Shared leadership and group identification in healthcare: The leadership beliefs of clinicians working in interprofessional teams.

Despite the proposed benefits of applying shared and distributed leadership models in healthcare, few studies have explored the leadership beliefs of clinicians and ascertained whether differences exist between professions. The current article aims to address these gaps and, additionally, examine whether clinicians' leadership beliefs are associated with the strength of their professional and team identifications. An online survey was responded to by 229 healthcare workers from community interprofessional teams in mental health settings across the East of England. No differences emerged between professional groups in their leadership beliefs; all professions reported a high level of agreement with shared leadership. A positive association emerged between professional identification and shared leadership in that participants who expressed the strongest level of profession identification also reported the greatest agreement with shared leadership. The same association was demonstrated for team identification and shared leadership. The findings highlight the important link between group identification and leadership beliefs, suggesting that strategies that promote strong professional and team identifications in interprofessional teams are likely to be conducive to clinicians supporting principles of shared leadership. Future research is needed to strengthen this link and examine the leadership practices of healthcare workers.


Introduction
The challenges involved in providing patient centred care during periods of financial instability have encouraged organisations to develop new models of delivering healthcare. One consequence has been the development of initiatives that aim to distribute greater leadership influence and responsibilities to frontline clinicians; an approach that differs from traditional hierarchies that locate influence and power with senior managers (Hurley & Linsley, 2007). Recent healthcare models have a clear focus on enhancing the quality and accountability of services while promoting cost effectiveness. Clinicians are argued to have an integral role in achieving these goals and in transforming healthcare services globally (Martin & Learmonth, 2012;Millward & Bryan, 2005). In the United Kingdom, the move towards greater clinician engagement in leadership is evidenced in prominent models of leadership development, such as the healthcare leadership model, which emphasises the relevance of the dimensions of clinical leadership across the spectrum of healthcare workers (NHS Leadership Academy, 2013).
The notion of diffusing leadership responsibilities to frontline workers is consistent with contemporary models in the leadership literature, notably the models of shared leadership and distributed leadership. The promotion of these models is representative of growing interest in exploring the relational and dynamic elements of leadership. Shared leadership involves the collective influence of team members which is embedded in social interactions (Carson, Tesluk, & Marrone, 2007;Currie & Lockett, 2011), whereas distributed leadership refers more explicitly to the distribution of influence to frontline workers (Spillane & Diamond, 2007). These leadership theories are relevant to healthcare, where there is an explicit focus on interprofessional teamwork. Evidence is emerging on the potential benefits of shared and distributed leadership, with several research studies documenting a positive association between these forms of leadership and team performance (Bergman, Rentsch, Small, Davenport, & Bergman, 2012;Ensley, Hmieleski, & Pearce, 2006;Pearce & Sims, 2002;Wang, Waldman, & Zhang, 2014). Comparable findings have been reported in healthcare studies that have demonstrated a link between features of shared/distributed leadership and a number of organisational outcomes including staff empowerment (Barden, Griffin, Donahue, & Fitzpatrick, 2011), staff satisfaction (Sherman & Pross, 2010) and improved service outcomes (Fitzgerald, Ferlie, McGivern, & Buchanan, 2013).
Despite the potential benefits of shared and distributed leadership, few studies have explored the leadership beliefs of healthcare clinicians and whether differences exist between professional groups in interprofessional teams. This is surprising since interprofessional teams, particularly in mental health, are comprised of professions from distinct training backgrounds and diverse philosophical stances (Clark, 1997;Smith et al., 2015). Through processes of group formation, socialisation and identification, healthcare professions are likely to have developed their own specific norms and stereotypes about other professional groups (Stull & Blue, 2016). These norms are likely to influence the beliefs and practices of clinicians belonging to these professions (Hogg, 2001). Social identity theory (SIT) proposes that people develop a group identity based on a shared set of norms, attitudes and behaviours, which results in them perceiving their own groups favourably and differentiating themselves from other groups (Tajfel & Turner, 1979). Group membership could act as a barrier to shared and distributed leadership being incorporated in teams since professions are likely to hold different beliefs about what leadership is and how it should be practiced (Dow, DiazGranados, Mazmanian, & Retchin, 2013). Findings from Steinert, Goebel, and Rieger (2006) provide support for this view as nurses working in psychiatric hospitals expressed greater levels of satisfaction with shared leadership when compared with their physician colleagues. The relationship between leadership beliefs and profession however is unlikely to be straightforward. Research has demonstrated the complexity of leadership and interprofessional team working; professional hierarchy, team characteristics, power and trust have all emerged as influential factors in the literature (Fox & Reeves, 2015;Jones & Jones, 2011;Sims, Hewitt, & Harris, 2015).
