Does your group matter? How group function impacts educational outcomes in problem-based learning: a scoping review

Abstract

Background

Problem-based learning (PBL) is a common instructional method in undergraduate health professions training. Group interactions with and within PBL curricula may influence learning outcomes, yet few studies have synthesized the existing evidence. This scoping review summarized the literature examining the influence of group function on individual student PBL outcomes. Following Kirkpatrick’s framework, experiential, academic, and behavioral outcomes were considered. The impacts of three aspects of group function were explored: (1) Group Composition (identities and diversity), (2) Group Processes (conduct and climate, motivation and confidence, and facilitation), and (3) PBL Processes (tutorial activities).

Methods

A literature search was conducted using Medline, CINAHL, and APA PsychInfo from 1980–2021, with the help of a librarian. English-language empirical studies and reviews that related group function to learning outcome, as defined, in undergraduate health professions PBL curricula were included. Relevant references from included articles were also added if eligibility criteria were met. The methods, results, discussions, and limitations of the sample were summarized narratively.

Results

The final sample (n = 48) varied greatly in context, design, and results. Most studies examined junior medical students (n = 32), used questionnaires for data collection (n = 29), and reported immediate cross-sectional outcomes (n = 34). Group Processes was the most frequently examined aspect of group function (n = 29), followed by Group Composition (n = 26) and PBL Processes (n = 12). The relationships between group function and outcomes were not consistent across studies. PBL experiences were generally highly rated, but favorable student experiences were not reliable indicators of better academic or behavioral outcomes. Conversely, problematic group behaviors were not predictors of poorer grades. Common confounders of outcome measurements included exam pressure and self-study.

Conclusions

The main findings of the review suggested that (1) group function is more predictive of experiential than academic or behavioral PBL outcomes, and (2) different Kirkpatrick levels of outcomes are not highly correlated to each other. More research is needed to understand the complexity of group function in PBL tutorials under variable study contexts and better inform curricular training and design. Standardized tools for measuring PBL group function may be required for more conclusive findings.

Introduction

Problem-based learning (PBL) is a method of teaching and learning that involves solving case problems in small groups. Built upon the philosophical foundation that knowledge is socially co-constructed, this pedagogy depends on self-directed learning, group discussion, problem-solving, and peer teaching as its primary processes [1]. Conceivably, group function is the backbone of this educational format. Since its inception in the late 1960’s at McMaster University [1], PBL has evolved and expanded on a global scale. Today, it finds its place as a central component of health professions education worldwide [2, 3]. Given the centrality and ubiquity of PBL in the training of health professionals, the practical implications of understanding its ability to produce individual student learning outcomes is of greater importance than ever before [4].

In a typical PBL tutorial, students are divided into small groups and presented with a relevant problem, most commonly in the form of a clinical case or problem [5]. The groups are tasked with identifying learning issues and objectives and formulating hypotheses about the case, then sent off for independent study on the topic [5]. Students later bring their findings back to the group and collaboratively synthesize knowledge through various cognitive processes that are evoked by group discussion [5, 6]. Courses tend to conclude with an assessment and reflection of group learning, with regard to both knowledge content and group process. A facilitator, also known as a tutor, is present throughout to guide tutorial activities [5]. At many institutions, tutors and/or students also undergo training prior to the start of a semester specifically to learn how to engage in a PBL setting [7], though the degrees of training and training satisfaction are variable [8]. The exact format of PBL also differs across institutions and programs [9]. Variations may involve different group sizes and compositions, different tutor identities, expertise, and roles [10], the presence or absence of individual and group assessments during various stages of the process, and additional components such as presenting small-group findings to a larger class [9].

