learning of the whole
group can be by turns inspiring, threatening, and overwhelming. Given the
underlying insecurity of most medical students, this regard for their potential is
enormously positive. (p. 93)

Knowing how to manage the paradoxical tension between inducing cognitive dissonance

while providing cognitive scaffolding is essential for efficacious tutoring.

Role of Small Group Tutorial

Rogoff (1991) describes a collaborative model of learning as "guided

participation" where students are active participants with guidance from more

knowledgeable others. Through mutual engagement, learners and their peers manage

activity with guidance from a tutor. This form of participation shares many

characteristics of cognitive apprenticeships described by Brown and colleagues (1989)

including communication (modeling), bridging student's current understanding and skills

with new ones (coaching), where problem solving is structured so that students are

supported and challenged in the tasks they undertake (through the interaction). Key to

this view of guided participation is the titrated transfer of responsibility as skills are

developed and deployed by students. A salient distinction in the structure of lecture-

based versus problem-based instruction is the amount of time scheduled for formal,

group-based learning (Richards & Cariaga, 1993).

Principles of constructivism and situated learning predict collaborative discussion

will facilitate the elaboration of prior knowledge and promote construction of richer more

meaningful knowledge representations of problem challenges and associated information.

In particular, collaborative learning assumes knowledge is situational and socially

negotiated (Vygotsky, 1926/1997), rather than individually constructed, by communities

of learners (Brown & Campione, 1990) and that generation and testing of information

and ideas are a process in which anyone can participate (Brown et al., 1989; Lave &

Wenger, 1991). Collaboration distributes the cognitive load among group members and

permits group en toto (collective knowledge) to engage problems that demand knowledge

and skills (e.g., learning issues) exceeding the immediate ability of individuals (Brown &

Campione; Gardner & Hatch, 1989; Stage et al., 1998). Informed by principles of

differential psychology (Knuth & Cunningham, 1993; Ryan, 1999; Sch6n, 1987), the

concept of distributed cognition recognizes not every member of a tutorial group will be

ready to learn material simultaneously (Brown, 1994; Brown & Campione).

Appreciating the varied intellectual standpoints and methods governing the dynamics of

students in a learning environment has important implications for medical teaching and


Capitalizing on personal differences and perspectives, collaboration enlists the

collective expertise of the group, as learning issues are distributed among members, who

through self-directed learning become quasi-experts in their assigned learning issues

(Brown, 1994). Encouraging students to become experts, or "major" in varied aspects of

content provides flexibility with respect to individual learner development (Brown, p. 7).

Collaboration also fosters idea sharing and encourages individuals to consider and

coordinate varied points of view, thereby facilitating reasoning and higher -..,k. Il,-i. i,,.

(e.g., analysis, syntheses, and evaluation) and promoting elaborated knowledge

construction (Bloom, 1984; Resnick, 1989, 1991; Ryan, 1999; Schmidt, 1983).

Ryan (1999) posits the cognitive constructivists' Principle of Multiplicity as a key

framework for understanding collaborative knowledge construction during problem-

based instruction. For Ryan, the Principle of Multiplicity accentuates the importance of

dialogue among tutorial group members. Different from argumentation, dialogue is a

process used to facilitate clarity and understanding. Dialogue is not simply a

communication technique, but a principle grounded in the premise that problem

perception, analysis, and resolution is coupled inextricably with a core of common

understanding (Dewey, 1933/1998; Schdn, 1987).

Insight for dialogue is provided by Dewey (1933/1998) who, as discussed

previously, advocated reflective inquiry as a method for enabling students to suspend

their assumptions and the process of responding reflexively or habitually. Thus, students

have to know how and when to wake up their patience and be sensitive to relevant

evidence. This requires practice. For Dewey, reflective inquiry is a mode of conduct for

collaborative problem solving. Initial response is arrested 1iI ih1 ,-,1 hesitation) to

assess and clarify the problem confronting the group. For collaborators to achieve

correspondence, varied points of view must be amalgamated (Lave & Wenger, 1991).

This "attitude represents what Mr. Pierce has happily termed the 'laboratory habit of

mind'" (Cited in Dewey, 1916, p.306).

Group activity provides students with the opportunity for self-reflection and joint

construction of knowledge (Brown & Campione, 1990; Brown & Palinscar, 1989; Lave

& Wenger, 1991). Theoretically, individual reflective inquiry can enhance the power of

the group to achieve the desired results. Thus, knowledge activation and elaboration,

dialogue with other students, and refining knowledge germane to the problem at hand, are

not only preparatory steps for individual learning, they also are instrumental in a

cooperative type of learning that fosters restructuring of knowledge or conceptual change

(Bransford, Brown, & Cocking, 1999; Brown, 1994; Brown et al., 1989; Schdn, 1987).

Moreover, this democratic model contrasts starkly with traditional teacher-centered

learning because it prescribes a shared process of educational dialogue, negotiation,

compromise, and consensus (Dewey, 1916).

