2007; 29: 855–871
Workplace-based assessment as an
educational tool: AMEE Guide No. 31
JOHN NORCINI1& VANESSA BURCH2
1Foundation for Advancement of International Medical Education and Research, Philadelphia, USA,2University of Cape Town,
Background: There has been concern that trainees are seldom observed, assessed, and given feedback during their workplace-
based education. This has led to an increasing interest in a variety of formative assessment methods that require observation and
offer the opportunity for feedback.
Aims: To review some of the literature on the efficacy and prevalence of formative feedback, describe the common formative
assessment methods, characterize the nature of feedback, examine the effect of faculty development on its quality, and summarize
the challenges still faced.
Results: The research literature on formative assessment and feedback suggests that it is a powerful means for changing the
behaviour of trainees. Several methods for assessing it have been developed and there is preliminary evidence of their reliability
and validity. A variety of factors enhance the efficacy of workplace-based assessment including the provision of feedback that is
consistent with the needs of the learner and focused on important aspects of the performance. Faculty plays a critical role and
successful implementation requires that they receive training.
Conclusions: There is a need for formative assessment which offers trainees the opportunity for feedback. Several good methods
exist and feedback has been shown to have a major influence on learning. The critical role of faculty is highlighted, as is the need
for strategies to enhance their participation and training.
For just over two decades leading educationists, including
medical educators, have highlighted the intimate relationship
between learning and assessment. Indeed, in an educational
context it is now argued that learning is the key purpose of
assessment (van der Vleuten 1996; Gronlund 1998, Shepard
2000). At the same time as this important connection was being
stressed in the education literature; there were increasing
concerns about the workplace-based training of doctors.
A study by Day et al. (1990) in the United States documented
that the vast majority of first-year trainees in internal medicine
were not observed more than once by a faculty member in a
patient encounter where they were taking a history or doing a
physical examination. Without this observation, there was no
opportunity for the assessment of basic clinical skills and, more
As one step in encouraging the observation of performance
by faculty, the American Board of Internal Medicine proposed
the use of the mini-Clinical Evaluation Exercise (mini-CEX)
(Norcini et al. 1995). In the mini-CEX, a faculty member
observes a trainee as he/she interacts with a patient around a
focused clinical task. Afterwards, the faculty member assesses
the performance and provides the trainee feedback. It was
expected that trainees would be assessed several time
throughout the year of training with different faculty and in
different clinical situations.
An advantage of the mini-CEX and other workplace-based
methods is that they fulfil the three basic requirements for
assessment techniques that facilitate learning (Frederiksen
1984; Crooks 1988; Swanson et al. 1995; Shepard 2000):(1) The
content of the training programme, the competencies expected
as outcomes, and the assessment practices are aligned (2)
Trainee feedback is provided during and/or after assessment
. The research literature on work-based formative assess-
ment and feedback suggests that it is a powerful means
for changing the behaviour of learners.
. Several formative assessment methods have been
developed for use in the workplace and there is
preliminary data evidence of their reliability and validity.
. The efficacy of feedback is enhanced if it is consistent
with the needs of the learner, focuses on important
aspects of the performance in the work-place, and has
characteristics such as being timely and specific.
. Faculty development is critical to the quality and
effectiveness of formative assessment.
. Strategies to encourage the participation of faculty are
critical to the successful implementation of formative
Correspondence: John Norcini, Foundation for Advancement of International Medical Education and Research (FAIMER) 4th Floor 3624 Market St,
Philadelphia PA 19104, USA. Tel: 1 215 823 2170; fax: 1 215 386 2321; email: JNorcini@faimer.org
ISSN 0142–159X print/ISSN 1466–187X online/07/09-100855–17 ? 2007 Informa UK Ltd.
events;(3) Assessment events are used strategically to steer
trainee learning towards the desired outcomes. Over the past
several years there has been growing interest in workplace-
based assessment and additional methods have been (re)in-
troduced to the setting of clinical training (National Health
Previous publications have focused on the advantages
and disadvantages of workplace-based methods from the
perspective of assessment alone (Norcini 2007). In this role,
the methods are best thought of as analogous to classroom
tests and they have much strength from this perspective.
However, it is difficult to assure equivalence across institutions
and the observations of faculty may be influenced by the
stakes and their relationships with trainees. Consequently,
their use faces challenges as national high stakes assessment
Perhaps more importantly, workplace-based assessment
can be instrumental in the provision of feedback to trainees to
improve their performance and steer their learning towards
desired outcomes. This paper focuses on the use of the
methods for this purpose and it is divided into five sections.
The first section briefly reviews the literature on the efficacy
and prevalence of formative assessment and feedback. This is
followed by a section that describes some of the more
common methods of work-based assessment. The third
section concentrates on feedback and it is explored from the
perspective of the learner, its focus, and which characteristics
make it effective in the context of formative assessment.
Faculty play a key role in the successful implementation of
formative assessment, so the fourth section describes strategies
to encourage their participation and training to improve their
performance. In the closing section we draw attention to the
challenges faced by medical educators implementing forma-
tive assessment strategies in routine clinical teaching practice.
Efficacy and prevalence of
formative assessment and
The purpose of formative assessment and feedback
Formative assessment is not merely intended to assign grades
to trainee performance at designated points in the curriculum;
rather it is designed to be an ongoing part of the instructional
process and to support and enhance learning (Shepard 2000).
Clearly, feedback is a core component of formative assessment
(Sadler 1989), central to learning, and at ‘the heart of medical
education’ (Branch & Paranjape 2002). In fact, it is useful to
consider feedback as part of an ongoing programme of
assessment and instruction rather than a separate educational
entity (Hattie & Timperley 2007).