Given its collaborative nature, PBL uniquely provides the opportunity to practice skills required for collective learning, such as the discussion and expansion of ideas, critical reflection, cooperation, communication with group members, and responding to the social influences of the group [11]. However, the inherent social dimension of groupwork introduces challenges [12] where students and groups may differ in academic skills, level of contribution, tolerance of other students, social status, etc [13]. Apart from the students, the tutor is also inherently involved in the functioning of the group. Tutors may contribute to both social functions, such as guiding the learning climate, and learning functions, such as planning tutorial activities, facilitating group processes, and evaluating outcomes [10]. The effectiveness or ineffectiveness of PBL towards individual learning outcomes may thus be conceptually dependent upon the abilities of the students and tutors to engage in group and curricular processes (Fig. 1) [12, 14].

Fig. 1

Conceptualization of group, PBL, and outcomes

The literature debates whether the merits of group learning are realized in standard practice [9, 15, 16]. Currently, there is no accessible knowledge base or tool to help researchers and educators understand the value of group function in contributing to student learning outcomes. Literature reviews on this topic are few, none of which focus on group function comprehensively in an undergraduate health professions population. It is currently unclear what aspects of group function specifically impact learning outcomes, to what degree, and in what manner. A new scoping review is thus valuable to help researchers understand the range of work that has been done so far on this topic and identify areas in need of further investigation, as well as help educators inform curriculum and training design.

Definitions

PBL

PBL has a broad range of definitions that span from the ideologic to the pedagogic. Several models of PBL have been identified, such as the Aalborg model, the Maastricht ‘Seven-Step’ model, and project-based PBL, which differ in details such as the frequency of attendance of the tutor at tutorials, the number of problems presented, and the level of tutorial structure [17]. Hung also conceptualized a nuanced PBL step-ladder that points from hybrid curricula with PBL components to pure PBL, based on the two dimensions of structuredness of cases and self-directedness of tutorials [9]. It has also been pointed out that the definition of PBL often manifests differently in theory than in practice, the latter being dependent on tutors’ and students’ human abilities to act out PBL in a real-world setting [9]. In recognizing the complications of PBL theory and implementation, this review considers the definition of PBL in its broadest sense and includes any format of learning that incorporates core PBL components (i.e. students work in groups and undergo self-directed learning to solve a clinical case or problem in the presence of a tutor [2]), within or without other curricular components.

PBL may also be differentiated from related pedagogy, such as team-based learning (TBL), small-group learning (SGL), and case-based learning (CBL), which are distinct in their origins, features, and goals, and yet may share overlapping elements [18, 19]. In some cases, the use of pedogeological terminologies have been blurred in practice across institutions and in research [9, 18, 19]. This review thus also considered literature using TBL or SGL terminology, on a case-by-case basis, so long as the curricular processes resemble those described above.

Group and group function

The PBL group is defined, according to Barrows’ model, as consisting of a number of students and a tutor, who engage collaboratively to solve tutorial problems [20]. Group function, then, refers to the processes of interactivity between the participants of the group, and between the participants and the PBL curriculum.

Learning outcome

Learning outcomes may be organized in several ways. For instance, PBL is recognized for its impact on both knowledge learning and social outcomes, at an individual and group level [11]. This review focuses on individual student learning outcomes, which most directly aligns with the qualifications of an independent health professional.

Research objective

The objective of this scoping review is to explore the breadth and depth of the existing literature that relate group function to PBL outcomes, to identify aspects of group function that are important in contributing to individual student learning outcomes in undergraduate health professional PBL.

Methods

This scoping review follows the PRISMA checklist for Transparent Reporting of Systematic Reviews and the JBI methodology for scoping reviews [21].

Search strategy

A preliminary search in Medline, CINAHL, and APA PsychInfo was conducted in October 2021 to identify relevant articles and inform the search strategy. The research question was broken down into search terms in natural language, which was then used to develop a full search strategy with the help of a librarian at McMaster Health Sciences Library. Search terms were adapted for each database, mapped to subject heading, and used as keywords to search titles and abstracts. Search strategy details are provided (Appendix 1).

Inclusion criteria

Included studies (1) are in English, (2) are published no earlier than 1980, out of consideration for practicality and relevance to the modern social context, (3) are empirical studies using any methodology or systematic or scoping reviews, (4) studied an undergraduate healthcare student population, and (5) explored aspects of group function, as defined, in relation to student learning outcome in a PBL setting.