Evidence for collaboration-induced elaboration in college science students is

provided by Schmidt et al. (1989). Randomly assigned students in small groups were

presented with either a problem related to or not related to a science text subsequently to

be studied. For a specified time, students discussed the problem, then studied the text

that provided novel information. Students participating in relevant problem discussion

before studying the science text exhibited significantly better free-recall, explanations

and descriptions (M = 27.2, SD = 9.1, n = 20 versus M = 19.7, SD = 7.5, n = 19, p= .008,

Effect size = 0.90). Thus, prospective group discussion of a problem matched with

incomplete prior knowledge appears an appropriate enabling condition for fostering

subsequent knowledge construction.

The achievement effect of collaborative learning also has been studied using

medical students. Martenson, Eriksson, and Ingelman-Sundberg (1985) designed a

medical chemistry course to sustain traditional content but using a modified instructional

method to induce students to obtain primary facts themselves, to stimulate problem

solving in small group settings, to foster understanding by asking students to integrate

chemistry and clinical practice, and to foster responsibility for self-directed learning. The

results of their instructional intervention were compared to a historical control, that is,

achievement exhibited by students completing the course in lecture-based format with

similarly structured assessments. Mean essay questions achievement was 72.7% and

60.7% in students undergoing problem-oriented, collaborative (11-cohorts, n = 1651)

versus traditional, lecture-based learning (5-cohorts, n = 818), respectively. This is an

effect size of 0.26 with a 95% confidence interval of 0.18 to 0.34. Mean short answer

questions achievement was 72.2% and 76.7% in students undergoing problem-oriented,

collaborative (11-cohorts, n = 1651) versus traditional, lecture-based learning (5-cohorts,

n = 818), respectively. This is an effect size of-0.10 with a 95% confidence interval of-

0.18 to -0.02. Mean final score for the course was 71.8% and 68.2% in students

undergoing problem-oriented, collaborative versus traditional, lecture-based learning,

respectively. This is an effect size of 0.08 with a 95% confidence interval of 0.00 to 0.16.

Results are limited by validity threats including unsubstantiated baseline equivalence and

unknown temporal effects associated with using a historical control. However,

significantly better essay question scores for students undergoing collaborative group-

based versus lecture-based learning is consistent with some views of situated learning

(Brown et al., 1989; Savery & Duffy, 1995) suggesting active, socially mediated learning

facilitates deep processing and concept elaboration.

Further evidence is provided by Lawry, Schuldt, Kreiter, Densen, and Albanese

(1999) who conducted a randomized investigation in second-year medical students and

physician assistants. Instruction in methods to perform a musculoskeletal examination

(physical examination) was completed in four formats including written materials only,

written materials and videotape, written materials and small-group sessions facilitated by

senior medical students, and all three methods. Facilitators were trained by one member

of the faculty. Written materials described musculoskeletal problems (e.g., deformity,

muscle atrophy) and the video presented examples of the problems and corresponding

physical examination procedures. Achievement was quantified using a written test and

objective standardized physical examinations before instructional sessions, and then

seven to ten days, three months, and 16 months after instruction. Instructional validity

was confirmed by an improvement of 2.5 and 5.0 standard deviations in mean written and

standardized patient examinations, respectively, between the mean pre-course scores and

all students independent of instructional group at seven to ten days. Intergroup written

examination achievement was statistically similar. However, students undergoing small-

group instruction exhibited significantly better patient examination skills compared to

those who underwent written material or videotape instruction at each of the examination

periods. The addition of small-group sessions to written materials affected standardized

patient examination scores 1.8 standard deviations greater than the mean score observed

for written materials alone and 1.0 standard deviation above value achieved for written

material and videotape instruction. The authors concluded small-group instruction with

hands-on supervised practice is superior to passive instruction for teaching

musculoskeletal examination skills, and this superior advantage is sustained for at least

sixteen months. Taken together, these findings and those of Martenson and associates

(1985), illustrate how varied achievement measures can provide varied interpretation of

instructional interventions; this is discussed later.

Boehler et al. (2001) designed a study to test the hypothesis that students utilizing

collaborative study methods exhibit greater achievement. Third-year medical students

were surveyed to quantify reading habits, utilization of lectures, group study, study

patterns, and resources employed during five consecutive ten week surgery clerkships.

Assessments during the clerkship included the surgery subtest ofNBME (NBME-II-S)

and the multiple stations clinical examination (MSCE). The MSCE is a nineteen station

assessment employing standardized patients and interpretation of clinical information

situated in the framework of clinical cases emphasizing problem solving. Group study

was selected by 18 (22%) and independent study by 63 (78%) of the participants. Groups

comprised two or three students and convened two to five (55.5%), six to ten (16.7%), or

11 to 22 (27.8%) times per month. Students from first and second year PBL track

comprised 33.3% of participants. Surprisingly, of twenty-seven students who underwent

reiterative, tutorial-group PBL, only eight (30%) chose collaborative learning. Of fifty-

four previous LBL students, forty-four selected individual study (81%) and ten (19%)

selected collaborative learning. Indeed, there was no significant relationship between

whether students underwent pre-clerkship PBL or LBL and their study patterns and use

of resources, except PBL students reported more frequent Internet access (p = .05).