Feedback promotes student learning in three ways (Gipps
1999, Shepard 2000):
. it informs trainees of their progress or lack thereof;
. it advises trainees regarding observed learning needs and
resources available to facilitate their learning; and
. it motivates trainees to engage in appropriate learning
Efficacy of feedback
Given these presumed benefits, it is appropriate to ask
whether there is a body of research supporting the efficacy
of feedback in changing trainees’ behaviour. Most compelling
is a synthesis of information on classroom education by Hattie
which included over 500 meta-analyses involving 1,800 studies
and approximately 25 million students (Hattie 1999). He
demonstrated that the typical effect size (ES) of schooling on
overall student achievement is about 0.40 (i.e. it increases the
mean on an achievement test by 0.4 of a standard deviation).
Using this as a benchmark or ‘gold standard’ on which to judge
the various factors that affect performance, Hattie summarized
the results of 12 meta-analyses that specifically included the
influence of feedback. The feedback effect size was 0.79,
which is certainly very powerful, and among the four biggest
influences on achievement. Hattie also found considerable
variability based on the type of feedback, with the largest
effect being generated by the provision of information around
a specific task.
Data to answer the question about the efficacy of
feedback are much more limited in the domain of medical
education but a recent meta-analysis by Veloski and collea-
gues looked at its effect on clinical performance (Veloski et al.
2006). Of the 41 studies meeting the criteria for inclusion,
74% demonstrated a positive effect for feedback alone.
When combined with other educational interventions, feed-
back had a positive effect in 106 of the 132 (77%) studies
A recent paper by Burch and colleagues reports on the
impact of a formative assessment strategy implemented in a
4th year undergraduate medical clerkship programme (Burch
et al. 2006). In this paper, students who engaged in an average
of 6 directly observed clinical encounters during a 14-week
clerkship reported that they more frequently undertook
blinded patient encounters (McLeod & Meagher 2001) in
which they did not consult the patient records before
interviewing and examining the patient. Prior to implementing
the formative assessment programme, students traditionally
interviewed and examined patients only after consulting
patient records. In addition they reported that they read
more frequently on topics only relevant to patients clerked in
the ward. While this paper provides information on self-
reported learning behaviour changes, it does suggest that
formative assessment may have the potential to strategically
direct student learning by reinforcing desirable learning
behaviour (Gibbs 1999).
A recent publication by Driessen and van der Vleuten
(2000) support the findings reported by Burch. In their study
they introduced a portfolio of learning assignments as an
educational tool in a legal skills training programme compris-
ing tutorials which were poorly attended and for which
students did not adequately complete the required pre-tutorial
work. The portfolio assignments, such as writing a legal
contract or drafting a legislative document, were reviewed by
peers and the tutor prior to being used as the teaching basis for
subsequent skills training sessions. This educational interven-
tion resulted in a twofold increase in time spent preparing for
skills training sessions.
J. Norcini & V. Burch
Prevalence of feedback
It is clear from these data that formative assessment and
feedback have a powerful influence on trainee performance.
However, there is a significant gap between what should be
done and ‘on the ground’ practice. Lack of assessment and
feedback, based on observation of performance in the
workplace, is one of the most serious deficiencies in current
medical education practice (Holmboe et al. 2004; Kassebaum
& Eaglen 1999). Indeed, direct observation of trainee
performance appears to be the exception rather than the rule.
In a survey of 97 United States medical schools, accredited
between 1993 and 1998, it was found that structured, observed
assessments of students’ clinical abilities were done across
clinical clerkships for only 7.4% to 23.1% of medical students
(Kassebaum and Eaglen 1999). A more recent survey of
medical graduates found that during any given core clerkship,
17% to 39% of student were not observed performing a clinical
examination (Association of American Medical Colleges 2004).
Likewise, Kogan & Hauer (2006) found that only 28% of
Internal Medicine clerkships included an in-course formative
assessment strategy involving observation of student perfor-
mance in the workplace setting. Outside the US, Daelmans
et al. (2004) reported that over a 6-month period, observation
of trainee performance occurred in less than 35% of
educational events in which observation and the provision of
feedback could have taken place.
Unfortunately the situation is no better in postgraduate
training programmes. In one study, 82% of residents reported
that they engaged in only one directly observed clinical
encounter in their first year of training; far fewer (32%)
engaged in more than one encounter (Day et al. 1990). In
another survey of postgraduate trainees 80% reported never or
only infrequently receiving feedback based on directly
observed performance (Isaacson et al. 1995).
Not only is assessment of directly observed performance
infrequently done as part of routine educational practice, but
the quality of feedback, when given, may be poor. Holmboe
colleagues evaluated the type of feedback given to residents
after mini-CEX encounters and observed that while 61% of
feedback sessions included a response from the trainee to the
feedback, only 34% elicited any form of self-evaluation by the
trainee. Of greatest concern, however, was the finding that
only 8% of mini-CEX encounters translated into a plan of
action (Holmboe et al. 2004a). The paper by Holmboe and
colleagues suggests that there are key reasons why clinician-
educators fail to give trainees effective feedback (see Box1):
In addition to finding that trainee observation and feedback
is infrequently given and often of limited value, it has also
been noted that the faculties’ assessment of trainee perfor-
mance may be less than completely accurate. Noel and
colleagues found that faculty failed to detect 68% of errors
committed by postgraduate trainees when observing a
videotape scripted to depict marginal competence (Noel
et al. 1992). The use of checklists prompting faculty to look
for specific skills increased error detection from 32% to 64%. It
was, however, noted that this did not improve the accuracy of
assessors. Approximately two thirds of faculty still scored the
overall performance of marginal postgraduate trainees as
satisfactory or superior. Similar observations attesting to the
poor accuracy of faculty observations have been made
elsewhere (Herbers et al. 1989; Kalet et al. 1992).
Based on the infrequency with which trainees are observed
and problems with the quality of the feedback they receive, it
is fair to ask whether observation of trainee performance is an
outdated approach to medical training and assessment. The
critical question, therefore, is whether clinical interviewing and
examination skills are still relevant to clinical practice such that
faculty should be trained to properly observe performance and
provide effective, useful feedback.