Opinion texts, unpublished literature, and experimental studies that took place within simulated environments were excluded. Studies that aimed to compare PBL outcomes to traditional learning outcomes were also excluded, as the aim of this review is to understand the role of group components within PBL, not to distinguish the merits of PBL from other learning contexts.

Study screening and inclusion

All identified citations were uploaded into the Covidence tool for removal of duplicate records and primary screening. A pilot test of the inclusion criteria was conducted on a small sample of articles by two independent reviewers. The inclusion criteria were revised for clarity after discussion. Primary screening of titles and abstracts by the two reviewers took place between October and December. 73 inter-reviewer conflicts were encountered, such as when there was a lack of clarity about the relevance to group function or the population being studied in the abstract. These were resolved by discussion, over two online meetings. By mutual agreement, irrelevant abstracts were excluded, and truly ambiguous abstracts were conservatively included for full text review.

The first author conducted a full text review for abstracts labeled “yes” or “maybe” for inclusion by both reviewers. Eligible articles were included in the analysis. Additional relevant articles were identified by back-searching the reference section of included studies. These were included in the analysis if inclusion criteria were met after full text screening.

Data extraction

A data extraction sheet, developed by the first author, was used to extract and organize details from the included full text sample. Articles were reviewed for publication details, participants, PBL context, aims/purpose, methods, relevant key findings, and limitations. Additional fields were added and revisited as data extraction proceeded. A final draft of the extraction form is provided (Appendix 2).

Quality appraisal

A basic appraisal was also conducted to inform readers of the methodological quality of the included articles (Appendix 3). Quantitative and mixed-methods studies were assessed using the Medical Education Research Study Quality Instrument (MERSQI) framework [22]. Qualitative studies were assessed using a 12-item grid by Côté and Turgeon [23]. These models were selected for their particular focus on medical education research. Scoping and systematic reviews were assessed using the Joanna Briggs Institute Critical Appraisal Checklist for Systematic Reviews and Research Syntheses [24]. Quality appraisal scores did not influence screening decisions.

Data analysis framework

Aspects of group function

Fonteijn and Dolmans previously identified several variables that affect group functioning in PBL and categorized these into three clusters [25]. The first, "the resource pool”, considers the intrinsic properties of the group, such as group size and individual differences in identity, ability, and experience. Next, “the learning task and group learning processes” includes both the quality of the tutorial problem and concepts of autonomy, group climate, and team learning behaviors that enable the co-construction of knowledge. Third, “the learning context” broadly refers to the situational factors that influence how groups may interact, including the discipline of study, cultural context, socialization and training of students, and role of the tutor [25].

To practically aggregate data, this review gains inspiration from Fonteijn and Dolmans [25] but considers tutor components as a human part of the group rather than as a part of the curricular context. As such, group function will be organized into three aspects: Group Composition, Group Processes, and PBL Processes. Group Composition aligns closely with “the resource pool” and includes tutor identities in addition to student identities and other human resource factors. Group Processes reflects the behavioral and cognitive dimensions of “group learning processes” above, including student conduct and social climate, and motivation and confidence. Group Processes additionally considers the role of tutor facilitation. Finally, “the learning task” is reworked into PBL Processes, which examines factors related to tutorial activities, such as case quality and feedback.

Levels of learning outcome

The Kirkpatrick (KP) Model is a recognized method of assessing the efficacy of education and professional training programs from a learner-focused perspective [26]. This model, developed by Donald Kirkpatrick in the 1950s, is valuable in its ability to provide a simple categorization of educational outcomes in four ascending levels [26, 27]. In the context of health professions PBL, these levels are: i) KP1, which refers to students’ immediate reaction to a training program, including their perceptions and experience of PBL, ii) KP2, which refers to the learning of information, is measured by indicators of academic performance, iii) KP3, which refers to the impact on students’ behavior, including leadership and conduct, and iv) KP4, which refers to long term professional results in the workplace [28].

