Perceiving Agency in Sustainability Transitions: A Case Study of a Police-Hospital Collaboration

: This paper explores how agency was used within a police-hospital collaboration to implement a planned change designed to increase the sustainability of a cross-sector collaboration. A longitudinal, qualitative case study involving pre-and-post interviews with 20 police o ﬃ cers and 20 healthcare workers allowed us to capture multiple perspectives of the planned change over time. Analysis of case study data reveals three major ﬁndings: (1) organizations with limited power can have agency in cross-sector collaborations when they are perceived to have legitimacy and urgency; (2) the extent to which the implementation of a planned change inﬂuences perceptions of agency depends on the organizational context of the perceiver; and (3) di ﬀ erent levels of analysis (i.e., meso versus micro) support di ﬀ erent conclusions with respect to the role of agency in the sustainability transition process. More broadly, our study highlights the role of perception when investigating agency within sustainability transitions.


Introduction
Finding solutions to many of today's pressing social problems, such as community mental health care that requires public, private, and/or not-for-profit organizations to collaborate with each other [1][2][3]. Such cross-sector collaboration requires "the linking or sharing of information, resources, activities, and capabilities by organizations in two or more sectors to jointly achieve an outcome that could not be achieved by organizations in one sector separately" [4] (p. 44). In theory, cross-sectoral collaborations enable participating organizations to address "wicked" problems more effectively by unlocking the benefits of competitive advantage [3]. To realize these benefits, however, they have to overcome the complexities and difficulties associated with these types of partnerships [3,5], as "there is a fine balance to be struck between gaining the benefits of collaborating and making the situation worse" [5] (p. 117). Many cross-sectoral collaborations have been found to be unsustainable over time [6], as firms are unable to juggle the advantages with the disadvantages of such partnerships [5].
The growing dependence on cross-sectorial partnerships, along with the high failure rate of such collaborations [5], increases the need for research that explores how these partnerships remain sustainable over time. Scholars increasingly cite agency (or lack thereof) as both an obstacle [7] and driver [8] of sustainability. While agency has been traditionally associated with higher level structures, such as politics and governance [8], there is growing evidence that other forms of agency (e.g., individual agency) can improve sustainability efforts [9,10]. Despite the growing body of research in this area, the link between agency and sustainability remains highly debated [11]. In addition, relevant to this present study are arguments made by researchers that most sustainable studies do not take into account psychological theories [12] or investigate behavioral dynamics [13,14]. In this paper, we maintain that the growing demand for cross-sector partnerships, the high failure rate of such collaborations [5], and lack of consensus on the role of agency in sustainability [11] increases the need for sustainability researchers to: (1) utilize broader conceptualizations of agency in their research designs and (2) explore how perceptions of agency impact the sustainability of cross-sector collaborations over time.
This paper reports on a longitudinal (two year) case study that examines how partners from two different sectors (health care, policing) engaged in a planned change process designed to make their collaboration more sustainable over time. These two organizations had worked together for many years dealing with the needs of persons with mental illness (PMIs) in a large Canadian community. Changes in the external environment, however, shifted the balance of the relationship between these two organizations so that one partner (the police) felt that the collaboration was no longer sustainable and desired change. To address these issues, the police service organized and drove a radical transformation to make the cross-sector collaboration more sustainable. In other words, these organizations were working towards a "sustainability transition" [9]. This paper examines how end-users from both organizations perceived agency changing over time as the sustainability transition unfolded. Three theoretical lenses were used to inform how we set up our study and analyzed our data: cross-sector collaborations [4], sustainability transitions [8], and stakeholder theory [15].
Transitions research attempts to understand the relationship between the "desire for change and forces for stability" in a system, and seeks to "conceptualize and explain how radical changes can occur in the way societal functions are fulfilled" [8] (p. 2). In their recent comprehensive review of the sustainability transitions field, Köhler and colleagues identify a number of areas that require future research, many of which we took to heart when designing this study.
First, Köhler et al. [8] note that, although agency is an important instrument for transitions, sustainability researchers have used limited a limited number of conceptualizations of this construct in their research [16]. Consistent with growing evidence that individuals [9,10] and groups [17,18] with little power can affect transitions, our study addresses this issue by examining a case where one partner (i.e., the police) with limited power is able to drive change within a cross-sector collaboration. Second, Köhler and his colleagues' [8] recommend that sustainability researchers widen their scope of study to more than a single system to allow for the fact that the social and environmental problems that require transitions affect many different sectors. While much of the transitions literature focuses on private sector settings [8], this study investigates a transition toward social sustainability by public sector organizations. More specifically, our study focuses on a multi-sector transition involving health and police systems. Third, there is a need for transitions research at the meso (e.g., organization) and micro (e.g., individual) [19][20][21] levels of analysis to supplement the already large body of transitions research that takes a macro-level perspective [8]. Our research uses both meso-level (i.e., police, hospital) and micro-level (i.e., end users in the two organizations who were intimately involved in the change) analyses to grow our understanding of sustainability transitions. Finally, by adapting a framework that is used by those working in the areas of strategy and management (i.e., Stakeholder Theory [15]) to frame our conceptualization of agency, we respond to Köhler et al.'s [8] recommendation that sustainability researchers should broaden their use of frameworks to mobilize insights from other social science fields. The use of Stakeholder Theory in this study means that our findings have applicability for both sustainability and business domains.
Our paper unfolds as follows. Section 2 reviews theory that influenced our thinking and interpretation of the data. We then, in Section 3, provide the reader with relevant details about the case study on which this research is based, followed by a description of how we collected and analyzed our data. Sections 4 and 5 present our results, and Section 6 discusses our findings. The final section of the paper, Section 7, outlines our key conclusions, summarizes the limitations of our work, and offers suggestions on how other researchers could build on our study.