Feedback in relation to history and physical
Despite major technological advances, the ability to compe-
tently interview and examine patients remains one of the
mainstays of clinical practice (Holmboe et al. 2004). Data
gathered over the past 30 years highlight the critical
importance of these skills. In 1975 Hampton and colleagues
demonstrated that a good medical history produced the final
clinical diagnosis in 82% of 80 patients interviewed and
examined. In only one of 80 cases did laboratory tests provide
the final diagnosis not made by history or physical examina-
tion (Hampton et al. 1975).
Technological advances over the past two decades have
not made the findings of this study irrelevant. In 1992 Peterson
and colleagues showed that among 80 patients presenting for
the first time to a primary care clinic, the patient’s history
provided the correct final diagnosis in 76% of cases (Peterson
et al. 1992). Even more recently, an autopsy study of 400 cases
showed that the combination of a history and physical
examination produced the correct diagnosis in 70% of cases.
Diagnostic imaging studies successfully indicated the correct
diagnosis in only 35% of cases (Kirch & Schafii 1996).
Beyond diagnostic accuracy, physician-patient communi-
cation is a key component of health care. In a review of the
literature, Beck et al. (2002) found that both verbal behaviours
(e.g., empathy, reassurance and support) and nonverbal
behaviours (e.g., nodding, forward lean) were positively
associated with patient outcomes. Likewise, a study by Little
et al. (2001) found that the patients of doctors who took a
patient-centred approach were more satisfied, more enabled,
had greater symptom relief, and had lower rates of referral.
The ability to competently interview a patient and
perform a physical examination thus remains the cornerstone
Box 1. Key reasons why clinician-educators fail to give trainees
? Current in-vivo assessment strategies such as the mini-CEX may be
focusing on assessment of performance at the expense of providing
? The scoring sheets currently used for in-vivo assessment events provide
only limited space for recording comments thereby limiting feedback
? Clinician-educators do not fully appreciate the role of feedback as a
fundamental clinical teaching tool.
? Clinician-educators may not be skilled in the process of providing high
of clinical practice. The ability of faculty to accurately observe
feedback is therefore one of the most important aspects of
medical training. Although methods such as standardised
patients certainly provide complementary assessment and
feedback information, they cannot replace the central role
of observation by faculty.
Formative assessment methods
A number of assessment methods, suitable for providing
feedback based on observation of trainee performance in the
workplace, have been developed or regained prominence
over the past decade. This section provides a brief description
of the essential features of some of them including:
. Mini-Clinical Evaluation Exercise (mini-CEX);
. Clinical Encounter Cards (CEC);
. Clinical Work Sampling (CWS);
. Blinded Patient Encounters (BPE);
. Direct Observation of Procedural Skills (DOPS);
. Case-based Discussion (CbD);
. MultiSource Feedback (MSF).
Mini-clinical evaluation exercise (mini-CEX)
As described above, the mini-CEX (Figure 1, Source:
www.hcat.nhs.uk) is an assessment method developed in the
United States (US) that is now in use in a number of institutions
around the world. It requires trainees to engage in authentic
workplace-based patient encounters while being observed by
faculty members (Norcini et al. 1995). Trainees perform clinical
tasks, such as taking a focused history or performing relevant
aspects of the physical examination, after which they provide a
summary of the patient encounter along with next steps (e.g.,
a clinical diagnosis and a management plan).
These encounters can take place in a variety of workplace
settings including inpatient, outpatient, and emergency depart-
ments. Patients presenting for the first time as well as those
returning for follow up visits are suitable encounters for the
mini-CEX. Not surprisingly, the method lends itself to a wide
range of clinical problems including: (1) presenting complaints
such as chest pain, shortness of breath, abdominal pain,
cough, dizziness, low back pain; or (2) clinical problems such
as arthritis, chronic obstructive airways disease, angina,
hypertension and diabetes mellitus (Norcini et al. 2003).
In the original work, each aspect of the clinical encounter is
scored by a faculty member using a 9–point rating scale where
1–3 is unsatisfactory, 4–6 is satisfactory and 7–9 is superior.
The parameters evaluated include: interviewing skill, physical
examination, professionalism, clinical judgement, counselling,
organization and efficiency, and overall competence. Different
scales and different parameters have been used successfully in
other settings (e.g., National Health Service).
The core purpose of the assessment method is to provide
structured feedback based on observed performance. Each
patient encounter takes roughly 15 minutes followed by 5–10
minutes of feedback. Trainees are expected to be evaluated
several times with different patients and by different faculty
members during their training period.
This assessment tool has been shown to be a reliable way
of assessing postgraduate trainee performance provided there
is sufficient sampling. Roughly 4 encounters are sufficient to
achieve a 95% confidence interval of less than 1 (on the
9-point scale) and approximately 12–14 are required for a
reliability coefficient of 0.8 (Norcini et al. 1995, 2003; Holmboe
et al. 2003).
In addition to the postgraduate setting, the mini-CEX has
been successfully implemented in undergraduate medical
training programmes (Hauer 2000; Kogan et al. 2003; Kogan
& Hauer 2006). In this context, the period of observation and
feedback is often longer, ranging from 30–45 minutes (Hauer
2000; Kogan et al. 2002).
There is a growing body of evidence supporting the validity
of the mini-CEX. Kogan et al. (2002, 2003) found that mini-CEX
performance was correlated with other assessments collected
as part of undergraduate training. Faculty ratings of videotapes
of student-standardized patient encounters, using the mini-
CEX forms, were correlated with the checklist scores and
standardized patient ratings of communication skills (Boulet
et al. 2002). In postgraduate training, mini-CEX performance
was correlated with a written in-training examination and
routine faculty ratings (Durning et al. 2002). Holmboe et al.
(2004) found that, using the mini-CEX form, they could
differentiate amongst videos, scripted to represent different
levels of ability. Finally, et al. (2006) found that mini-CEX
scores were correlated with the results of a Royal College oral
Clinical encounter cards (CEC)
The CEC system, developed at McMaster University in Canada
(Hatala & Norman 1999) and subsequently implemented in
other centres (Paukert et al. 2002), is similar to the mini-CEX.