A previous systematic review on the effectiveness of TBL by Fatmi et al. interestingly pointed out a lack of correlation between KP1 and higher level outcomes [29], suggesting the importance of examining each level of KP outcomes separately.

Summarizing and reporting findings

Included articles were characterized numerically for year of publication, studied population and geographical distribution, study methods, aspects of group function studied, and KP level of outcomes reported. Overall results pertaining to the objective of this review were organized by aspects of group function and summarized narratively.

Results

Literature scope and characteristics

Results of the initial search, inclusion process, and exclusion details for ineligible full text articles are reported in Fig. 2. The final sample consisted of 48 articles [30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77].

Fig. 2

PRISMA-ScR Flow Diagram for study selection and inclusion

The final sample of studies (n = 48) was highly varied in context, design, and outcomes. Details are provided in Table 1. The search results yielded studies published between 1992 and 2020.

Table 1 Study characteristics

Studied population and geographical distribution

Most studies focused on medical students (n = 32). Others examined a mixture of programs including nursing, medicine and dentistry, and undergraduate health sciences, or a mixture of student and tutor populations. Geographically, there was representation across Europe (n = 16), North America (n = 12), East Asia (n = 7), the Middle East (n = 6), and Australia (n = 5).

Study methods

Two articles were systematic or scoping reviews. The remaining 46 articles were empirical studies, reliant on quantitative (n = 31), qualitative (n = 8), or mixed (n = 7) methods to evaluate various group functions and learning outcomes. Field study questionnaires using Likert-type scales were the most common study design (n = 29), followed by randomized control trials (RCTs) (n = 6), semi-structured interviews (n = 4), and observational studies (n = 4). Reported outcomes were largely cross-sectional (n = 39), and longitudinal studies did not exceed two years in length of follow-up.

Aspects of group function studied

For ease of categorization, a count of mentions of each aspect of group function is tallied in Table 1. The greatest number of articles examined elements related to Group Processes (n = 29), and the least number touched on PBL Processes (n = 12).

Learning outcomes

Most frequently reported outcomes were student experiences or perceptions of learning (KP1) (n = 38), followed by knowledge acquisition (KP 2) (n = 24). Very few studies reported behavioral outcomes (KP3) (n = 6), and no studies followed KP4 outcomes into the workplace.

Individual results of each included study are summarized in Table 2 [30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77].

Table 2 Study details and summary of results

Group Composition: Tutor identity

Studies looking at tutor identity compared expert and non-expert tutors, or faculty and student tutors. Student tutors were direct peers from the same cohort, or near-peers, which describes upper years or recent graduates. One study also compared tutored and untutored PBL [30].

Students did not show discriminatory preference (KP1) for content expert or non-expert tutors [31, 32], and appreciated both on different dimensions [33]. Non-experts were valued for group-management skills [31, 34, 35], and experts were rated as better able to navigate discussion around case objectives [34, 35]. However, these KP1 outcomes were not consistently predictive of KP2 learning outcomes [34]. Some studies found expert tutored students scored higher on exams [31, 33], but the majority of the literature reported no significant difference across tutor types in terms of attitudes towards learning and content [34, 36], knowledge and skills acquisition [32, 34, 37,38,39], and confidence with the material [40]. Similarly, tutors who were more perceived as more socially and cognitively congruent were rated more favorably by students (KP1), but did not produce better recall or module scores (KP2) [41].

Studies comparing faculty to student tutors found a slight preference for the latter (KP1) [34, 37, 40]. Peers and near-peers were rated as better at creating relaxing learning climates [37], more understanding of learner needs [42], more socially congruent [42], and thus more easily accepted by students [40]. Student tutors had greater trouble during case problem discussion and analysis [37, 38], but did not produce worse academic KP2 outcomes than faculty tutors [37, 39, 43].

Tutored PBL was preferred over untutored PBL [30]. On average, untutored students did not perform worse than tutored students on an exam (KP2), though there was greater variability in individual student scores [30].

Group Composition: Student identity

Students were characterized on dimensions such as baseline achievement and demographic factors.

