Although problem-based learning has been at the forefront of reforms to medical curricula since its inception more than thirty years ago, conclusive evidence about its effectiveness as an educational approach remains elusive. In addition to research-related issues, such as non-randomised groups and small sample sizes, are variables inherent in the implementation of PBL curricula. These include variability in selection criteria for prospective students and the precise model of PBL employed [1]. An associated issue for which definitive answers have yet to be found relates to the qualifications and backgrounds of staff employed as facilitators in PBL tutorials.

The function of the tutor in PBL differs considerably from that of the tutor in conventional tutorials in which the tutor assumes a comparatively didactic role. A major feature of PBL is that learning is student-centred in that students take responsibility for identifying and addressing their own learning needs; tutors are required to facilitate this rather than adopt the position of content expert. Facilitation requires understanding of the learning process and primarily involves monitoring of student learning and promotion of effective group function. The student-centred learning approach of PBL means that for tutors, content knowledge should be subordinate to proficiency in group facilitation [2]. Thus, effective tutors promote student learning by creating a supportive environment which encourages active participation by all members of the group, by monitoring the quality of learning through questions and feedback and by encouraging the development of students' metacognitive skills [3].

Most of the studies [4–8] that have looked at the characteristics of skilled PBL tutors have compared the effectiveness of content experts with that of non-experts as measured either by student satisfaction or academic achievement. (In medical curricula, the problems that students address are presented as integrated clinical scenarios. Thus, content experts are seen as those having relevant subject matter knowledge. This refers not only to clinical skills such as history taking and physical examination but also to knowledge of basic science, public health and ethico-medico-legal issues as required by the problem. Thus, although non-clinically qualified academic staff may be seen as having expertise in the teaching/facilitation process and/or particular aspects of the curriculum, when it comes to subject matter knowledge, it is clinicians who are considered as experts.) Although findings from these studies are inconclusive, overall there is a slight preponderance in favour of content expertise over group facilitation skill with regard to both academic achievement and student satisfaction [4, 6–8]. This is not a surprising outcome given the subjective nature of student perceptions of tutor effectiveness, possible conflict between students' perception of effective tutors and those characteristics essential to PBL (such as student-centredness, a non-didactic approach, emphasis on group function and self-directed learning) and the large number of factors that influence student academic achievement in addition to the skill of their PBL tutor. Findings from research into the relative effectiveness of staff versus student tutors are also inconclusive [9–11]. Additionally, Schmidt and Moust [12] in a study of PBL in a series of six-week health science courses looked at the influence of tutor style on student learning behaviour and academic achievement. Their findings suggested that the most effective tutors, as judged by the students, were those with both content knowledge and the ability to empathise with their students' circumstances. As far as we are aware, there have been no studies which have specifically explored the relationship between students' perception of overall effectiveness, particular tutor behaviours, clinical qualifications and academic appointment.

The medical program at the University of Queensland is a four-year, graduate entry program and features a PBL curriculum. The first year of the MBBS has an enrolment of approximately 270 students, divided into 26 PBL groups, and employs a range of PBL tutors over three teaching terms of about 11 weeks each. Although tutors may have medical, basic science or educational qualifications, the majority have expertise in one or other of the basic sciences, reflecting the dominant focus of the First Year curriculum. Tutors may be full-time academic staff or postgraduate students and others employed on a casual basis. All tutors are specifically trained in PBL before appointment to a student group and may teach up to three terms each year.

As stated above, tutoring in PBL tutoring has two components: facilitation skill and content knowledge. It may be expected that students would consider the principal strength of clinically qualified tutors to be their greater relevant content knowledge. In contrast, the principal strength of non-clinically qualified academic staff to the PBL process would be the facilitation skills derived from (often extensive) teaching experience. This study therefore has two aims. First, to compare the tutoring style and overall effectiveness (as perceived by their students) of Year 1 PBL tutors based on academic qualifications and appointment category; and second, to determine which of six specific tutor styles contributed most to students' perception of effective tutoring. The following hypotheses were tested: a) clinicians would be seen as more effective tutors than non-clinicians; b) full-time academic staff would be seen as more effective tutors than non-staff (casually-employed) tutors; and c) students would view subject matter knowledge as a more important determinant of effective tutoring than group facilitation skill. Thus, it is hoped that this study will provide a useful contribution to the PBL literature specifically by allowing both PBL tutors and their students to increase their understanding of the PBL process and how it may be optimally applied in medical education.

Correlation coefficients were calculated to determine the strength of the association between students' perception of their tutor's overall effectiveness and each of the tutor behaviours assessed. With the exception of the scale measuring use of authority, all scales showed a significant, although variable, association with overall effectiveness score. Use of expertise and role congruence correlated most strongly. When the multivariate relationship between individual scales and overall score was analysed, there was a close association between five scales (Use of Expertise, Cognitive Congruence, Test Orientation, Role Congruence and Cooperation Orientation). A final regression model retained only Use of Expertise (β-coefficient = 0.53, p <0.01), Authority (β-coefficient = -0.20, p <0.01) and Role Congruence (β-coefficient = 0.39, p <0.01).