Further differences have been outlined between professions in their attitudes to teamwork, preferred leadership roles and status in interprofessional teams (Nembhard & Edmondson, 2006). For example, Liberman, Hilty and colleagues (2001) found that psychiatrists tend to work in a directive manner and view leadership as a prominent part of their role in teams, whereas Cohen (2003) argued that social workers and psychologists favour practices associated with shared decision making and joint working. Comparable findings emerged in Braithwaite et al. (2013), with allied health professionals (i.e. psychologists, speech therapists and physiotherapists) expressing more favourable attitudes to team working and collaboration when compared with physicians and nurses. Differences between nurses and other professions have not always been consistently demonstrated in the literature. Cleary, Horsfall, Deacon, and Jackson (2011) found mental health nurses to favour a leadership style associated with sharing responsibilities. However, in general the accumulation of research in healthcare settings suggests that allied health professionals and social workers will express greater agreement with shared and distributed leadership when compared with other professions.
Research studies have outlined the importance of clinicians maintaining strong identifications with their professions to promote learning and development (Weller, Boyd, & Cumin, 2014). The structural setup of interprofessional teams, with their diverse composition of professions, has been associated with both positive and negative outcomes (Mitchell, Parker, & Giles, 2011). In situations where professional identification is strengthened by intergroup differentiation (i.e. larger differences are perceived between "in" and "out" groups), this has been found to impede joint working practices between healthcare professionals (Lidskog, Löfmark, & Ahlström, 2008). The potential for professional identification processes to undermine features of shared leadership in interprofessional teams is concerning since a lack of teamwork in healthcare settings has been linked to poor patient safety and heightened conflict between workers (Weller et al., 2014). The formation of negative "out group" stereotyping has also been linked to poorer team working between professions (Davies et al., 2011).
Research has indicated that group membership alone is insufficient to determine the leadership beliefs and practices of individuals (Haslam, 2004). More specifically, group membership is likely to influence leadership perceptions when people identify strongly with a group and this membership is salient (van Knippenberg, 2011;van Knippenberg & Hogg, 2003). These elements are a core part of self-categorisation theory (SCT), which outlines the processes that influence whether people define themselves through individual terms or through group membership (Turner, 1985). The core assertions of SIT and SCT have been confirmed in a number of studies examining leadership (Duck & Fielding, 1999;Haslam, Reicher, & Platow, 2010;van Knippenberg & Hogg, 2003) but have been largely absent in healthcare research. These theories offer testable applications as they predict that clinicians' leadership beliefs will be influenced by the norms of their professional groups, particularly for clinicians who identify strongly with their professions. Based on SIT and SCT, psychologists and social workers are likely to report agreement with shared and distributed leadership when they identify strongly with their own professions since these beliefs will likely compliment group norms. The leadership norms of other professions could be incompatible with elements of shared and distributed leadership, suggesting that a strong professional identification will be associated with lower levels of agreement.
Clinicians will experience varying degrees of identification with their professional groups and with their teams, suggesting that both forms of identification are likely to influence their beliefs and practices. SCT contends that individuals can hold a number of group identities simultaneously, a prediction that has been validated in healthcare research that has shown clinicians to express strong identifications with both their teams and professions (Baxter & Brumfitt, 2008). Periods of instability in healthcare services (e.g. cuts, service changes) may pose a threat to duel identifications as this could increase interprofessional rivalry between groups and result in clinicians seeking greater unity with members of their own professions (Hogg, 2001). This assertion is supported by research that has highlighted the negative impact of threat to professional identity on the strength of clinicians' identification with their teams (Mitchell et al., 2011).
This article aims to: 1) explore whether there are differences between professions in their beliefs about shared and distributed leadership; and 2) ascertain whether SIT and SCT can help predict the leadership beliefs of professional groups. The current authors predicted that psychologists, social workers and occupational therapists would report greater agreement with shared and distributed leadership when compared with consultant psychiatrists. Additionally, it was hypothesised that a significant association would emerge between the strength of clinicians' professional identification and their leadership beliefs i.e. a strong professional identification will be associated with leadership beliefs that are congruent with the norms of the professional group clinicians belong to.