Cross-Sector Collaborations
Cross-sector collaborations are generally seen as a means to improve public value [22]. Much of the research in this area focuses on what makes these types of collaborations successful at the partnership level [23,24]. One insight gleamed from this body of research is how challenging it can be to sustain stable cross-sector collaborations over time as partners find that their organizational incentives, structures, modes of governance, goals, and processes ultimately do not fit with those of their cross-sector partners [25][26][27]. Other research provides support for the idea that these differences may lead one or more of the partners to act with agency and put their needs above their partners' needs [4,28].
Our review of the literature determined that police services and hospitals frequently work together to serve the needs of their community. While research in this area has traditionally focused on police-hospital partnerships connected to crime-reduction initiatives [29], recent research has focused on collaborations relating to mental health care in the community. In 2018, Parker and colleagues completed a systematic review of 125 empirical studies on this topic and concluded that most researchers working in this area study the enactment of tactical elements related to police-hospital partnerships (e.g., pre-arrest diversion of persons with mental illness, jail diversion following arrests) [30]. Parker et al. [30] did not, however, identify any research focusing on the sustainability of police-hospital collaborations over time. Our study, which explores how a police-driven change impacted cross-sector collaboration over time, increases our understanding of the challenges of sustaining such partnerships.

Sustainability Transitions
Sustainability transitions are multidimensional, multi-actor processes that involve many kinds of agency, a high degree of uncertainty, and considerable conflict, as actors/social groups tend to disagree about the most desirable transition pathways for sustainability [8]. Köhler et al. [8] point out that transitions are "inherently political processes in the sense that different individuals and groups will disagree about desirable directions of transitions, about appropriate ways to steer such processes and in the sense that transitions potentially lead to winners and losers" (p. 6). They go on to recommend researchers who study transitions to pay attention to "who gets what, when and how." In other words, they need to make note of who wins and who loses, when changes emerge, and when those changes are implemented. These ideas support the need for researchers to examine transitions using both meso and micro perspectives; the approach taken in this paper.
It is well established in the literature that transitions involve various manifestations of agency [4,8,31,32]. While a considerable body of research exists investigating what makes agency successful in enacting change at macro levels [33][34][35], our knowledge on the role of agency is limited by the fact that few sustainability researchers consider individual context [36][37][38], individual agency [39][40][41][42], and/or the dynamics of agency development [43].
In 2010, Geels argued the need for Multi-Level Perspectives of agency within sustainability transitions research [44]. This "call to arms" resulted in a growing number of recent transitions studies that examined agency from a micro level. For example, Antadze and McGowan [9] show how moral entrepreneurs can trigger macro changes by gradually disassociating rules and practices from their moral foundations through engaging in a macro-systemic discourse. Pesch, Vernay, van Bueren, and Pandis Iverot [10] highlight the importance of actors who connect elements that are needed to develop successful niches (e.g., project champions who build coalitions and safe spaces for learning). Bögel and Upham [12] examine how technology users can bring about sustainability change based on their consumption and acceptance of technology. In line with growing conceptualizations of agency, Köhler et al. [8] suggest that researchers need to continue to expand our understanding of agency in the transition process by: (1) using frameworks from other disciplines, (2) exploring how power manifests itself at different levels of the system, (3) undertaking in-depth empirical case-study analyses that explore agency from different perspectives, and (4) by scrutinizing "the (un)intended political implications of transition processes" [8] (p. 8). These recommendations were adhered to when designing the case study reported on in this manuscript.