Full-time academic staff are more likely than non-staff (casual and postgraduate student) tutors both to emphasise summative assessment as a factor in learning, and to display greater skill in establishing and maintaining a team approach among their students. These findings may be due to the considerably greater time that faculty spend in teaching and assessment tasks in general, as well as experience in dealing with student issues relative to casual tutors and postgraduate students. Although age was shown not to have a significant impact on any of these behaviours, it is worth noting that most (85.7%) staff tutors were in the 40+ age group, compared to 53.8% for non-staff tutors.

Using the same instrument, Moust and Schmidt [10] investigated differences in tutoring style between staff and non-staff tutors. They found that staff tutors scored significantly lower than non-staff tutors in three of the six behaviours: cognitive congruence, test orientation and role congruence, with staff tutors scoring higher on the authority scale only. This is in contrast to our findings which found significant differences in only two of the scales (and those in favour of staff tutors), test orientation and cooperation orientation. The discrepancy between the two studies may be explained, at least in part, by the different composition of the non-staff group between the two studies. The non-staff group in our study was more heterogeneous, consisting of both casual and student tutors, thereby blurring the distinction between staff and non-staff, particularly with regard to age – only 8.3% of student tutors were over 40 years compared with 85.7% of casual tutors.

With regard to clinical versus non-clinical tutors, medically-qualified PBL tutors are significantly more likely to make use of their expertise in facilitating their groups' learning and to empathise with their students' lives as medical students. This is not surprising, given their own prior experience as medical students themselves and that they can be assumed to also have a large body of clinically-relevant knowledge which they are able to apply to addressing the problem being studied.

None of these findings are reflected in the students' ratings for the last item in the questionnaire: how well they perceived that their tutor played his/her role (overall effectiveness). This may be due to the relative crudity of the four-point scale for each item in the questionnaire. Such an effect would be particularly noticeable with regard to the item on overall effectiveness which has a maximum score of only four points, as well as a relatively higher standard deviation than those for the other scales. Additionally, when taken in conjunction with the finding of significant correlations between five of the six behaviours and perceived effectiveness, it may be that characteristics other than those measured in this study also contribute to the students' perception of their tutor's effectiveness.

Nevertheless, the lack of a significant difference between clinicians and non-clinicians in students' ratings for overall effectiveness, given that they rate clinicians' use of expertise significantly higher, is striking. This is especially so in view of the substantial correlation between use of expertise and overall effectiveness, and raises the question of what exactly is meant by the term "expertise" in relation to problem-based learning. The issue of whether content experts are more effective as PBL tutors than those without content expertise has been, and continues to be, a subject of debate in the literature with many studies finding in favour of the experts [4, 7]. In research into PBL in integrated medical curricula, the experts are assumed to be clinicians, with basic scientists and others designated as non-experts for the reasons discussed earlier. However, it is possible that, in First Year of the MBBS where the principal emphasis is on the basic sciences, expertise in the basic sciences is seen as an equally valuable characteristic in PBL tutors as clinical expertise. In contrast, the Second-Year curriculum has a much greater clinical focus and the majority of PBL tutors are medically qualified. A follow-up study of tutors in Year 2 is planned.

An alternative explanation is that it is the way a tutor uses his/her expertise, rather than the degree of expertise, which is important in determining a tutor's score on this scale. That is, it is possible that clinicians adopt a tutor-centred, rather than student-centred approach to PBL tutoring but that this characteristic is not regarded by students as having a major impact on their tutor's effectiveness. Indeed, a recent study has shown that clinicians tend to ask questions directly of students, whereas non-clinicians are more likely to encourage students to ask questions of each other [6].

The same argument can be applied to the finding regarding staff versus non-staff tutors. That is, that although full-time staff are more test-oriented and better at establishing a collaborative environment for student learning, these differences are not enough in themselves to impact significantly on overall effectiveness, again suggesting that characteristics other than those measured at least partially influence this assessment.

In summary, these findings suggest that, although clinicians and staff tutors consistently scored higher on each of the measured behaviours than did non-clinicians and non-staff, most of these differences are not statistically significant and do not have a substantial impact on students' assessment of their effectiveness as PBL tutors. This conclusion has implications for the recruitment and training of PBL tutors in that it suggests that both subject matter knowledge and process facilitation skills are necessary but not individually sufficient characteristics of effective tutors. Developing a broad range of strategies to encourage optimal group functioning and to stimulate student learning should therefore be a major focus of tutor preparation.