Student perception of PBL (KP1) were generally favorable and did not significantly increase with achievement [44, 45]. Achievement was variably measured across studies as admission averages, GPA [46], final course grades [39, 44, 45], prior in-class process assessment scores [46, 47], or prior module scores [40, 46]. Overall student achievement appeared to be stable over time [47]. Individual student differences in academic ability [46, 47] and behavior within groups [47] persisted across PBL sessions, suggesting that PBL did not preferentially benefit high or low achieving students over time [47]. However, one study found that feedback improved the quality of student contributions to the group for baseline low-achievers, but not mid- or high-achievers [48].

Demographic diversity such as gender [46, 49, 50], race [49], ethnicity [49], and socioeconomic status [49] did not reliably influence student experience (KP1) or achievement (KP2). Diverse groups attributed a greater importance for diversity, but this had no impact on group function or academic performance [49]. However, the studies displaying these results were primarily conducted in countries with high baseline diversity. Contrastingly, two studies in the Middle East, where students are gender-separated for cultural reasons, found significant gender differences in student perception and behavior [51, 52]. For example, females placed greater priority on acquiring academic knowledge, whereas males valued the opportunity to participate in tutorials [51]. Female groups were also higher in motivation, productivity, and display of positive professional behaviors than male groups [52]. It is thus suggested demographics-related outcomes may be culturally dependent.

Students with more or less PBL experience did not differ in enjoyment of PBL (KP1) or exam scores (KP2) [35]. Some students have also experienced PBL as true-peer or near-peer tutors, which allowed them to facilitate their own cohort or a younger cohort of students, respectively. A systematic review of students with facilitating experiences reported favorable KP1 outcomes such as self-perceived development of professional identity, leadership skills, a sense of confidence and autonomy in their learning, and the ability to reflect on their learning gaps [53]. However, no significant improvements to academic outcomes (KP2) were found [53].

Group Composition: Other human resource factors

One study examining various PBL groups factors found that groups’ academic performances (KP2) positively correlated to that of its individual members [50]. Both individual and group performance also increased with group size, with the groups ranging from five to seven members [50]. A separate study examining the presence of librarians in PBL tutorials found no impact of the librarian on students attitudes towards PBL (KP1), exam performance (KP2), and confidence in information-searching skills (KP3) [54].

Group Processes: Student conduct and social climate

Certain student behaviors were favorably perceived (KP1) by students and tutors alike. These centered around themes of respectful assertiveness [55], listening to each other [43, 51, 55], setting clear group goals [55], giving constructive feedback [55, 56], defining and sharing leadership within the group [51, 55], maintaining self-awareness [55, 57], and being prepared for tutorials [55]. Both students and tutors reported good group dynamics and fair allocation of tasks overall [45, 58, 59].

Some favorable behaviors and pleasant group interactions made for enjoyable experiences (KP1) but were not necessarily beneficial to knowledge acquisition (KP2) [55]. For example, contributing to group discussions was viewed by students as important [45, 51, 55], but students who did not actively participate were found to perform equally well on academic assessments [55, 60]. Having group members who were caring or accommodating helped the group feel comfortable, but did not improve learning [55]. Groups may also bond well unproductively, as is the case when students shared a mutual disdain for the learning activity [55].

Contrastingly, unfavorably perceived (KP1) student behaviors included condescension, stubbornness, passivity, and unreliability [55]. Other common student-related issues included quiet and dominant group members [55, 61, 62], lateness and absenteeism [51, 55, 61], rushing through tutorial activities [61], and lack of effective group interactions [61]. Interestingly, both lack of effort for conflict resolution [63] and the act of conflict resolution [64] were perceived by students as undesirable (KP1), in two separate studies.

Similar to the above, unfavorable student behaviors were not linked to worse KP2 outcomes. For instance, several studies found that both students and tutors felt unprepared to deal with conflicts between group members [51, 55, 61, 64, 65], and thus dysfunctional group members and behaviors were most often ignored [51, 63, 66]. Yet, persistent disruptive behaviors were not necessarily harmful to learning or grades (KP2) [55, 61, 66]. In some cases, groups with perceivably poor group dynamics, such as those containing students with aggressive or critical personalities, may even have better KP2 outcomes, as the dominance helps kick-start group discussions and other learning processes [55].