Agency: A Stakeholder Salience Perspective
Recently a number of scholars have conceptualized agency using a variety of micro lenses including morality [9], behavior [10,12], and subjective experience [13]. Consistent with these trends, in our study we utilize Mitchell, Agle, and Wood's [15] theory of stakeholder salience to conceptualize agency based on how actors perceive one another. Mitchell, Agle, and Wood's [15] theory describes how the perception of three attributes-power (i.e., the ability to get an actor to do something), legitimacy (i.e., the perception that actions are appropriate), and urgency (i.e., the perception that actions require immediate attention)-can be used to identify and prioritize the claims of relevant stakeholders, defined as "any group or individual who can affect or is affected by the achievement of an organization's purpose" [45] (p. 53). Mitchell et al. [15] define stakeholder salience as the degree to which managers give priority to competing stakeholders' claims in their decision-making process.
A stakeholder perspective describes how groups and organizations can influence others in their system [46]. According to this framework, stakeholders may be categorized into classes based on whether or not they possess (or are perceived to possess) one, two, or all three of the above attributes. The more attributes that the stakeholder has, the more salient their concerns. While Mitchell et al. [15] define seven stakeholder combinations, we identified only four that have relevance to this research: (1) definitive stakeholders possess all three attributes (i.e., power, legitimacy, urgency) and are presumed to be the most salient partner in the collaboration; (2) dominant stakeholders possess power and legitimacy, as well as typically have some formal mechanism in place to help them act on their desires; (3) dependent stakeholders have legitimacy and urgency but little power to obtain what they need; and (4) discretionary stakeholders have legitimate needs and demands, but the fact that they have no power or urgency means that often get little attention from other stakeholders.
Despite vast support for Mitchell et al.'s theorization, scholars have critiqued their static depiction of stakeholder salience, arguing that stakeholder salience may be dynamic, as stakeholders' attributes change over time [47][48][49] in response to shifts in the external environment [50]. This study, which looked at how a planned change designed to improve sustainability impacted stakeholder salience over time, addresses this gap in our understanding.
In this study, we conceptualize agency as the extent to which an actor is perceived to have power, legitimacy, and urgency. This view is supported by Stakeholder Theory and consistent with the sustainability transition literature which has previously examined: (1) how power (and to much less of an extent legitimacy and urgency) can be exercised by different actors [8] and (2) how several other forms of agency can be used [9,10,12,13] to achieve or obstruct sustainability transitions. Regardless of how agency is conceptualized, however, the sustainability transitions literature predominantly investigates how agency impacts transitions [43]. In this paper, however, we explore the reverse: how the transition process affects perceptions of power, legitimacy, and urgency in a cross-sector collaboration over time. We anticipate that our study will give those working within the sustainability transition domain a better understanding of the interconnected relationship between agency-operationalized as perceptions of power, legitimacy and urgency-and planned change within cross-sector collaborations over time [51,52].