However, investigation revealed superior MSCE achievement for students who studied in

groups (p = .001). Collaborative learning afforded no advantage for performance on

NBME-I-S. Since a primary goal of PBL is to foster appreciation for collaboration in

practice, educational implications of this research, if corroborated, are significant.

Possibly more important for medical education in general, results of this research suggest

collaborative learning enhances clinical reasoning competence, independent of previous

collaborative learning experience.

De Grave, Boshuizen, and Schmidt (1996) studied cognitive and metacognitive

processes during problem analysis by examining conversations (verbal protocols)

between group members and their thinking processes. Thinking processes were revealed

through stimulated recall methods (recall protocols). Analysis of verbal protocols

revealed that considerable time was expended for data exploration and problem

formulation coupled with a dominant tendency to construct theories, generate causal

relationships, and test ,i. r.. I6. ,. Investigators concluded students in a problem-based

collaborative process undergo alternating periods of theory construction and data

exploration. Moreover, investigators provide evidence for conceptual change resulting

from collaborative interactions. Apparently students critically appraise information

proposed and are persuaded by arguments advanced by other group members, resulting in

conceptual change. Important observations reported but not discussed by the authors,

indicate students in collaboration tend to employ abductive reasoning (generation of

tentative hypotheses from best available data) and inductive reasoning (testing tentative

hypotheses with current data and data accrued) rather than hypothetico-deductive

reasoning (Chamberlin, 1890/1965). This matches with the authors' description of

collaborative process as theory construction, and method of inquiry employing multiple

working hypotheses (Chamberlin) and reflective inquiry (Dewey, 1896). For


.. under the working hypothesis, the facts are sought for the purpose of ultimate
induction and demonstration, the hypothesis being but a means for the more ready
development of facts and of their relationship, and the arrangement and
preservation of material for the final induction. (p. 755)

This way of knowing affords students the opportunity to consider equally all evidence

and to accept the interpretation most suggested by the evidence, or a "warranted

assertion" pending better data (Dewey). Implications for medical education of this

method of thinking and learning are important because participating in such a process

acknowledges that solutions are tentative. This process might facilitate dialogue and

promote effective collaboration.

Clearly, collaborative learning affords the potential to engage medical students in

activities that are valuable in the process of concept learning, and developing a

disposition toward intelligent inquiry. It is not guaranteed, however, that such high-

quality dialogue will transpire in a collaborative learning environment. Medical student

collaboration can be characterized by competition, asymmetrical participation, and the

focus can be on finishing a task instead of understanding concepts (Steele, Medder, &

Turner, 2000). For example, Steele et al. observed that peer-tutored groups "take

shortcuts in the PBL process that may undermine some of the intended goals of PBL" (p.

23). Consequently, the learning process must be verified to facilitate the valid

interpretation of outcomes. This is discussed later.

Tutor characteristics and tutoring responsibilities. Ingredients hypothesized

as important to fostering collaborative learning during problem-based instruction include

authenticity of the problem, the role of the student, and activity of the tutor (Dolmans,

Wolfhagen, Schmidt, & van der Vleuten, 1998; Schmidt & Moust, 1995). For teachers,

problem-based instruction represents a pedagogical shift from a transmission-oriented

perspective, in which the emphasis is on instructors and what they teach (instructivism),

to an interaction-oriented perspective, in which the emphasis is on students, and how they

build knowledge constructivistm) (Brown & Campione, 1990). These concepts are

relevant to medical education in general and problem-based learning in particular, and are

expressed in the following passage by Dewey (1916):

The alternative to furnishing ready-made subject matter and listening to the
accuracy with which it is reproduced is not quiescence, but participation, sharing

in an activity. Inside shared activity the teacher is the learner, and the learner is,
without knowing it, teacher. (p. 160)

Extending the Dewey tradition, a tutorial framework for problem-based teaching

and learning theoretically renders each pupil both student and teacher (Federman, 1999).

This promotes group processing of information rather than an imparting of information

by a tutor (Vernon & Blake, 1993). Such a pedagogy converges with Dewey's (1974)

belief that a student cannot be taught what to know, but can be coached.

He has to see on his own behalf and in his own way the relations between means
and methods employed and results achieved. Nobody else can see for him, and he
can't see just by being 'told,' although the right kind of telling may guide his
seeing and thus help him see what he needs to see. (p. 151)

To this end, quite possibly a PBL tutor's greatest challenge is deciding when to

engage and when to disengage in direct participation in PBL group activities (Neville,

1998). For example, an important role of the tutor is to evaluate the metacognitive

competence of students (Norman & Schmidt, 1992). Tutors face the challenge of

adapting their level of engagement to their students' varied levels of metacognition in

ways that scaffold under-developed self-directed learning skills. Consequently,

metacognitive coaching is essential and metacognitively challenged students must be

identified and helped immediately (Brown, 1994; Brown et al., 1989), or as mentioned

earlier, academic frustration and anxiety may manifest.