Design and participants
A cross sectional online survey was developed to explore the leadership beliefs of healthcare clinicians working in interprofessional teams in the East of England. Quantitative data was obtained from participants through the use of closed/multiple choice questions and questionnaires featuring Likert scales. Three hundred and thirty six participants provided consent to take part in the study. A number of participants dropped out after completing a few questions (n = 95), resulting in a dropout rate of 28%. Of the remaining 241 participants, 12 clinicians did not meet the eligibility criteria and their responses were excluded from the study. Overall, a total of 229 participants were included in the study.
The majority of the 224 participants who provided their gender were women (n = 165, 74%) and predominantly worked in mental health teams (n = 207, 90%) across child, learning disability, adult and older adult settings. Table 1 displays the frequencies and percentages of participants from each profession. Based on 223 responses (6 participants did not respond to the question), 76 participants (34%) had worked in their interprofessional teams for two years or less, 42 participants (19%) between two years one month and five years, 56 (25%) participants between 5 years one month and 10 years, and 49 participants (22%) for over ten years. A similar trend emerged for the number of years participants had been qualified in their professions ( Figure 1). Approximately a quarter of the participants were team leaders of their interprofessional teams (n = 54, 24%) and the majority of participants had experienced at least one form of service change in the previous year (n = 185, 81%).

Eligibility criteria
Participants were required to be qualified healthcare clinicians working in interprofessional teams in mental health, neuropsychological or clinical health settings. Individuals working solely in inpatient settings were excluded from the study due to the nature of shift work in these settings which could limit the scope for workers to develop cohesion and a team identity, in addition to the current authors wanting to promote heterogeneity in the sample.

Recruitment strategies
Participants were recruited from five trusts in the UK's National Health Service (NHS) between October 2014 and February 2015. A number of senior healthcare managers forwarded recruitments e-mails to clinicians in their organisations. This was supplemented by clinical contacts of the researchers sending recruitment e-mails to staff members in their services. E-mails were evenly spread across the recruitment period to capture new staff members. Team leaders were also contacted and a number of these individuals forwarded recruitment e-mails to clinicians in their teams. Personnel from the communication departments of all five NHS trusts provided a link to the survey in their newsletters.