Research Methodology
Our research uses a longitudinal, in-depth, single case-based approach to explore how the implementation of a planned change designed to improve the sustainability of a police-hospital collaboration impacts perceptions of power, legitimacy, and urgency within this partnership over time. Our approach is consistent with that used by most sustainable transition researchers [8] (p. 18), and "fits assumptions of complex causation, emergent realities and non-linear development trajectories." We begin by describing the case on which this study is based. We then discuss how our data were collected and analyzed.

The Case
This research reports on a cross-sector collaboration involving a police service and a hospital operating in a large city in Canada. The police service participating in this case study has over two thousand sworn and one thousand non-sworn members. The hospital has 600 beds, 900 physicians, 4300 staff, and 1000 volunteers. Both organizations are mandated to work together by the Ontario Mental Health Act (OMHA), which regulates the involuntary admission of people into a psychiatric hospital for a 72-h observation period. The focus of this research is on a change to one particular aspect of the police-hospital collaboration, a change to the PMI process.
The PMI Process unfolds as follows. Police officers respond to calls for service based on a 1 to 5 priority system, where 1 is prioritized most urgent and typically involves immediate and emergent threats of endangerment. Calls for service involving PMIs are generally prioritized as a one. Typically, when the police receive a call for service involving a PMI, they need to respond immediately to manage the crisis. After the police officers first arrive at the scene, they assess the situation. If the officers feel the PMI is a risk to themselves and/or others, they are required by Section 33 of the OMHA to bring the PMI to the nearest hospital and wait with them until a doctor becomes available. Once the doctor assesses the PMI, the police transfer the custody of this individual to the hospital and are free to leave.
The hospital involved in this case study has a very busy emergency department (ED) with over 500 visitors a day. The cumulative length of time a PMI waits to see an ED doctor or specialist (i.e., wait time) is determined by three factors: (1) the number of patients more seriously ill than the PMI; (2) the number of tests that are required; and (3) whether the patient needs a consultation with a specialist, such as a psychiatrist. The Canadian Triage and Acuity Scale (CTAS) is used by the hospital during this stage of the PMI process to prioritize the need for the PMI to see a physician. This is problematic as mental health patients "present a unique challenge for triage" and there is only a vague understanding of how to prioritize physical injuries and mental health cases in relation to each other [53] (p. 229). This challenge contributes to confusion with respect to how to prioritize and manage PMIs brought into the ED by the police.
The police are required by law to wait at the hospital until a doctor psychiatrically assesses the PMI and either admits them to hospital or discharges them. While this law ensures that the transfer of PMI custody from the police to the hospital is seamless, it also gives healthcare workers (specifically ED physicians) the power to dictate how long police officers will be delayed at the hospital before they could return to their regular service. Anecdotally, we heard from many at the hospital that they liked having the police in the ED as it encouraged other people who were waiting for treatment to be on "their best behavior." Not surprisingly, this waiting period results in a considerable loss of police resources (i.e., person and vehicle hours). The hospital is also dependent on the good will of the police, as they are required by law to assess all PMIs brought into the ED by the police even if the PMI does not require medical attention. The more PMIs brought into the hospital by the police, the more hospital resources are devoted to people that the police (but perhaps not the medical staff) deem to require care.
Over the past several decades, several major changes in the external environment have conspired to make the PMI process unsustainable. The closure of mental health hospitals [54] and the reduction of social service funding [55] has increased the number of PMIs that the police and health care providers have to manage within a given year, and increased the financial and reputational costs of managing the PMI process for both organizations. Police services are responding to more calls for service involving PMIs now than in the past [56]. This has, in turn, increased the PMI patient load at the hospital to the extent that health care providers are now less able to treat PMIs in a timely manner. The trickle-down effect of all of this is that the police are now being made to wait at the hospital for a longer time period of time than in the past. When this case study began, the police calculated that the average wait time for a police-escorted PMI to see an ED physician at the hospital was over 4 h. The direct cost of this wait was estimated to be $1.1 million annually. The police also drew a link between the PMI process and a decline in service to the community, noting officers waiting at the hospital were not available to handle other calls in the community.
The high costs associated with their collaboration in the PMI process drove the Deputy Chief of police to initiate a planned change involving both the police and the hospital. The goal of the planned change was to collaboratively determine the best way to manage PMIs so that it was sustainable for both organizations over time. Top decision-makers from both organizations agreed on a pair of complimentary changes: one to be implemented by the police organization, the other by the hospital. It was agreed by both partners that: (1) police officers would complete a background information form about the PMI prior to their arrival at the ED in order to facilitate information transfer from police officers to healthcare workers, and (2) the hospital would implement "an escalation and intervention policy" requiring ED doctors to make every effort to assess PMIs within 2 h of the police officers bringing them into the ED. If this time deadline was not honored, the police officer's sergeant would escalate the situation to involve the hospital's senior administration.
One of the authors of this paper had worked with the police service in the past and was invited to undertake research on how these changes unfolded over time. This presented us with the opportunity to observe the phenomenon of interest for this case study-a cross-sector collaboration that was introducing a planned change in order to make the system more sustainable over time-an important factor when selecting cases for case study research [57]. Focusing the research on a particular setting has the advantage of minimizing the risk that sources of extraneous variation will conflate the findings [58]. The engagement and willingness of the organizations to participate in the study facilitated the use of the qualitative techniques that are recommended for process-based change research [59] and meant that the longitudinal data necessary for addressing the research objectives was accessible.