As previously suggested, reflective inquiry may be a useful methodology for the

tutor. Reflective inquiry guides students away from impulsive thinking by encouraging

dialogue and structuring thinking about thinking (metacognition). In this way, the tutor

encourages metacognitive awareness to facilitate self-assessment to determine what a

student or group knows and does not know. The basic challenge for the tutor is to

develop in students an interpretive (interactive) rather than reactive disposition (Dewey,

1933/1998; Lindblom-Yl nne & Lonka, 1999) that enables them to perceive affordances

(situational cues) in complex, often chaotic and unpredictable clinical problem-spaces so

they can learn to practice competently on the fringe of surprise or in non-routine

situations (Sch6n, 1987).

Tutor activity. Savery and Duffy (1995) argue, "The most critical teaching

activity is in the questions the tutor asks the learner" (p. 33). This pedagogical approach

approximates the technique of recitation advocated by Dewey (1933/1998). Indeed,

Dewey indicated that the art of conducting a recitation rests primarily in the art of

questioning learners to develop in them independent disposition of inquiry "in both of its

directions; namely inquiry in'observation and recollection for the subject matter that is

pertinent and inquiry through reasoning into the meaning of material that is present" (p.

266). These questions aim to challenge the learner thinking and support intellectual

development via scaffolding and extending the zone of proximal development (ZPD) as

described by Vygotsky (1987, 1926/1997). Vygotsky (1987) described ZPD as "the

distance between actual developmental level as determined by independent problem

solving and the level of potential development as determined through problem solving

under adult guidance or in collaboration with more capable peers" (p. 86). Thus, for

individuals, psychological function appears twice, first on a social plane (interpsychical)

and then on a psychological plane (intrapsychical). For Vygotsky (1926/1997), an

individual's ZPD represents boundaries of knowing space, that is, the gulf between actual

knowledge or performance (e.g., independent problem-solving competence) and potential

knowledge or performance (e.g., guided problem-solving competence). Traditionally

these psychological functions (e.g., attention, memory, and cognition) have been treated

as properties of an individual's mind. The social dimension of psychological functioning

proposed by Vygotsky (1978) helps explain how collaborative learning occurs and

theoretically sponsors efficacy of collaborative learning. From this perspective,

scaffolding at once strives to proximate ZPD, while appropriately extending the upper

bound of ZPD to facilitate the personal construction of cognitive representations. A

formidable responsibility for tutors is to balance scaffolding while inducing cognitive

conflict to mediate knowledge construction. Such a pedagogical modus operandi

recognizes that knowledge is perpetually under construction with titrated or "demand

triggered" scaffolding via a tutor.

According to Bandura (1997), tutors serve as powerful role models. Initially the

tutor is a model, then a facilitator. As students develop expertise, the tutor transitions to a

monitoring role and engages only to insure quality of learning and attainment of learning

outcomes (Barrows, 1996, 2000). Thus, an essential initial responsibility of the tutor is to

establish with students a commitment to, and skills required for, evaluating their learning

and performance. Moreover, a tutor models and encourages students to evaluate the

group's learning and performance. Eventually, evaluation becomes part of what Dewey

called a complete act of learning (Dewey, 1933/1998), as students assist each other in

developing a disposition toward the critical consideration of the quality of all elements of

learning and its application.

Efficacy of tutoring. What are the qualities of an effective tutor? Bloom (1984)

regarded individual tutoring as the gold standard of education against which instructional

effectiveness can be judged. Summarizing data from two of his doctoral students'

research, Bloom observed one-to-one tutoring produced an effect size of 2.0.

Importantly, Bloom illustrated that most students undergoing one-to-one tutoring have

the potential to achieve this effect-size. Bloom provides little guidance for the process of

efficacious tutoring, particularly in a manner employed for PBL, that is, group tutorial.

In their literature review of PBL, Albanese and Mitchell (1993) reported that students

preferred to be tutored by experts. In particular, expert tutors were observed to augment

students' ability to leam via facilitating identification of relevant learning issues and by

assisting in correcting gaps in knowledge and errors in processing information.

Dolmans et al. (1998) revealed three elements of efficacious PBL tutoring

including guiding students through the learning process, content knowledge input and

commitment to the group's learning. Schmidt and Moust (1995) defined quintessential

dimensions of PBL tutoring to affect student learning as "social congruence" and

"cognitive congruence," which they posit are inextricably coupled. Cognitive

congruence is defined as a tutor's sensitivity (social congruence) to problem-induced

difficulties encountered by students. Wilkerson (1995) assessed ratings from first-year

medical students undergoing PBL at Harvard Medical School and found four distinct

attributes of the highest rated tutors. First, effective tutors did not control the group and

were able to balance student directions with assistance, only intervening when

appropriate to foster student focus vis-A-vis problem context and to facilitate learning

(situational awareness). Second, these tutors contributed their expertise to the group by

recommending resources and providing anecdotal clinical experiences. Third, these

tutors had the ability to create a pleasant learning environment via appropriately

responding to students. Fourth, effective tutors stimulated the critical evaluation of ideas

by encouraging students to think beyond the obvious and to examine a problem from

multiple perspectives to enhance their learning.