The basic purpose of this assessment strategy is also to score
trainee performance based on direct observation of a patient
encounter. The encounter card system scores the following
dimensions of observed clinical practice: history-taking,
physical examination, professional behaviour, technical skill,
case presentation, problem formulation (diagnosis) and
problem solving (therapy). Each dimension is scored using
a 6-point rating scale describing performance as 1: unsatisfac-
tory, 2: below the expected level of student performance, 3: at
the expected level of student performance, 4: above the
expected level of student performance, 5: outstanding student
performance, and 6: performance at the level of a medical
In addition to capturing the quality of the performance, the
4?6 inch score cards also provide space for assessors to
record the feedback given to the trainee at the end of the
This system has been shown to be a feasible, valid, and
reliable measure of clinical competence, provided that a
sufficient number of encounters (approximately 8 encounters
for a reliability coefficient of 0.8 or more) are collected (Hatala
& Norman 1999). Moreover, introduction of the system was
found to increase student satisfaction with the feedback
J. Norcini & V. Burch
Mini-clinical evaluation exercise form. Source: www.hcat.nhs.uk.
process (Paukert et al. 2002) and to have modest correlations
with other forms of assessment (Richards et al. 2007).
Clinical work sampling (CWS)
Thisassessment method,developed inCanada,isalso basedon
direct observation of clinical performance in the workplace
(Turnbull et al. 2000). The method requires collection of data
domains either at the time of admission (admission rating form)
or during the hospital stay (ward rating form). These forms are
completed by faculty members directly observing trainee
performance. The domains assessed by faculty include:
communication skills, physical examination skills, diagnostic
acumen, consultation skills, management skills, interpersonal
behaviour, continued learning skills and health advocacy skills.
Not all skills are evaluated on each occasion.
Trainees are also assessed by ward nursing staff (using the
multidisciplinary team rating form) and the patients (using the
patient rating form) who are in the care of the trainees. These
rating forms, also completed on the basis of directly observed
behaviour, require a global assessment and ratings of the
following domains: therapeutic strategies, communications
skills, consultation with other health care professionals,
management of resources, discharge planning, interpersonal
relations, collaboration skills, and health advocacy skills and
All rating forms use a 5-point rating scale ranging from
unsatisfactory to excellent performance. This assessment
method has also been shown to be valid and reliable provided
a sufficient number (approximately 7 encounters for a
reliability coefficient of 0.7) of encounters are observed
(Turnbull et al. 2000).
A later study found that the CWS strategy could be adapted
to radiology residency using a handheld computerised device
(Finlay et al. 2006). Compliance with voluntary participation
was not as great as expected but this evaluation format
included the opportunity to discuss performance at the time of
data entry, rather than at the end of rotation. The investigators
found the method less useful for summative purposes
although the sample size was small (N¼14).
Blinded patient encounters
This formative assessment method is based on the same
principle as the three assessment methods already mentioned.
It is unique, however, in that it forms part of undergraduate
bedside teaching sessions. (Burch et al. 2006). Students, in
groups of 4–5, participate in a bedside tutorial. It starts with a
period of direct observation in which one of the students in the
group is observed performing a focused interview or physical
examination as instructed by the clinician educator conducting
the teaching session. Thereafter the student is expected to
provide a diagnosis, including a differential diagnosis, based
on the clinical findings.
The patient is unknown to the student, hence the term
‘blinded’ patient encounter (McLeod & Meagher 2001). This
type of patient encounter has the advantage of safely allowing
the trainee to practice information gathering, hypothesis
generation, and problem solving without access to the
workup by more senior doctors.
After the presentation, the session focuses on demonstrat-
ing the important clinical features of the case as well as
discussing various issues, for example appropriate investiga-
tion and treatment relevant to the patient’s presenting clinical
problem. It concludes with a feedback session in which the
student receives personal private advice about his/her
Feedback is provided using a 9-point rating scale for
assessment of clinical interviewing and examination skills
as well as clinical reasoning skills. The rating scale ranges from
1–3 for poor performance, 4–6 for adequate performance and
7–9 for good performance. Space is provided on the score
sheet to add other written comments. Students keep the score
sheets which are only used for feedback purposes.
Direct observation of procedural skills (DOPS)
This assessment method (Figure 2, Source: www.hcat.nhs.uk),
developed in the UK, focuses on evaluating the procedural
skills of postgraduate trainees by observing them in the
workplace setting (Wragg et al. 2003). Just as in CWS and the
Encounter Card Assessment systems, trainees’ performance is
scored using a 6-point rating scale where 1–2 is below the
expected level of competency, 3 reflects a borderline level of
are above the expected level of competency. The assessment
procedure is generally expected to require 15 minutes of
observation time and 5 minutes dedicated to feedback.
Trainees are provided with a list of commonly performed
procedures for which they are expected to demonstrate
competence such as endotracheal intubation, nasogastric
tube insertion, administration of intravenous medication,
venepuncture, peripheral venous cannulation and arterial
blood sampling. They are assessed by multiple clinicians on
multiple occasions throughout the training period.
This method of procedural skills assessment is not limited
to postgraduate training programmes. Paukert and colleagues
have included basic surgical skills to be mastered by under-
graduate students in their clinical encounter card system
(Paukert et al. 2002).
Although DOPS is similar to procedural skills log books, the
purpose and nature of these methods differ significantly. The
recording of procedures is common to both of them, but log
books are usually designed to ensure that trainees have simply
performed the minimum number required to be considered
competent. The provision of structured feedback based on
observation of a performance is not necessarily part of the log
book process. Moreover, the procedure is not necessarily
performed under direct observation and little feedback, if any,
is expected to be given. In contrast, DOPS ensures that trainees
are given specific feedback based on direct observation so as
to improve their procedural skills.
Case-based discussion (CbD)
This assessment method is an anglicised version of Chart-
Stimulated Recall (CSR) developed for use by the American
J. Norcini & V. Burch
Directly observed procedural skills form. Source: www.hcat.nhs.uk.