Data Collection
Interview data were collected from both police officers and healthcare workers to improve theory-building potential [58]. We conducted semi-structured interviews with 20 police officers and 20 healthcare workers prior to the introduction of the planned change to the cross-sector collaboration (Time 1). We did a second set of interviews with the same 40 individuals one year after the implementation of the change (Time 2). The interview process was designed to help the researchers better understand: (1) how organizational partners viewed each other with respect to stakeholder attributes (i.e., agency) and (2) how the implementation of the planned change changed views of agency over time. The police sample consisted of 14 constables (i.e., frontline police officers) who brought PMIs to the hospital and the 6 officers (i.e., sergeants, staff sergeants) who supervised their work. The hospital sample included 6 ED nurses, 5 ED physicians who treated the PMIs once they were brought to the hospital, 4 security guards within the ED, and 5 administrators (e.g., managers, directors). All respondents interacted with PMIs on a regular basis (i.e., once a week or more).
The first author of this paper conducted all interviews. All but 4 of the interviews were done by phone. The other 4 respondents (all ED nurses from the hospital) chose to be interviewed in person, on site, in a private room at the hospital. One average, interviews took approximately 45 min to complete. All interviews were audio recorded and transcribed and followed a script which included questions regarding the workers' backgrounds and general experiences within the cross-sector collaboration. The data used in this paper comes from the portion of the interview that focused on stakeholder relationships (i.e., power, legitimacy, urgency) at Time 1 and Time 2. This section of the interview unfolded as follows. We began by providing participants with a summary of how researchers, such as Mitchell et al. [15], view stakeholder relationships. We then asked informants about perceptions of power, legitimacy, and urgency (their own and their partners) within the cross-sector collaboration. To identify perceptions of who had power within the cross-sector collaboration, informants were asked, "How is power displayed when managing PMIs". We then prompted this question by asking, "How does the police display power?" and "How does the hospital display power?" To identify perceptions of legitimacy within the cross-sector collaboration, we followed suggestions provided by Drori and Honig [60] and asked broadly about the roles key actors played in the collaboration. Specifically, we asked informants "What role do police officers/healthcare workers perform when managing PMIs?" We then examined their responses to determine which behaviors had legitimacy. To identify perceptions of urgency with respect to the management of PMIs, we asked informants, "When managing PMIs, what actions would you consider most time sensitive?" The transcribed responses to these 6 questions from these 40 respondents totaled approximately 80,000 words.