Extending Wilkerson's (1995) work, De Grave et al. (1999) argue "tutors are not

tabula rasa" (p. 902), thus personal theories and perspectives of teaching and learning

influence their instructional modus operandi. Despite equivocal pedagogical approaches

to group tutoring, there appears to be a relationship between a tutor's role and the quality

of learning adopted by students (Wilkerson). Facilitating through scaffolding appears to

be an essential role of the tutor (De Grave et al.), particularly because students are

saddled with independence and ownership of their learning, which as noted earlier, may

be anxiogenic and frustrating (Albanese & Mitchell, 1992; Barrows, 1986, 1990, 1996,

2000; Gijselaers, 1996; Kaufman & Mann, 1998; Moore et al., 1994; Norman & Schmidt,

1992; Vernon & Blake, 1992). Student competence in coping with independence and

ownership of their learning may vary (De Grave et al.). Thus, as mentioned previously,

individualized goal directed scaffolding might be an important aspect of efficacious

tutoring in PBL groups.

De Grave and colleagues (1999) wished to discover the behavior of effective and

less effective tutors. They employed a Tutor Intervention Profile, a valid and reliable

instrument which quantifies four-dimensions of tutor behavior. These include stimulates

elaboration, directs the learning process, stimulates integration, and stimulates interaction

and individual accountability. Not surprisingly, they observed varied styles of tutoring.

For instance, one pattern of tutor behavior was characterized by depending more on the

use of expert knowledge, compared to another behavior pattern that predominantly relied

on the ability to stimulate learning in a tutorial group. Tutors who rated highest on each

of the four-dimensions were perceived to be most effective by students. Moreover, tutors

who focused learning processes were perceived to be more effective than those who

focused content. De Grave et al. suggest viewing four dimensions of effective tutoring as

an operationalization of scaffolding. If scaffolding is considered an essential role of the

tutor, future research seems warranted to clarify the strengths and weakness of these

dimensions to facilitate faculty development, reveal misconceptions about teaching and

learning, and optimize the tutorial process.

Aligned with the perspectives of Vygotsky (1987, 1926/1997) and Bandura (1986,

1997), Lepper, Drake, and O'Donnell-Johnson (1997) discovered effective tutors equally

emphasize intra-psychical (cognitive), inter-psychical (social), and motivational factors.

Interaction of these factors is illustrated in the observed behavior of effective tutoring

(Lepper et al., p. 122).

1. Expert tutors have content knowledge and topic-specific pedagogical knowledge
to manage the difficulties learners encounter, and expert tutors have general
pedagogical knowledge.
2. Expert tutors exhibit high level affective support and nurturance when interacting
with students (similar to the concept of social congruence advocated by Schmidt
and Moust [1995] discussed below).
3. Expert tutors employ a Socratic style.
4. Expert tutors are committed to extending the intellectual envelope of each student
during each tutorial.
5. Expert tutors accomplish these activities via unimposing and indirect methods
(e.g., via strategic use of the Socratic style).
6. Expert tutors encourage reasoned and meaningful thinking (e.g., reflective
inquiry) to foster self-generated explanations.
7. Expert tutors are exceptionally committed to encouraging and motivating

How does level of tutor expertise affect achievement? Schmidt, van der Arend,

Moust, Kokx, and Boon (1993) studied 1,120 medical students in four curricular years,

152 tutors and 336 tutorials. Students in expert-tutor groups spent more time on self-

directed study and exhibited higher achievement scores. These findings were most

impressive for first-year students. A tutor's process-facilitation skills also were related to

achievement scores and both tutor's knowledge-related behaviors and group-dynamic

skills were correlated, indicating that both were essential elements of effective tutoring.

In a follow-up investigation, Schmidt and Moust (1995) observed that subject matter

expert tutors were able to employ more effective process facilitative behaviors such as

asking stimulating questions, offering counter examples, or seeking clarification, and that

these behaviors were related to achievement, as quantified by written test scores. They

concluded that to be effective, tutors must be good facilitators and content experts, with

content expertise a pre-condition of efficacious behavior.

Sobral (1994) compared faculty-tutored groups with peer-tutored groups in a PBL

medicine course at the University of Brasilia and observed no significant intergroup

differences in either problem solving success or student self-evaluation of skills.

Although, scores were greater significantly for meaningfulness of learning and usefulness

of group work in peer-tutored groups.

Dolmans, Wolfhagen, and Schmidt (1996) asked tutors to rate themselves as

experts and non-experts. Then they examined data from 119 tutors participating in 135

tutorial groups. Students were asked to rate their prior knowledge of clinical problems

and to rate structure of information presented with clinical problems. Problems were

examined for magnitude of structure. Student achievement was quantified using 150

true-false questions at the end of each curricular unit. Student achievement was

independent of tutor expertise and curricular structure.

In a similarly designed study using students as non-experts, Steele et al. (2000)

compared achievement in student-led and faculty-led problem-based learning.