Board of Emergency Medicine (Maatsch et al. 1983). It is
currently part of the Foundation Programme implemented for
postgraduate training in the UK National Health Service.
In CbD, the trainee selects two case records of patients in
which they had made notes and presents them to an assessor.
The assessor selects one of the two for discussion and explores
one or more aspects of the case, including: clinical assessment,
investigation and referral of the patient, treatment, follow-up
and future planning, and professionalism (Figure 3, Source:
www.mmc.nhs.uk). Since the case record is available at the
time of assessment, medical record keeping can also be
assessed by the examiner.
This type of performance assessment focuses on evaluating
the clinical reasoning of trainees so as to understand the
rationale behind decisions made in authentic clinical practice.
As with other assessment methods described, each encounter
is expected to last no more than 20 minutes, including
5 minutes of feedback. Trainees are expected to engage in
multiple encounters with multiple different examiners during
the training period.
There are several studies supporting the validity of this
measure. Maatsch et al. 1983) collected several assessments for
a group of practicing doctors eligible for recertification in
Emergency Medicine. They found that CbD correlated with a
number of the other measures, including chart audit. The score
distribution and pass-fail results were consistent with scores on
initial certification, ten years earlier. As importantly, CbD was
considered the most valid of the measures by the practicing
doctors participating in the study.
A study by Norman and colleagues compared a volunteer
group of doctors to those referred for practice difficulties
(Norman et al. 1989). CbD was highly correlated with a
standardised patient examination and with an oral examina-
tion. More importantly, it was able to separate the volunteer
group from the doctors who were referred. Likewise, Solomon
et al. (1990) collected data from several different assessments
on practicing doctors eligible for recertification. CbD was
correlated with the oral examination as well as written and oral
exams administered 10 years earlier.
MultiSource feedback (MSF)
More commonly referred to as 360-degree assessment, this
method represents a systematic collection of performance data
and feedback for an individual trainee, using structured
questionnaires completed by a number of stakeholders. The
assessments are all based on directly observed behaviour
(Wragg et al. 2003) but they differ from the methods presented
above in that they reflect routine performance, rather than
performance during a specific patient encounter.
Although there are a number of different ways of conducting
this form of assessment, the mini-peer assessment tool (mini-
PAT) that has been selected for use in the Foundation
Programme in the UK is a good example. Trainees nominate 8
assessors including senior consultants, junior specialists, nurses
and allied health service professionals. Each of the nominated
assessors receives a structured questionnaire (Figure 4) which is
completed and returned to a central location for processing.
Trainees also complete self-assessments, using the same
questionnaires, and submit these for processing. The categories
of assessment include: good clinical care, maintaining good
clinical practice, teaching and training, relationships with
patients, working with colleagues and an overall assessment.
The questionnaires are collated and individual feedback is
prepared for trainees. Data are provided in a graphic form
which depicts the meanratings of the assessors and the national
mean rating. All comments are included verbatim, but they
remain anonymous. Trainees review this feedback with their
supervisor and together work on developing an action plan.
This process is repeated twice yearly during the training period.
This method is widely used in industry and business, but has
also been found to be useful in medicine. Applied to practicing
doctors, it was able to distinguish certified from non-certified
written examination (Ramsey et al. 1989; Wenrich et al. 1993).
In a follow-up study, two subscales were identified—one
focused on technical/cognitive skills and the other focused on
professionalism (Ramsey et al. 1993). Written examination
performance was correlated with the former but not the latter.
Multisource feedback has been applied to postgraduate
trainees as well as practicing doctors. The Sheffield Peer
Review Assessment Tool, which is the full scale version of
mini-PAT as shown in Figure 4 (Source: www.mmc.nhs.uk),
was studied with paediatricians and found to be feasible and
reliable (Archer et al. 2005). It also separated doctors by grade
and tended to be insensitive to potential biasing factors such as
the length of the working relationship. Whitehouse et al.
(2002) also applied multisource feedback to postgraduate
trainees with reasonable results.
Finally, this form of assessment has also been used
successfully with medical students (Arnold et al. 1981, Small
et al. 1993). Both positive and negative reports from peers
have influenced academic actions.
Overall, reasonably reliable results can be achieved with
the assessments of 8 to 12 peers.
Nature of the feedback
For the purpose of this discussion, feedback can be conceptua-
lised as ‘information provided by an agent (teacher, peer, self,
etc.) regarding aspects of one’s performance or understanding’
(Hattie & Timperley 2007). This information can be used by the
learner to ‘confirm, add to, overwrite, tune or restructure
information in memory, whether that information is domain
knowledge, meta-cognitive knowledge, belief about self and
tasks or cognitive tactics and strategies’ (Winnie & Butler 1994).
The main purpose of feedback is, therefore, to reduce the
discrepancy between current practices or understandings and
desired practices or understandings (Hattie & Timperley 2007).
Perspective of the learner
In order for feedback to fulfil this purpose, it needs to address
three fundamental questions for the learner:
. Where am I going?
. How am I going?
. Where to next?
J. Norcini & V. Burch
Case-based assessment form. Source: www.mmc.nhs.uk.
Mini-peer assessment questionnaire. Source: www.mmc.nhs.uk.
J. Norcini & V. Burch
To address the first question, it is critical that there be
clearly defined learning goals. If the goals are not clearly
articulated then ‘the gap between current learning and
intended learning is unlikely to be sufficiently clear for
students to see a need to reduce it’ (Hattie & Timperley 2007).
Goals can be wide ranging and variable, but without them
students are less likely to engage in properly directed action,
persist at tasks in the face of difficulties, or resume the task if
disrupted (Bargh et al. 2001). The existence of goals is also
more likely to lead students to seek and receive feedback,
especially if they have a shared commitment to achieving them
(Locke & Latham 1990). So, medical trainees need to have a
clear understanding of desired practice or competence in order
to seek feedback and stay focused on the task of achieving
competence in the domain of interest.