Measures
The Leadership Attitudes and Beliefs Scale (LABS) was used to evaluate leadership beliefs (Wielkiewicz, 2000). The LABS is a 28item measure, allowing respondents to select five possible responses ranging from 1 (strongly agree) to 5 (strongly disagree). The LABS measures two leadership dimensions: "Hierarchical Thinking" and "Systemic Thinking." Items in the "Hierarchical Thinking" dimension are phrased in support of the view that leadership influence should reside with individuals in positions of authority. A low score indicates agreement with hierarchical leadership and disagreement with distributed leadership. Items in the "Systemic Thinking" dimension are phrased in support of the view that leadership influence should be shared between team members and conveys the relational elements of leadership. A low score in this dimension indicates agreement with shared  leadership. The current study demonstrated alpha coefficients of .80 for the Hierarchical Thinking dimension and .90 for the Systemic Thinking dimension. Permission was provided by the author to use the questionnaire. A group identification measure from Brown, Condor, Mathews, Wade, and Williams (1986) was used to measure professional identification and team identification. With permission, five items were taken from the ten-item questionnaire. Professor Brown provided advice in adapting this questionnaire. One example item from the original measure is: "I am a person who identifies with the group." This item was changed to: "I identity with my profession/team." Participants rated their level of agreement with each statement by responding on a five-point scale ranging from 1 (never) to 5 (very often). The current study demonstrated Cronbach's alpha coefficients of .85 for the professional identification measure and .90 for the team identification measure.
Threat to professional identity was measured by adapting a validated six-item measure from Ethier and Deaux (1990). Permission was provided by the authors to adapt their questionnaire. An example item is: "I try not to show the parts of me that are ethnically based." This item was changed to: "I avoid showing the parts of me that are connected to my profession to other team members." Participants rated their agreement with each statement by responding on a seven point scale ranging from 1 (not at all) to 7 (a great deal). Three items from the original questionnaire were adapted and an additional two items were developed by the current researchers. The professional threat scale used in the current study reported a high level of internal reliability, Cronbach's alpha of .85.
Demographic information was also obtained from participants including gender, staff grade, the number of years qualified in their profession and work setting. Participants were also required to rate their familiarity with the NHS Healthcare Leadership Model (NHS Leadership Academy, 2013). The Healthcare Leadership Model features nine dimensions and encourages healthcare workers to become more effective leaders, irrespective of seniority.

Data analysis
Analyses were completed using SPSS Version 22 for Windows. ANOVA tests were used to explore differences between professional groups in their leadership beliefs. The assumptions of normality and equal variances were achieved when examining the Hierarchical Thinking dimension scores of all professional groups. The same assumptions were not confirmed when viewing the distribution of participants' Systemic Thinking dimension scores due to the presence of an outlier in one group. Removing this outlier resulted in a normal distribution in this group and equal variances between professions.
Measures of group identification and professional threat were treated as ordinal data due to concerns about the symmetry of the response options. Non parametric correlation analyses could not be undertaken since the assumption of a monotonic relationship between variables was not met. Participants' responses on each questionnaire were divided into binary categories of scores that fell above and below the median as the data from each measure were skewed at two points. The scores of participants on the LABS were similarly categorised into "high" and "low" scoring groups in order that Fisher's exact tests could be used to examine the association between group identification variables and leadership beliefs.
Logistic regression analyses were completed to determine the impact of a number of dichotomous predictor variables on participants' categorised LABS scores. The regression analyses also helped examine the effects of possible confounding variables on the association between profession and leadership beliefs. All the necessary assumptions were met for the logistic regression analyses.

Ethical considerations
The study received ethical approval from the Health and Human Sciences committee at the University of Hertfordshire.

Profession and leadership beliefs
The mean score for all participants, irrespective of profession, was 45 in the Hierarchical Thinking dimension (n = 214, SD = 6) and 26 in the Systemic Thinking dimension (n = 214, SD = 5.5). Fifteen participants were not included in these analyses as they completed only a few items from the LABS questionnaire.
Tables 2 and 3 provide the descriptive statistics separately for the five largest professional groups in the study. Of the 204 participants who made up the largest professional groups, 14 were excluded from analyses as these participants answered fewer than 7 items from the 28-item questionnaire. This left a total of 190 participants in the largest professional groups (N = 190). One outlier response was removed from the analysis of the mean Systemic Thinking dimension score of the psychiatric nursing group as this response was greater then three standard deviations from the mean score of this group (i.e. N = 189).