Group Processes: Motivation and confidence

Studies exploring group processes from a cognitive perspective identified motivation and confidence as factors that effectively map onto experiential (KP1) [32, 44, 53, 67], academic (KP2) [60], and behavioral (KP3) outcomes [53].

Both tutors and students who were more confident in their skills or content knowledge were more likely to engage productively in PBL activities [62, 68] and reported more positive experiences [58, 67]. For example, one study exploring the implementation of a wiki platform for students to share resources found that students with greater confidence in their information-searching skills and understanding of the content were more likely to contribute to the wiki [68]. In several studies, perception of PBL (KP 1), though not learning (KP2), improved slightly with year of study as students became more confident with the process of group formation [69, 70] and with self-direction [32, 51, 67].

Higher motivation and sense of contribution in students mapped onto more favorable KP1 outcomes [67, 71] and higher group productivity [52]. Motivation may be intrinsic, such as pre-existing interest for the content [72], or developed through favorable group interactions [60]. Meaningful group discussions, for example, excited students’ interests toward the content and their desires to learn more [60], while superficial group discussions and unresponsive group members were demotivating and inhibitory to learning (KP2) [60, 63]. As well, tutor qualities may play a role [72, 73]. Students encountered with authentic, passionate, professional, and dependable tutors reported being more incentivized to learn [72, 73], while those who were faced with unprofessional or disengaging tutors reported strongly worded narratives of demotivation from the PBL material, or even the profession at large [73].

Group Processes: Tutor facilitation

Tutors and students acknowledged that facilitation practice is not standardized [42, 62]. Variations in facilitation practice was not necessarily tied to the level of content expertise of the tutor [33, 42]. Rather, facilitation behaviors were reactionary to the performance of the group [71, 74], the tutor’s perception of their students’ cognitive and practical needs [62], the tutor’s own levels of background knowledge and experience with PBL [34, 58, 62]. For example, some tutors were more willing to actively help resolve student concerns while others favored giving groups the freedom to self-regulate, meaning the nature of group interactions were fundamentally different tutor-to-tutor [62]. Neither content facilitation nor process facilitation consistently produced better PBL outcomes at any KP level [34, 57], though the tutor’s ability to stimulate self-directed learning improved students’ perceptions of PBL case quality and group performance [57].

There exist certain facilitation practices that were generally identified by students as highly problematic in PBL, including both excessively imposing and excessively uninvolved tutors [61, 65, 66] and tutors who were unprepared for tutorial [63]. These views may be moderated by students’ year of study. Students’ facilitation values (KP1) were found to shift slightly over time, where junior students preferred tutors with more content expertise and willingness to provide guidance, and senior students felt a greater need for autonomy from their tutors [73].

Some studies also found that students perceived the need and expectation for tutors to play a greater role in managing group process-related issues, such as dominant or uncommitted students, clashes in student personalities, or lack of group productivity [59, 65, 66]. However, tutors generally felt less prepared to deal with these [61, 62] as compared to content-related issues involving curriculum or case design [61].

Tutors were additionally viewed as role models of professional conduct, such as the handling of sensitive case issues [65].

PBL Processes: Tutorial activities

Studied PBL processes included a wide range of independent factors such as different natures of tutorial activities [70] and phases [75, 76], group testing [45], knowledge sharing [68], and reflection and feedback [48, 57]. Perceptions of PBL processes were generally favorable across studies [45, 58, 70]. There was good understanding of the intention of PBL to train professional skills, such as teamwork, competence, and caring [66, 77]. Students were able to recognize the importance of these skills in later professional settings [45] and made up for any gaps in PBL curriculum through extracurricular involvements [77]. Engagement with knowledge sharing resources also contributed to students’ exploration of professional identity as knowledge experts and increased their confidence with the content [68].