The second question focuses on the provision of concrete
information, derived from an assessment of the performance,
relative to a task or goal. To do so well requires criteria that
properly. The answer to this question addresses the traditional,
restricted definition of feedback. Nonetheless, it is critical to the
provision of effective feedback. Ironically, it is precisely this
aspect of feedback which is usually poorly done. Clinician-
educators are often reluctant to provide honest feedback,
particularly in the face of poor performance. Having a set of
clearly defined criteria makes it somewhat easier to provide
guidance based strictly on observed performance, rather than
interpretations of the trainee’s intentions.
The final important question from the perspective of the
trainee is what actions need to be taken in order to close the
gap between actual performance and desired performance.
Trainees need an action plan; specific information about how
to proceed in order to achieve desired learning outcomes. As
indicated previously, without honest feedback regarding
actual performance, trainees are unlikely to seek advice
about how to proceed in order to close the learning gap.
The interrelatedness of these questions becomes apparent
when attempting to address this final question. Indeed,
without clearly defined learning outcomes, including criteria
which make achievement of the learning goals explicit, and
honest feedback about observed performance, planning aimed
at improving performance will not take place. Closing the gap
between where trainees are and where they need to be is
both the purpose of feedback and the source of its influence
Focus of feedback
How effectively feedback addresses the three questions for
learners is dependent in part on what aspects of the
performance are addressed. Specifically, there are four foci
for feedback (Hattie & Timperley 2007):
. feedback about the task;
. feedback about the process of the task;
. feedback about self-regulation;
. feedback about the self as a person.
The most basic focus of feedback addresses the quality of
the task performed. Using well defined criteria, trainees are
given specific information about whether they achieved the
required level of performance. This type of feedback is easiest
to give, and is consequently the most frequently provided. It is
most helpful when it concentrates on the performance, rather
than the knowledge required for the task. The latter is best
dealt with by providing direct instruction and it is not regarded
as feedback (Hattie & Timperley 2007).
One of the limitations of providing feedback focused only
on the task is that it is necessarily context-specific or task-
specific. Consequently, it does not generalise readily to other
tasks (Thompson 1998). On the other hand, providing
feedback that focuses on the process can be of more value
because it encourages a deeper appreciation of the perfor-
mance. This involves giving feedback that enhances an
understanding of relationships (the construction of meaning),
cognitive processes, and transfer to different or novel
situations (Marton et al. 1993). This focus for feedback is
also more likely to promote deep learning (Balzer et al. 1989).
A major component of this type of feedback is the provision
of strategies for error detection and correction, in other words
developing the trainee’s ability to provide self-feedback (Hattie
& Timperley 2007). Feedback about the process underlying
the task can also serve as a cueing mechanism leading to more
effective information search strategies. Cueing is most useful
when it assists trainees in detecting faulty hypotheses and
provides direction for further searching and strategising
Feedback that focuses on self-regulation addresses the
interplay between commitment, control, and confidence.
It concentrates on the way trainees monitor, direct, and
regulate their actions relative to the learning goal. It implies
a measure of autonomy, self-control, self-direction, and
self-discipline (Hattie & Timperley 2007). Effective learners
are able to generate internal feedback and cognitive routines
while engaged in a task (Butler & Winnie 1995).
Students who are able to self-appraise and self-manage are
able to seek and receive feedback from others. At the other
end of the spectrum are less effective learners who, having
minimal self-regulation strategies, are more dependent on
external factors, such as teachers, to provide feedback. For
these learners, feedback is more effective if it directs attention
back to the task and enhances feelings of self-efficacy such
that trainees are likely to invest more time and become more
committed to mastering the task (Kluger & DeNisi 1996).
Trainees’ attributions of success and failure can have more
impact than actual success or failure. Feelings of self-efficacy
can be adversely affected if students are unable to relate
feedback to the cause of their poor performance. In other
words, feedback that does not specify the grounds on which
students have achieved success or not, is likely to engender
personal uncertainties and may ultimately lead to poorer
performance (Thompson 1998). On the other hand, feedback
that attributes performance to effort or ability is likely to
increase engagement and task performance (Craven et al.
1991). Thus, when giving feedback it is critical that the assessor
clearly directs the feedback to observed performance, while
being aware of the impact feedback has on the self-efficacy of
The final focus of feedback is discussed not because
of its educational value but rather because it often has
J. Norcini & V. Burch
adverse consequences. This feedback is typically concentrated
on the personal attributes of the trainee and seldom contains
task-related information, strategies to improve commitment to
the task, or a better understanding of self or the task itself
(Hattie & Timperley 2007). This focus for feedback is generally
not effective, its impact is unpredictable, and it can have an
adverse effect on learning. This is particularly true of negative
feedback directed at a personal level.
Characteristics of effective feedback in the context of
Formative assessment strategies are thought to best prompt
change when they are integral to the learning process,
performance assessment criteria are clearly articulated, feed-
back is provided immediately after the assessment event, and
trainees engage in multiple assessment opportunities (Crooks
1988; Gibbs & Simpson 2004). In addition to these features,
Ende (1983) suggested that specific conditions could make
feedback more conducive to learning as described in Box 2.
In addition to the strategies suggested by Ende, it has also
been suggested that the efficacy of feedback may be further
improved by promoting trainee ‘ownership’ of feedback
(Holmboe et al. 2004). Strategies to achieve this include:
. encouraging trainees to engage in a process of self-
assessment prior to receiving external feedback;
. permitting trainees to respond to feedback;
. ensuring that feedback translates into a plan of action for
Based on a large qualitative study, including 83 academics
involved in education, Hewson & Little (1998) validated many
of these literature-based recommendations. They developed a
useful list of bipolar descriptors outlining feedback techniques
to be adopted and avoided (Box 3).
As already mentioned, formulating an action plan at the end
of a feedback session is critical to the success of formative
assessment. If a plan addressing the deficiencies is not
formulated, it results in failure to close the ‘learning loop’
and correct the identified problems (Holmboe et al. 2004).