Group identification and leadership beliefs
The majority of participants reported strong identifications with their teams (Mdn = 22, maximum score of 25) and their professions (Mdn = 22.5, maximum score of 25). The majority of participants also reported a low level of threat to their professional identities (Mdn = 10.5, maximum score of 35). Professional identification and participants' Hierarchical Thinking dimension scores were not found to be significantly associated, p = .68, Fisher's exact two-sided test (n = 213). Separate analyses for each of the five largest professions did not report any significant associations. Similarly, no significant association emerged between team identification and Hierarchical Thinking dimension scores, p = .41, Fisher's exact two-sided test (n = 214).
A significant association was found between professional identification and participants' Systemic Thinking dimension scores, p < .001, Fisher's exact two-sided test. The odds of participants being assigned to the "low scoring" Systemic Thinking dimension group was 2.35 times higher for clinicians who reported the "strongest professional identification" than for clinicians who reported a "weaker professional identification." Separate analyses were completed for the five largest professions but these analyses did not report any statistically significant associations. The same trend emerged for each profession, participants were more likely to be assigned to the "low scoring" Systemic Thinking dimension group when they reported the "strongest professional identification." Team identification and Systemic Thinking dimension scores were significantly associated, p = .04, Fisher's exact two-sided test. The odds of participants being assigned to the "lower scoring" Systemic Thinking dimension group was 1.76 times higher for clinicians who reported the "strongest team identification" than for clinicians who reported a "weaker team identification."

Group identification associations
Professional identification was positively associated with team identification, p = .03, Fisher's exact two-sided test. The odds of participants reporting the "strongest team identification" was 1.89 times higher for clinicians who reported the "strongest professional identification" than for clinicians who reported a "weaker professional identification." This association was only evident when participants were assigned to the "lowest professional threat" group, p = .01, Fisher's exact one-sided test.
Threat to professional identity was negatively associated with both team identification (p < .001, Fisher's exact one-sided test) and professional identification (p = .03, Fisher's exact two-sided test). The odds of participants reporting the "strongest team identification" was 4.13 times higher for clinicians who reported the "lowest professional threat" level than for clinicians who reported a "higher professional threat" level (Table 4).

Regression analyses
Variables were dummy coded and placed into the regression models in order of relevance to leadership. Only one variable, professional group (psychiatric nurses), was significant in predicting the categorised Hierarchical Thinking dimension scores of participants (b = −1.04, SE = 0.4, p = .02, Exp (b) = 0.36, 95% CI: 0.15-0.83). This factor predicted approximately 8% of variation in the Hierarchical Thinking dimension scores of participants (Nagelkerke R 2 = .079). The odds of reporting a Hierarchical Thinking dimension score in the "high scoring" category was 0.64 times lower for psychiatric nurses when compared with other professions.
Three variables proved to be significant in predicting the categorised Systemic Thinking dimension scores of participants (Table 5). This model predicted approximately 15% of variation in the scores of participants (Nagelkerke R 2 = .147). When including the effects of gender and professional identification, the odds of reporting a Systemic Thinking dimension score in the "higher scoring" category was 2.8 times higher for participants who reported a moderate or unfamiliar rating with the Healthcare Leadership Model. When including the effects of familiarity with the Healthcare Leadership Model and professional identification, the odds of reporting a Systemic Thinking dimension score in the "higher scoring" category was 2.3 times higher for men than women.

Supplementary analyses
While gender was predictive of participants' categorised Systemic Thinking dimension scores in the regression model, an independent t-test test did not report significant differences between men and women in their leadership scores, t (209) = −1.66, p = .098, d = .22. No significant differences emerged between professions in their familiarity with the Healthcare Leadership Model, p = .23, Fisher's exact two-sided test.