Some studies suggested students cared a great deal about peer and tutor-given feedback [55, 56]. Providing, reflecting on, and responding to feedback was perceived as helpful for learning and group interaction [55, 56]. In one instance, reflection and feedback on group leadership did not impact leadership behavior, but did increase students’ self-awareness of group roles [56]. Feedback itself was thought to be most effective when received in-person rather than on paper, in the context of clearly defined improvement goals and a willingness to change [48].

Quality of the case-study problems and level of tutorial organization were important to tutors’, but not students’, perceptions of learning effectiveness (KP1) [58]. Case quality was also directly or indirectly important to academic outcomes (KP2) [56, 71]. In one study, tutors reported that case problems, though appropriate for students’ knowledge levels, did not reliably motivate the use of external resources for further learning [58]. Superficial cases and disorganized tutorials were identified in one study as some of the most hindering problems to learning [61].

Discussion

This review identified several key findings on the role of group function in determining PBL outcomes for students at several KP levels.

Tutor and student demographics did not consistently influence experiential (KP1) or academic (KP2) outcomes. Group diversity did not mark a difference at large, either. This finding may be largely because the nature and degree to which these individuals engaged in a group setting were not significantly influenced by sociodemographic variables [30, 33, 42]. For example, the average expert and non-expert tutors were not different in their teaching strategies, degree of direction, amount and nature of social interaction with students, and ability to utilize their expertise [33, 42]. How the group interacts may have a greater impact on PBL outcome than who the group is composed of, though these interactions were highly context-dependent.

Both extra- and intra-tutorial context has major implications in determining the learning environment in which groups engage. External components in some hybrid PBL curricula, such as concurrent lectures and labs, may confound important outcome variables such as GPA [41]. In the included RCT studies, cross-contamination of students from different tutorials was a frequently reported concern [30, 40, 54]. Within PBL tutorials, aspects of group function may be directly important to learning outcomes or moderate the effects of other aspects. For example, the influence of tutor ability on outcome was moderated by curricular factors (e.g., module structure) [43], characteristics of the group [72, 74], or individual student differences (e.g., comfort with PBL [35], self-study effort [34, 35], innate academic ability [41], and prior knowledge [43]). Facilitation expertise was more important to learning outcome in structurally disorganized than well-organized tutorials [43], in junior than senior students [34], and in students with less experience with PBL [35].

Individual student learning behaviors further complicate the effort to sparse out group function-related outcomes. Length of time engaged in self-study was briefly explored in the literature both as an outcome of group function, either as a favorable indicator of increased interest in the learning [34] or an unfavorable indicator of poor group efficacy [72], and as a confounder to group function-dependent outcomes [71]. For instance, rather than group conflict resolution, self-study was frequently cited as the easier and preferred solution for unsatisfactory teamwork experiences [51, 61, 64, 66]. Dysfunctional group members and processes can thus be left unaddressed, yet remain unharmful to academic KP2 outcomes [51, 55, 63, 66]. Concerns over sustaining individual grades and the pressure to pass exams [62, 69] was suggested to be the cause of bypassing group conflict resolution [66]. This may additionally imply challenges for the ability to measure professionally representative behavioral KP3 outcomes in a PBL classroom [63].

The one factor that seemed to overwrite the situational dependence of learning outcomes appeared to be baseline student achievement, which was consistently linked to several levels of KP outcomes. High-achieving students tended to remain high-achievers throughout their studies, suggesting that some stable factors prevail in successful students regardless of the education process [46, 47]. This may include having more general intrinsic motivation or specific interest in the subject matter [41, 46]. High achievers may also be more confident in their knowledge or have higher self-esteem in academic settings. They may thus find more enjoyment in the learning process, be more motivated to learn, and ultimately perform better, in a positive feedback loop. Indeed, confident students demonstrated higher levels of engagement with PBL activities [68], while students who felt less secure in their social or academic positions were more hesitant to engage in “risk-taking” behavior, such as raise contradictory opinions [75, 76], attempt conflict resolution [66], take on leadership roles [66], and instigate new activities [68]. Interestingly, this extended to tutors as well, such that tutors who were confident in their facilitation skills and training found PBL more enjoyable [58] and were better able to optimize their performance [62]. Thus, it seemed that a sense of social and intellectual safety may be important for all group members to maximize their PBL outcomes at any KP level.