Data Analysis
Our data analysis involved the examination of 80 cases (i.e., n = 20 police Time 1; n = 20 police Time 2; n = 20 hospital staff Time 1; n = 20 hospital staff Time 2). QSR-Nvivo Version 10, a software package that facilitates qualitative data analysis, was used to code and sort the interview data (i.e., assign labels to pieces of text from the transcriptions). The data were coded using the content coding techniques and steps suggested by Reference [61]. Content coding involves the researcher developing a set of categories (i.e., codes) for words and phrases [62]. Miles and Huberman [61] describe codes as abbreviations, or tags, for assigning concise meanings to a segment of descriptive data. Assigning codes in this way facilitates the process of identifying and aggregating all data segments relevant to a particular theme, construct or research question. Codes can also be counted, organized, aggregated, and displayed to support theoretical inferences [61]. The outcomes of this coding process along with exemplar quotes are shown in Table 1. For the sake of parsimony, this paper focuses on those responses that were identified by at least 4 informants from an organization (i.e., 20% of the organizational sample). Other responses are not reported on in this manuscript but available from the authors on request. Once the data were coded, we used Nvivo's classification function to associate each informant (and their respective codes) with their organization and the time frame the interview was done.
Data analysis began once the data were coded and sorted. Data analysis proceeded in two stages. The meso-level analyses focused on perceptions at the organizational level (i.e., the unit of analysis is the police service and the hospital) [63]. The "voice" of each organization was determined by aggregating responses to each interview question [64] for each organization (i.e., police, hospital) at each data collection period. To help us understand shifts in perceptions of agency over time at the organizational level, we report our findings with respect to the following two comparisons. First, within-group comparisons over time involve comparing the responses informants in the police/hospital sample gave at Time 1 (pre-change) to the responses the informants in these samples gave at Time 2 (post change). Second, between-group comparisons at Time 1 and Time 2 involve comparing the responses the informants in the hospital sample gave at Time1 and Time 2 to the responses given by the informants in the police sample at Time1 and Time 2. Results from the meso-level analysis are shown in Table 2. Within-group and between-group differences over time of 20% or more (i.e., n = 4) are noted in the discussion below as being potentially important.
We performed trajectory analysis [65] on the panel data collected as part of this research to conduct our micro-level analysis of the data. In this case, we compared the responses each informant gave to the same question at Time 1 and Time 2, and identified informants whose responses changed over time (i.e., gave a response at T1 that they did not provide at T2, gave a response at T2 that they had not given at T1). Changes in responses over time of at least 20% (i.e., n = 4) are again noted in the discussion as being of potential importance. Results of the micro-level analysis are shown in Table 3. Table 1. Perceptions of power, legitimacy, and urgency.

Perceptions of Power (How is power displayed?)
Doctor decides PMI's treatment "Once triaged, they will be moved to an assessment unit for the doctor and from there they will be put in whatever mental health unit is best suited for their needs." (Hospital Administrator); "At times the Hospital won't even give them anything, and it's clear they need meds..." (Police Constable) Hospital staff decides when police officers can leave "The staff are in control of when the police will get to speak with the physician or with their patient and how long they will wait there." (Hospital ED Nurse); "I don't know if you'd say the administrative side or the physician side has the ultimate power . . . he's got the power to make the decision: yes, I'm going to see this person now, or no, it's not important enough, and I'm going to move on." (Police Constable) Doctor decides whether the PMI is formed "We have the power to keep them [PMIs] for 72 hours . . . we can decide whether we're going to keep you or whether we're going to let you go." (Hospital Psychiatrist); "The hospital does have the power to issue a form on this patient to keep them on site for the next 72 hours for further site assessment." (Police Constable) Police presence influences PMI process flow/wait time "They can actually escalate the patient. When they're alone with the patient, they will, and it's been known to happen, they will escalate the patient to be seen faster or to be placed faster." (Hospital Nurse)