Employing a prospective, Latin-square crossover design, they investigated 177-second-

year medical students participating in eleven problems over one-year. For each problem,

half of the students were led by students and the other half by faculty. Each problem

achievement was quantified as performance on an examination incorporating multiple-

choice and matching questions aligned with curricular objectives. No differential

achievement was observed between students who underwent faculty-led versus peer-led

problem-based learning. Data from student evaluations indicated they preferred peer

tutoring, but, these same data indicated, and faculty observation confirmed, instances

when peer-tutored groups "take shortcuts in the PBL process that may undermine some

of the intended goals of PBL" (p. 23). Apparently, for students collaborating in short cuts

to truncate problem-based learning, answers were more salient than the process of

inquiry. A limitation of this study is that all participating students had a year of faculty-

tutored PBL experience and the curriculum is a hybrid of PBL and LBL, thus students

may have viewed the processes of PBL as less important and ancillary to traditional

didactics (Steele et al.). This explanation is supported by previous research conducted by

Schmidt who revealed students entering medical school with versus students without

tutorial-group or self-directed learning experience exhibited differential study strategies.

This developmental nature of self-directed learning also is acknowledged in educational

psychology literature (Winne, 1995a, 1995b; Zimmerman, 1986). In addition, apparently

the depth and breadth of domain knowledge affects capability and the efficacy of self-

directed learning.

Schmidt (1994) compared medical student tutors, non-expert faculty tutors, and

expert faculty tutors as a function of medical student achievement. Students with low

prior knowledge in relation to unit content benefited from having expert faculty tutors.

Tutor expertise also was important when problem-situations were ill-structured, but was

independent of achievement when problems were more structured. Unexpectedly,

students with low prior knowledge in an ill-structured problem situation performed

reasonably well having undergone peer tutoring. Presumably, peer-tutors can effectively

facilitate information translation to ill-informed students. This observation however,

requires additional research to substantiate and clarify the implications for medical

education. Particularly regarding initial learning competence, the presence of expert

tutors diminishes the likelihood of misconceptions (Neville, 1998). However, Albanese

and Mitchell (1993) indicate an expert tutor can reduce student attempts at self-directed

learning because experts tend to be more directive during the problem-solving process.

This observation is substantiated by Schmidt and Moust (1995) who, using structural-

equation modeling, found a slightly negative (-.12) association between tutor expertise

and self-study time. This suggests an inverse relationship between a tutor's expertise

contributions to group discussion and time students spend in self-directed learning. A

similar inverse effect has been observed between tutorial group functioning and self-

directed learning (Schmidt & Gijselaers, 1990). Taken together, these reports indicate an

inverse relationship between tutorial group efficacy and required extra-tutorial studying.

Paradoxically, these results support the efficacy of tutoring and small-group learning, but

blunt, or obviate the aims of problem-based instruction to foster self-directed learning.

Li,,..,i... ill research literature comparing the advantages and disadvantages

associated with level of tutor expertise is contradictory. Establishing an appropriate level

of tutor expertise requires further research; but, rather than dichotomize issues, research

designed to examine the differential efficacy of tutor expertise vis-a-vis student

experience and targeted learning outcomes might be more fruitful. Possibly, expert tutors

and peer tutors provide uniquely effective roles in given contexts. For example, evidence

presented earlier suggests students with low prior knowledge challenged with ill-

structured problem-settings may be better served via peer tutoring (Schmidt, 1994).

However, given the importance of corrective I...... which requires some degree of

subject-area expertise, serious questions remain -..,,1nd 1. r-, use of content-experts as


Summary. Situated learning theorists emphasize the inherent benefits of working

together. Germane to medical education, collaborative learning establishes conditions

relevant to clinical practice, where students and physicians must provide explanations and

take advice. Theory and c .;..1 : 0..| :I ii information is best retained for relevant

retrieval in memory when it is linked to prior knowledge. This process is facilitated by

cognitive elaboration or reinterpretation and re-organization of material (e.g., recitation,

writing summaries, and explaining synthesis). Moreover, peer discussion is one means of

elaboration. In peer discussion, students interact with material and with each other,

generating more information processing than when interacting with material alone. In

addition, tutor scaffolding appears to influence significantly the quality of collaborative

learning. Problem-based learning environments immerse both students and tutors.

Continual use and reflection on the PBL processes broadens and deepens educators'

knowledge and ability to facilitate teaching. Optimal tutoring may be realized when

effective dimensions of scaffolding are operationalized (De Grave et al., 1999). These

include stimulating elaboration, directing the learning process, integration of knowledge,

and stimulating interaction and individual accountability. An interesting and potentially

useful paradox of PBL is that students learn to work collaboratively (small-group

tutorials) while they are developing as autonomous learners (self-directed learning), a

process that incorporates the educational pluralism advocated by Dewey (19 i' I

legitimate peripheral participation (Lave & Wenger, 1991), and the social genesis of

individual understanding advocated by Brown and Palinscar (1989). Indeed, PBL

appears to be a suitable framework for investigating the continuum of learning anchored

with individual learning endogenouss learning) and collaborative learning (exogenous

learning) (Moshman, 1982).