Indeed, formulation of an action plan may constitute the
most critical step in providing feedback.
Beyond these actions, it is becoming increasingly recog-
nised that ongoing coaching or mentoring improves the
efficacy of feedback. This is particularly true of 360-degree
feedback strategies (Luthans & Peterson 2004). Current
literature in the business world reports that the role of the
workplace managers has been reconceptualised such that they
are seen to be facilitators of learning, creativity, and innovation
rather than directors or controllers of activity. Furthermore,
learning leaders or managers should foster interconnections
between people and systems so as to create collective learning
networks (Walker 2001). While this research has not been
replicated in the medical workplace setting, the emerging
success of these strategies in business suggests that similar
methods merit further consideration in clinical training settings.
From the preceding discussion it is clear that there is a need to
increase the frequency of observation of trainee performance
in order to provide feedback aimed at improving the quality of
the services they later render in clinical practice. To this end a
number of strategies have recently been implemented, but the
studies of their efficacy are limited in number and they report
Holmboe and colleagues examined the impact of a scoring
sheet specifically designed to remind faculty both of the
dimensions of feedback and that its main purpose is to provide
Box 3. Feedback techniques to be avoided and adopted.
Feedback techniques to be avoidedFeedback techniques to be adopted
Creating a disrespectful, unfriendly, closed, threatening climate Creating a respectful, open minded, non-threatening climate
Not eliciting thoughts or feelings before giving feedbackEliciting thoughts and feelings before giving feedback
Being judgemental Being non-judgemental
Focusing on personality Focusing on behaviours
Basing feedback on hearsay Basing feedback on observed facts
Basing feedback on generalizationsBasing feedback on specifics
Giving too much/too little feedback Giving the right amount of feedback
Not suggesting ideas for improvement Suggesting ideas for improvement
Basing feedback on unknown, non-negotiated goalsBasing feedback on well-defined, negotiated goals
Taken from Hewson & Little, 1998.
Box 2. Specific conditions to make feedback more conducive to
? Set an appropriate time and place for feedback.
feedbackregardingspecific behaviours,not general
? Give feedback on decisions and actions, not one’s interpretation of the
trainees motives or intentions.
? Give feedback in small digestible quantities.
? Use language that is non-evaluative and non-judgemental.
trainees with information about their performance aimed at
improving it (Holmboe et al. 2001). In the study, the faculty
control group did not receive any instruction regarding the
use of the score sheet, while the intervention group received
20 minutes of instruction at the start of the clinical rotation.
This information session outlined the characteristics of
effective feedback and stressed the importance of direct
observation of trainees to evaluate clinical competence.
Results of the study indicated that while the intervention
group did not provide more frequent feedback, their trainees
were more satisfied with the quality of feedback they received.
Two recent studies in the Netherlands have produced
similar findings. In one of the studies an undergraduate
surgical clerkship was restructured in an attempt to increase
the observation of trainee performance and the provision of
feedback by senior faculty members (van der Hem-Stokroos
et al. 2004). Restructuring of the clerkship included the
introduction of a log book, a form documenting observation
of skill performance, and individual appraisal by senior staff.
Faculty was informed of the changes but they were not given
formal instruction in trainee observation and how to provide
feedback. The results indicated no significant increase in
trainee observation or the provision of feedback. The authors
suggest that the lack of impact of the intervention may be
partly attributed to the limited input received by faculty
involved in the study, particularly limited involvement in the
process of restructuring the clerkship.
In the other study, Daelmans et al. (2005) introduced
in-training assessment in an undergraduate medical clerkship
programme. Senior clinical staff was informed about the
introduction at a meeting held at the beginning of the
clerkship. They also received a letter outlining the in-training
assessment programme. The findings indicated that despite
implementing this new programme, students were not more
examinations in the workplace. In their discussion of the
results they suggest that observation and feedback regarding
student performance may have been improved if faculty
members had been more frequently reminded of the
programme, for example daily meetings could have been
used to alert faculty to the importance and potential
educational value of the programme.
In contrast to these studies, Turnbull et al. (2000) describe
a strategy using clinical work sampling in which students
received feedback based on directly observed patient encoun-
ters an average of eight times during a 4-week clerkship
rotation. In this study, faculty members observing students in
the workplace attended a 2-hour workshop outlining the
assessment and feedback strategy. In addition, they received
monthly communications reminding them of the project.
Students were also oriented to the project before it started,
and met with the research associate on a weekly basis during
the clerkship rotation. Results indicated that the ongoing
collection of performance data was feasible.
In another study using the clinical encounter card system,
students engaged in a directly observed assessment event an
average of 35 times during a 12-week surgery clerkship
(Paukert et al. 2002). As in the other study, evaluators involved
in the project were briefed about the project in a number of
short 15-minute meetings outlining the purpose and impor-
tance of the intervention implemented. These information
sessions formed part of other meetings routinely held in the
department, for example morbidity and mortality meetings. At
each of these information sessions, faculty were asked to raise
any issues or concerns they had regarding the project. They
also received a letter explaining the assessment and feedback
system prior to implementation. At the end of the clerkship,
students were more satisfied with the feedback they received.
Based on these studies it is clear that a number of strategies
need to be employed to successfully implement an assessment
process in which trainees receive feedback based on directly
observed performance in the workplace. First, it is apparent
that involvement of faculty in planning an in-course formative
assessment strategy is likely to enhance their engagement in
the process. Second, faculty need to be thoroughly briefed
about the purpose and process of the observation and
feedback strategy implemented. Third, students need to be
properly informed about the purpose and format of the
assessment method used. In particular, it is critical that the
potential learning benefits of the system are emphasized rather
than the assessment aspects of the methods being used.
Finally, faculty and students need to be regularly reminded of
the benefit of formative assessment and the importance of
keeping the assessment strategy active in the workplace.
While successfully implementing a formative assessment
strategy in the workplace is an achievement in its own right,
it is important to ensure that the quality of the observations
made by attending faculty are accurate and that the feedback
received by students is effective. As was highlighted earlier,
faculty observations of student performance may not be
sufficiently accurate to identify errors in student performance.