Perceptions of Legitimacy (What legitimate role do police/health care providers perform?)
Healthcare workers assess PMIs "We have training, expertise, and a mandate to provide a comprehensive psychiatric assessment of the patient." (Hospital MHESU Psychiatrist); "They're the ones that have the resources available for us to turn that individual over for assessment." (Police Sergeant)

Police apprehend PMIs under Mental Health Act
"The Police have the authority to apprehend someone or to detain them against their will for their own wellbeing." (Hospital ED Physician); "The officers will apprehend them under the Metal Health Act, the party is handcuffed, searched as per our directives . . . to look for any object that may endanger him or us and then they'll be transported to the nearest emergency ward with a mental health facility attached to it." (Police Sergeant)

Perceptions of Urgency (Which actions are most time sensitive?)
Physicians need to assess PMIs in a timely manner "I think there is a time sensitivity out of consideration for the resources of the Police, allow them to be back on the street." (Hospital Administrator); "If the person overdosed or they've taken a whole bunch of medication, then usually once you actually get to the Hospital, they are a little better . . . that's really the time sensitive-stuff." (Police Constable) Physicians need to discharge PMIs in a timely manner "The urgency is more once the decision is made that they're not going to be admitted to get them out the door with [community] services." (Hospital Manager); "There seems to be a lot of urgency to get them out [of the Hospital] because of limited bed space."   Police presence influences PMI process flow/wait time 1 (5%) 7 (35%) 0 (0%) 0 (0%) Hospital informants more likely to perceive that police displays power in this manner post-change

Examining Police-Hospital Collaborations through a Stakeholder Salience Lens
The findings from this in-depth case study are reported as follows. Our analysis begins by looking at the data through a stakeholder theory lens (i.e., meso-level analysis). Specifically, we examine perceptions of power, legitimacy, and urgency within the cross-sector collaboration prior to the planned change (Time 1) and one year after the implementation of the planned change (Time 2), highlighting similarities and differences in perceptions between-organizations, as well as within each organization across time. The data discussed in this section of the paper are presented in Table 2. When we asked employees how power was displayed in the collaboration at Time 1, all respondents identified at least one way the hospital displayed power, but only four respondents indicated ways in which the police demonstrated power. It would appear that, prior to the introduction of the change, both police and hospital informants perceived the hospital as the more powerful partner in the partnership.
Responses suggest that perceptions of the hospital's power are primarily due to the ED physician's role in the PMI process. Two actions taken by the doctor were mentioned by a majority of respondents as examples of power. Three-quarters of police officers and nearly all healthcare workers sampled perceived that the doctor's decision-making with respect to when they will treat the PMI was an example of power exhibited by the hospital. One in three of those in the hospital sample and a one in five of the police informants also perceived that doctors displayed a second form of power-the ability to form a PMI. Forming means the PMI has their liberties suspended (i.e., must stay at the hospital), and that a psychiatrist must assess this individual within 72 h.
Half of the informants in the hospital sample and three-quarters of the police officers we talked to also felt that healthcare workers other than the ED physician also displayed power. These informants talked about how hospital staff controlled when the doctor assessed the PMI, which gave them the ability to influence when police officers were allowed to leave the hospital.
Finally, we note that, prior to the implementation of the planned changes, only three informants (all healthcare workers) commented that police officers displayed power when their actions escalated (and sometimes de-escalated) the need for the PMI to be seen quickly by the ED doctor. It was reported officers commonly acted in a manner that would encourage the PMI to "act out" in a manner that would require health care providers to see the PMI promptly.
We observed only one between-organization difference in perceptions of power in our Time 1 data-a greater proportion of the informants in the police sample (70%) than in the hospital sample (45%) perceived that hospital staff displayed power. These findings support the idea that, prior to the introduction of planned change, there was a high degree of consensus between the two partners that the hospital was the partner with power.