Role of Self-Directed Learning

Medical educators are challenged to provide potential physicians with the breadth

and depth of current medical knowledge and the skills required to use this knowledge,

while simultaneously aiming to produce physicians who can sustain and further develop

their competency over a career (Mann, 1994). Compounding this challenge to medical

educators is the efficacy of current medical knowledge as characterized in the statement

by Dean Sydney Burwell (Harvard Medical School) nearly a half-century ago, "My

students are dismayed when I say to them half of what you are taught as medical students

will in 10-years have been shown to be wrong. And the trouble is, none of your teachers

know which half' (cited in Pickering, 1956, p. 115). Consequently, medical educators

aim to foster in their students the motivation and capability to continue learning beyond

their formal training (Barrows, 2000).

Developing self-directed learning skills is sine qua non to becoming an expert

learner in medical practice (Barrows, 1986, 2000; Schon, 1987). Indeed, a hallmark of

intellectual competence is the ability to recognize personal knowledge gaps and endeavor

to close them (Vygotsky, 1987, 1926/1997). From a cognitive perspective, self-directed

learning is triggered when the learner recognizes that the knowledge demands of a

problem exceed personal knowledge. The perceived knowledge differential represents

potential stimulation to acquire requisite knowledge to address problem solution

(Zimmerman & LeBeau, 2000).

...ii r. ,,i.,i.,r. in learning is often related to meaningful learning, whereas

external-, or other-regulation is more likely associated with knowledge reproduction or

rote memorization (Bandura, 1997; Lindblom-Yltnne & Lonka, 1999; Regan-Smith et

al., 1994; Winne, 1995a, 1995b). Conceptually self-directed learning is very similar to

what is called self-regulated learning in educational psychology literature (Zimmerman &

LeBeau, 2000). Numerous characterizations of self-regulated learning permeate the

literature (Bandura; Winne; Zimmerman, 1990, 1998). For illustration, social cognitive

theory views self-regulation as incorporating three components including self-

observation, self-judgment, and self-reaction (Bandura). Bandura claims these processes

interact with one another and influence a sense of self-efficacy that is required to attain

learning goals. Related activities are choosing or adopting learning goals, and self-

evaluation. Self-observation refers to a willingness to become aware of one's behavior

and to cope with resulting emotions. Self-judgment calls for making a choice between

alternative evaluation standards. Self-reactions can motivate students by creating or

maintaining positive feelings and expectations as well as giving self-rewards.

Extending Bandura's (1997) position, Zimmerman and Lebeau (2000)

synthesized common features among varied perspectives of self-regulated learning

including student's selective use of learning strategies, responsiveness to feedback

regarding learning effectiveness, and self-initiated efforts to seeks out opportunities to

leam. For Zimmerman (1986), such learners are "metacognitively, motivationally, or

behaviorally active promoters of their academic achievement" (p. 308). From this

vantage, .* I. r.: ,ii. d learners employ three kinds of strategies to execute their learning

including metacognitive (e.g., i i-,v...,i..rhr self-evaluating), motivational (e.g., setting

goals, self-reinforcement, positive private talk), and behavioral (e.g., scheduling study

time, employing external resources).

Barrows (1986, 2000) argues that a potent stimulus to fostering metacognition is

to have medical students learn collaboratively via problem-oriented inquiry. Such a

situation presents an opportunity for a learner to discover what they do not know and in

what areas they must focus study. Thi... r, i, .ij. this process can foster a disposition to

identify learning issues, which in addition to fostering activation of prior knowledge and

elaboration of knowledge (Coles, 1985), is thought to stimulate the development of self-

directed learning and to promote lifelong learning (Barrows; Blumberg et al., 1990;

Walton & Mathews, 1989). Developing a pattern of lifelong learning is required to

sustain practice competence in the continually changing field of medicine (Bok, 1984;

Schdn, 1987; Tosteson, 1994). Indeed, because of the medical information explosion,

becoming an expert learner may be the most powerful outcome of formal medical

education. However, this outcome may not be adequately fostered in traditional, teacher-

centered medical curricula that emphasize rote memorization, or what is often described

as "learning for school" rather than "learning for life".

Lindblom-Ylinne and Lonka (1999, 2002) investigated learning practices ("study

orchestration," p. 122) in medical students enrolled in a traditional curriculum. Their

study revealed a traditional learning environment fostered misleading cues about what

and how to study efficiently and forced many students to study in inadequate ways,

driving them toward externally regulated and surface learning. Study orchestration is

conceptually defined as a "contextualized study approach adopted by individual students

or groups of students" (Meyer, 1991, p. 297, cited in Lindblom-Ylanne & Lonka). This

concept includes three dimensions of student learning: (a) the existence of qualitative

individual differences in method of engaging learning environment challenges; (b) the

influence of context on method of learning environment engagement; and (c) individual

differences in student perceptions of learning i'.. .. ri Thus, study orchestration reflects

the interaction between the learner and the learning environment and functionally

captures differential strategies employed by students as they allocate their cognitive

resources in a specific learning context. Operationally, study orchestration may be

characterized as a student exhibiting either conceptual consonance or conceptual

dissonance. Lindblom-Yliinne and Lonka employed a comprehension monitoring probe

(responses classified as knowledge criteria or comprehension / application criteria),