While the use of checklists has been shown to improve the
ability of assessors to detect errors in performance (Noel et al.
1992), they have not been shown to improve the overall
accuracy of assessors. This is an issue that requires further
research; effective strategies to address this problem clearly
need to be found.
While the accuracy of examiners remains an issue needing
further work, the stringency of examiners can be improved
with training. A recent paper by Boulet et al. (2002) examined
the stringency of examiners using the mini-CEX to evaluate
directly observed trainee performance. They reported signifi-
cant variability among the examiners even when they were
observing the same event. Holmboe and colleagues have
shown that assessor training can address this issue. In their
paper, study participants engaged in a one-day video-based
training session aimed at reducing variability among faculty
when providing assessments and feedback on observed
performance. Participants engaged in performance dimension
training and frame-of-reference training (Holmboe et al. 2004).
The former was accomplished by getting faculty to discuss and
define key components of competence for specific clinical
skills and develop criteria for satisfactory performance. The
latter was addressed by giving individual faculty members the
opportunity to score real-time trainee performance using
J. Norcini & V. Burch
standardised patients and standardised trainees. While one Download full-text
faculty member scored the performance of the trainee and
provided feedback, other faculty members scored the trainee’s
performance by watching the interview and examination on a
video monitor. The encounter ended with a group discussion
of how each member of the group rated the performance and
reasons for the scores allocated. Finally the facilitator
described what type of trainee performance the case scenario
was scripted to depict.
Eight months after this faculty development effort, a set of
video recordings of scripted patient encounters were again
used to compare the performance of trained faculty as
compared to a cohort of untrained faculty. Trained faculty
were more stringent than untrained faculty members and they
also reported feeling more comfortable providing trainee
feedback. This study is one of the first demonstrating the
beneficial impact of faculty training for the purpose of scoring
performance with the intention of providing trainee feedback.
In this closing section of the paper we wish to highlight areas
where further work is needed to address some pivotal
questions regarding workplace-based formative assessment
and feedback. First and foremost, we need to develop
strategies that will ensure successful and sustainable imple-
mentation of formative assessment in the workplace. Most of
what has been done to date has been research-based, short
term projects. We need studies that identify the determinants
of successful, sustainable assessment and feedback strategies
so that we can better understand factors that promote trainee
feedback as a routine feature of training programmes rather
than a unique feature of selected programmes only. Long term
use may require further modification and simplification of
existing methods so as to make them more user-friendly in
busy clinical settings where patient care is the first priority and
trainee assessment of less importance.
Based on current literature it is apparent that poor faculty
participation in formative assessment and feedback strategies
is probably the most significant limiting factor currently
identified. Why faculty do not routinely engage in trainee
assessment and feedback needs to be better understood if we
wish to improve the situation. One strategy that may be of
benefit would be a reward structure for busy clinicians that
appropriately recognises their educational contributions and/
or provides them protected time to engage in teaching
activities. Another strategy would be to identify a core group
of faculty whose only educational job is assessment and
formative feedback. Other strategies clearly need to be
identified. In any event, these realities need to be addressed
before formative assessment is likely to be a routine feature of
workplace-based training programmes.
Second, we need to improve the quality of the assessments
and feedback given to trainees through a concerted faculty
development effort. Current work indicates that feedback
rarely results in the formulation of an action plan, a critical
component of effective feedback, and only sometimes
involves self-assessment by the trainee. Both these issues
need to be addressed if feedback is to be owned by the trainee
and remedial action undertaken to improve performance.
In addition, the accuracy and stringency of feedback need to
be improved. Innovative strategies to address this important
aspect of formative assessment need to be developed.
Finally, the impact of feedback on trainee learning
behaviour and performance needs to be determined. To date
there is very little information about the strategic use of
formative assessment in the workplace context to drive the
learning of medical trainees. The need for such data is
apparent. Not only do we need to determine the impact of
feedback on learning behaviour, but we also need to know
what the performance-in-the-workplace benefits can be
expected to be achieved by successful formative assessment
In the context of the workplace-based education of doctors,
there has been concern that trainees are seldom observed,
assessed, and given feedback. This has led to increasing
interest in a variety of formative assessment methods that
require observation and offer the opportunity for feedback,
including the mini-clinical evaluation exercise, clinical encoun-
ter cards, clinical work sampling, blinded patient encounters,
direct observation of procedural skills, case-based discussion,
and multisource feedback. The research literature on formative
assessment and feedback suggests that it is a powerful means
for changing the behaviour of students and trainees.
To enhance the efficacy of the methods of workplace-
based assessment, it is critical that the feedback which is
provided be consistent with the needs of the learner, focus on
important aspects of the performance (while avoiding personal
issues), and have a series of characteristics which make it
maximally effective. Since faculty play a key role in the
successful implementation of formative assessment, strategies
to provide training and encourage their participation are
Notes on contributors
JOHN J. NORCINI, PhD has been President and CEO of the Foundation for
Advancement of International Medical Education and Research (FAIMER?)
since May 2002. For the 25 years before joining the Foundation, Dr. Norcini
held a number of senior positions at the American Board of Internal
Medicine. His principal academic interest is in the area of the assessment of
VANESSA C. BURCH, MBChB, PhD is Associate Professor of Medicine at
the University of Cape Town, South Africa. She convenes the under-
graduate medical degree programme in the Faculty of Health Sciences and
is also actively involved in postgraduate education in the Faculty. Her main
academic interests are in the assessment of clinical competence and
innovative methods of medical education in resource-constrained educa-
tional environments typical of developing countries.
Arnold L, Willoughby L, Calkins V, Eberhart G. 1981. Use of peer evaluation
in the assessment of medical students. Med Educ 56:35–41.
Archer JC, Norcini JJ, Davies HA. 2005. Peer review of paediatricians in
training using SPRAT. Br Med J 330:1251–1253.