Discussion
Contrary to predictions, no differences emerged between professions in their beliefs about shared or distributed leadership. In Mean values, 95% confidence intervals, standard errors, and standard deviations. general, all professions reported a high level of agreement with shared leadership. These findings contradict previous research that has highlighted differences between professions in their attitudes towards interprofessional working, their team practices and leadership beliefs (Atwal & Caldwell, 2005;Braithwaite et al., 2013;George, Frush, & Michener, 2013;Steinert et al., 2006). The discrepancy could be explained by methodological differences between the research studies and the specific variables under investigation. Atwal and Caldwell (2005) used a direct observational design to record the nature of interactions in interprofessional teams and therefore, they were examining the actual behaviours of healthcare professionals. While Braithwaite and colleagues (2013) reported differences between professionals in their attitudes, items from their questionnaires explicitly focused on healthcare workers' beliefs about the perceived benefits of interprofessional working and collaboration. In contrast, the items used to measure attitudes to leadership in the current study (LABS questionnaire) featured less overt emphasis on the perceived benefits to services and patients. The items adapted from the LABS questionnaire could have resulted in less emotive responses from participants, potentially explaining the lack of differences that emerged between professions. However, the similarities expressed by various professionals in their leadership beliefs could also represent a shift in the attitudes of professional groups in recent years and the increasingly interdependent nature of healthcare delivery in interprofessional teams. While the results of the current study require replication with larger group sizes, the findings suggest that healthcare professionals are likely to hold positive views about shared leadership. This is an encouraging finding given the emphasis on shared and distributed leadership in contemporary health service initiatives, in addition to the positive outcomes associated with concepts such as shared purpose, interprofessional working and the transfer of knowledge across professions (Bateman, Bailey, & McLellan, 2003;McComb, 2013). These concepts, while obviously distinct entities, compliment key elements of shared leadership.
A statistically significant association emerged between the strength of professional identification and participants' Systemic Thinking dimension scores. Participants who reported the strongest professional identification were more likely to express the greatest level of agreement with shared leadership when compared with participants who reported a weaker professional identification. The same trend emerged for each profession separately although none of these analyses reached the .05 level of significance, likely due to the reduced sample sizes and the associated loss in power. These findings contradict the notion that a strong professional identification will impede clinicians' openness to work together and share decision making in teams; a finding that is inferred from the results of Stull and Blue's (2016) study that examined the influence of professional identity on the attitudes of healthcare students towards interprofessional learning. The incongruence in results could reflect the different responsibilities and pressures experienced by students and qualified healthcare clinicians. In addition, healthcare students and qualified clinicians are likely to be at different stages of identity formation within their professions.
The significant association between professional identification and shared leadership beliefs provides partial support for social identity theory (SIT) and self-categorisation theory (SCT), since these theories predict that people's leadership beliefs will be influenced by the strength of their group identifications. However, these theories also state that the nature of these beliefs is influenced by group norms. The trend that emerged for consultant psychiatrists contradicts the hypothesis that a strong professional identification in this profession would be associated with lower levels of agreement with shared leadership. This suggests that the leadership norms of consultant psychiatrists warrant further exploration as the results from this study suggest that their norms may be congruent with aspects of shared leadership. This assertion compliments the findings of Gair and Hartery (2001) as they highlighted the important role played by psychiatrists in facilitating shared decision making in interprofessional teams. Furthermore, the favourable attitudes expressed by psychiatrists towards shared leadership in the current study align with broader attempts within psychiatry to improve services by encouraging greater sharing of leadership roles and responsibilities in teams (Bhugra, Ruiz, & Gupta, 2013).
No significant associations emerged in the Hierarchical Thinking dimension, with high mean scores across groups indicating broad disagreement with hierarchical approaches to leadership. This outcome could be related to the content of this dimension as it focuses on themes of authority and control. These themes may attract greater uniformity in the responses from clinicians, who are likely to disagree with statements that undermine clinical judgement and autonomy.
The positive association that emerged between team identification and agreement with shared leadership suggests that a strong team identification is conducive to clinicians holding positive views about shared leadership. This conclusion compliments previous research that has demonstrated a positive association between team identification and attitudes to interprofessional working (Mitchell et al., 2011). The emergence of a collective team identity in healthcare teams is likely to reduce interprofessional rivalry and lessen professional boundaries by preventing the emergence of subgroups forming (Hobman & Bordia, 2006;McNeil, Mitchell, & Parker, 2013). This offers specific relevance to mental health settings where interprofessional teams feature a wide range of professional groups who have received distinct forms of training and operate from different epistemological positions. Reconciling these differences is unlikely in mental health teams where a collective identity is absent, which ultimately could prove detrimental to the application of shared leadership and team working in this setting.