survey instrument (71 items concerning learning approach, regulation of learning, and

conception of learning) and semi-structured interviews concentrating on learning

strategies, study patterns, and perceptions of learning environment. Stratifying their

results, investigators categorized thirty-five senior medical students into four groups: (a)

meaning-oriented independent students (n = 4; 3 males, 1 female), (b) meaning-oriented

with novice-like conception of knowledge (n = 12; 4 males, 8 females), (c) reproduction-

oriented and application-directed students (n = 11; 4 males, 7 females), and (d)

reproduction-oriented and externally regulated students subdivided into two-groups

including coherent (n = 4 females) and dissonant (n = 4 females). Age (M = 25.3 years)

and gender distributions were statistically similar among groups. Corresponding with the

institution's medical student population, 69 percent of study participants were female.

Meaning-oriented independent students actively sought to construct meaning and

integrate information despite factually oriented assessment practices. They also exhibited

metacognitive awareness and conveyed concise goals for studying. Similarly, meaning-

oriented students with novice-like conception of knowledge focused on understanding

instead of memorization, and exhibited intentional study strategies. Reproductive-

oriented and application-directed students emphasized the importance of acquiring

knowledge and applying it later in clinical situations. These students, contrary to

meaning-oriented students, emphasized factual learning more than understanding.

Reproduction-oriented and externally regulated students were sub-grouped according to

coherence of their study orchestration, that is, conceptual consonance or dissonance. In

the conceptual consonance subgroup, students engaged in learning "in a 'purely'

superficial way" (p. 131) and were rigidly regulated by the demands of a traditional

examination process. Whereas, despite manifesting some metacognitive awareness,

students categorized in the dissonance sub-group experienced problems adapting to their

medical education in general, and the assessment process in particular. These students

viewed their experience more as coping with the learning environment, thus they

concentrated on memorizing facts, were unable to integrate information, and had

problems in meaningful learning. These outcomes clearly are undesirable and potentially

harmful (Whitehead, 1929). Importantly, the authors found meaning-oriented

independent students enjoyed above-average study success, whereas reproduction-

oriented and externally regulated students achieved the lowest grades. The authors

concluded a traditional medical curriculum creates an environment "in which students

who manage not to adapt accordingly and who do not become reproduction-directed will

do well" (p. 137). Fortunately, some students' study practices are immune to knowledge

reproduction-orientation fostered by teacher-centered, assessment processes in a

traditional curriculum (Lindblom-Yltinne & Lonka). Despite the relatively small sample

and idiosyncratic students (Finnish nationals), these data are comparable with other

reports (Miller, 1961; Neame, 1984; Regan-Smith et al., 1994; Schmidt, 1983; Schdn,


Taken together these findings suggest pedagogical methods substantially

influence learning strategies adopted by medical students. Evidence suggests traditional

lecture-based instruction requires teachers to be active and medical students to be

passive, thus sponsoring superficial learning. Medical students are not challenged to

exercise critical thinking, as they are only required to recall basic information. In

contrast, problem-oriented instruction requires students to exercise inquiry and p:. r ..,,,.ll.

construct meaningful knowledge, and requires tutors to be relatively passive while

encouraging responsibility for self-directed learning.

Manifestations of self-directed learning. What instructional conditions

facilitate self-directed learning? Problem-based learning theorists (Barrows, 1986, 2000;

Blumberg et al., 1990; Dolmans & Schmidt, 1994; Walton & Mathews, 1989) derive

insight for self-directed learning from adult learning theory (Knowles, 1975, 1984). For

Knowles, self-directedness depended on locus of control for learning, that is, who decides

what should be learned, who should learn it, what methods and resources should be used,

and how success of effort should be measured. Therefore, to the extent the learner makes

those decisions, learning is considered self-directed (Barrows; Knowles).

Characterizations and operationalization of self-directed learning are varied in

PBL literature, but in general, as briefed earlier, conceptually they are similar to what

cognitive psychologists describe as ...I-r. ..It i i ..1 learning -/,mnn.;nn. ,, & LeBeau,

2000). Similar to locus of control principle advanced by Knowles (1975), Zimmerman

and LeBeau suggest three dimensions or controlling influences of PBL on self-directed

learning processes including tutor-driven, group-driven, and self-directed. Self-directed

learning processes are defined as identifying learning objectives, pursuing learning

issues, and self-evaluating learning. Differential implementations of problem-based

instruction affect the magnitude of these influences on self-directed learning (Barrows,

1986, 2000). That is, depending upon how problem-based instruction is implemented,

defining what to learn occurs on multiple levels from tutor-driven objectives to students'

collaboratively or individually generated objectives, which affect varied degrees of self-

direction (Dolmans & Schmidt, 1994; Zimmerman & Lebeau).

Cognitive science research suggests that the power of problem-based instruction

to drive self-directed learning varies with the locus of control (Knowles, 1975) for

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EZ9H9JGU8_3KTFHE INGEST_TIME 2013-03-25T12:51:26Z PACKAGE AA00013618_00001