Limitations of the literature pool

This review identified several limitations of the literature. Most striking was the variability of studies on this topic, which may help explain the vast array of contradictory findings. Some contextual details are provided for each included study in Table 2. Methodologically, the studies were highly different in sample sizes (15 to 9923 participants), response rates (25% to 100%), and studied PBL topic (basic sciences, anatomy, immunology, psychiatry, epidemiology, nursing, etc.). PBL curricula design was also systematically different across institutions. Examples include duration and frequency of tutorials, number of tutorials per case or problem, sizes of the student groups (3 to 21 students), formation of groups (stratified by background or random), degree of student and tutor training, amount of prior PBL experience, and nature and frequency of assessments and feedback. It is expected that these contextual variations in PBL design and implementation, in addition to cultural considerations [78], may complicate results of comprehensive literature reviews of the literature [9]. It is difficult to determine whether these results may be generalized to any singular institution.

A second limitation was lack of reporting. For instance, most studies did not report how groups were formed (n = 27), which was notable since group composition influences at least some dimensions of learning outcomes. Though perceptions of PBL were largely favorable, gross curricular details and the range of individual tutorial activities were not well reported by most studies. The effect of these aspects on learning outcome were thus not well characterized.

Limitations of this review

Finally, this review has limitations as a function of its methodology. This summarization and analysis of the literature is restricted by its chosen frameworks. For example, the KP model is one of many possible lenses of viewing learning outcomes. Other frameworks may include those focusing exclusively on knowledge acquisition [79], identity related outcomes [80], or social outcomes [81], or those that examine outcome at a group rather than individual level. These may all yield uniquely interesting and relevant results in the context of PBL. The definition of PBL in this review may be overly broad in the attempt to be inclusive. A stricter criterion for PBL structure may yield more consistent results. The framework used for group function is a modified version of Fonteijn and Dolmans [25], though other divisions of group function may exist.

Conclusion

Research on group function in PBL has been broad and diverse, offering a great pool of perspectives on its educational efficacy. This scoping review summarized and structuralized existing literature to provide an organized overview of what aspects of group function are important to individual student outcomes in undergraduate health professional PBL. Such an overview may help educators and researchers navigate this rich field of literature. This review additionally identified the fundamental challenges in linking learning outcomes to group processes, and the gaps where future research may be focused.

The Kirkpatrick framework established experiential (KP1), learning (KP2), and behavioral outcomes (KP3) as independent constructs, which this review found were not reliably correlated to each other in a PBL setting. Key findings showed that student perceptions of PBL and groupwork were generally favorable. Tutor and student demographics did not systematically predict outcomes, and facilitation style and group dynamics were predictors of KP1 but not KP2. Individual student tendencies in KP2 and KP3 persisted. PBL design, such as case quality, tutorial organization, and feedback, were important for KP2 but were not well reported in most studies.

In summary, group function was most directly important to students’ experiences and perceptions of the group and learning (KP1). Knowledge acquisition (KP2) and behavioral (KP3) outcomes were more subject to moderation by stable characteristics in individual students, such as ability, motivation, demotivation, and confidence, as well as external stressors such as exam pressure (Fig. 3).

Fig. 3

Revised concept of group, PBL, and outcomes to incorporate factors that enhance and impair group and PBL processes

Future directions

More research is needed to substantiate the findings of this paper. Reviews of studies at institutions that share similar PBL training and curricula may be required for more conclusive results. Future empirical research is also encouraged to focus on longitudinal behavioral (KP3) outcomes that span the duration of health professional training, or even follow-up into clinical settings, for which there is currently no knowledge. Publications are encouraged to better report the educational, curricular, and training contexts of the institution at which the studies were conducted. Finally, a standardized assessment method for group function in PBL has yet to be developed and may be helpful to future research in this area.