The majority of participants reported strong identifications with their professions and their teams. This is congruent with previous research and supports the assertion that individuals can hold a number of identities simultaneously and these identities can be complimentary to both the values of the team and individual professions (Jones & Jones, 2011;Tajfel & Turner, 1986). Since both professional and team identification were positively associated with higher levels of agreement with shared leadership, strategies that encourage duel identifications are likely to promote a culture of shared leadership in interprofessional teams. This emphasises the importance of promoting group identifications in healthcare as clinicians are more likely to agree with shared leadership practices when they express strong identifications with their teams and professions. This aspiration is not without its challenges as large caseloads and service targets could result in healthcare organisations prioritising clinical contact over team development practices. These pressures would likely undermine attempts to develop a shared purpose in teams and hinder effective collaboration between professionals (O'Carroll, McSwiggan, & Campbell, 2016). Without adequate resources and time being allocated to team development, teams are likely to become less cohesive and more fragmented in their delivery of care to patients. Furthermore, team leadership is also likely to influence the extent to which team members identify with their teams (Huettermann, Doering, & Boerner, 2014), providing evidence that the relationship between team identification and leadership is not unidirectional. This point is further illustrated by research that has shown leader inclusiveness 2 to be crucial in promoting a shared team identity in healthcare teams, particularly when teams are comprised of a large number of professional groups (Mitchell et al., 2015).
The current study reiterates the importance of clinicians feeling secure in their professional identities as the positive association between profession and team identification was only evident when the level of professional identity threat was low. This highlights the detrimental impact of professional identity threat on clinicians' identifications with their teams, an important finding when considering the benefits of developing a shared team identity in healthcare and effective interprofessional working (Kreindler, Dowd, Star, & Gottschalk, 2012;O'Leary, Sehgal, Terrell, & Williams, 2012). These findings offer a valuable contribution to the literature as they indicate that strong team and professional identifications will increase the likelihood of clinicians agreeing with shared leadership and conversely, high levels of professional threat will likely undermine this process.
The current study has some limitations that need to be highlighted, firstly with reference to the sample size. The number of participants recruited to the study was modest and future survey research in this area would benefit from recruiting a larger number of clinicians. Doing so could enhance the representativeness of the sample and allow linearity to be demonstrated in the relationship between group identification and leadership beliefs, as few clinicians in the current study expressed weak group identifications and a high level of professional threat.
Secondly, although the findings of the study are promising in establishing a link between group identification and agreement with shared leadership, it cannot be assumed that clinicians would act in full accordance with their beliefs. This provides an interesting avenue for future research as previous studies have highlighted the complexity involved in the relationship between beliefs and practices (Ajzen & Fishbein, 2005). Establishing a strong link between shared leadership attitudes and practices in healthcare would emphasise the importance of promoting a strong team identity in interprofessional teams.
Finally, while the LABS questionnaire has been validated in organisational settings, it has not been applied in healthcare studies and therefore it requires further evaluation to measure its convergent validity with other measures of shared leadership.

Concluding comments
Despite the limitations of the current study, the findings are important in demonstrating a link between group identification and clinicians' leadership beliefs. This link has been validated extensively in social psychology but represents a novel finding in healthcare. The current study reaffirms previous research that has demonstrated it is possible for clinicians to hold strong duel identifications with their teams and professions in situations when the level of professional threat is low. Strategies that help promote dual identifications in interprofessional teams will be important in reducing the likelihood of clinicians feeling their professional identities are under threat. This offers particular relevance in community mental health settings where recent policies have arguably sought to lessen specialised roles between professions and create greater uniformity in the practices of clinicians. While this approach could encourage greater mobility and flexibility in the delivery of healthcare services, should clinicians feel threatened by these changes it is likely to undermine the emergence of strong team identities and the associated benefits of interprofessional teamwork and collaboration. Furthermore, the findings of the current study suggest that a strong team identification is required to promote favourable attitudes to shared leadership. Without the necessary resources and attention being paid to promote a collective identity in healthcare teams, initiatives that aim to share leadership influence among frontline clinicians are unlikely to transfer into actual practices. Notes 1. N = 225 in Figure 1 as four participants did not respond to the question. 2. Leader inclusiveness was defined as the extent to which leaders encourage and value the expression of different viewpoints in teams.