Pre-Change Perceptions of Legitimacy
The data support the idea that, pre-change, both partners were perceived to have a high degree of legitimacy within the cross-sector collaboration-both by themselves and their partners. Informants in both samples recognized healthcare workers needed to access PMIs brought into the hospital (Hospital: 65%; Police: 75%) and that the OMHA required the police to apprehend PMIs and bring them to the hospital for treatment (Hospital: 80%; Police: 100%). We observed one between-organization difference in perceptions of legitimacy: a higher proportion of those in the police sample than in the hospital sample spoke about the legitimate role the police played in the PMI process (i.e., apprehending PMIs under the OMHA).

Pre-Change Perceptions of Urgency
There was a high degree of consensus within our samples that, prior to the implementation of the planned change, the police demonstrated a high level of urgency within the cross-sector collaboration and the hospital did not. These findings are not a surprise and are consistent with why the deputy chief of police coordinated a change in the first place.
The vast majority of informants agreed that the police transporting PMIs to the hospital in timely manner (Police: 100%; Hospital: 65%) and that the police needing to leave the hospital and return to the road in a timely manner (Police: 100%; Hospital: 100%) were time sensitive actions. The high degree of urgency noted for the police stands in stark contrast to our data showing that neither the police officers or the health care providers in our samples felt that the ED physicians perceived that it was urgent for them to assess (Hospital: 45%; Police: 5%) or discharge (Hospital: 0%; Police: 0%) the PMI in a timely manner.
Two between-organizational differences with respect to perceptions of urgency are noted in the Time 1 data. Specifically, we note that: (1) a greater proportion of those in the hospital sample than in the police sample felt that physicians assessed PMIs in a timely manner, and (2) a greater number of police officers than healthcare workers in our samples mentioned that police officers transported PMIs to the hospital in a timely manner.

Summary of Pre-Change Perceptions of Agency
There was a high degree of consensus within both police and hospital samples that, prior to the implementation of the planned change, the hospital was the dominant stakeholder (i.e., high power, high legitimacy, low urgency) and the police service the dependent stakeholder (i.e., low power, high legitimacy, high urgency) within the cross-sector collaboration [15].

Perceptions of Power, Legitimacy, and Urgency: Time 2
As noted earlier, the police service initiated two changes to the PMI process: (1) the implementation of an "intervention and escalation" policy that prompted ED physicians to assess PMIs within two hours of arrival and (2) the requirement that police officers complete a mandatory background check of the PMI to facilitate information transfer from police officer to healthcare worker. The following sections describe informants' perceptions of power, legitimacy, and urgency one year after the implementation of these changes. Data discussed in this section are shown in Table 2.

Post-Change Perceptions of Power
Analysis of the Time 2 data shows that, one year after the changes to the PMI process had been introduced, relatively few respondents perceived that the hospital displayed power by having the ED physician decide the PMI's treatment (Hospital: 30%; Police: 15%) or by having hospital staff decide when the police can leave the hospital (Hospital: 25%; Police: 40%). There was, however, one area where the majority of police officers in our sample (65%) perceived that the hospital still displayed power in the partnership post change-the doctor still decided whether or not the PMI was to be formed. Relatively few of the respondents in the hospital sample (30%) shared this view. Finally, we note that, post-change, half the informants in the hospital sample perceived that the police displayed power by influencing the PMI's wait time by way of the new escalation police. Interestingly, none of the police officers we interviewed shared this view (i.e., perceived that their behavior was a demonstration of power).
Two between-organization differences in power are observed in the Time 2 data: (1) more informants in the police than the hospital sample perceived that that the doctor's ability to decide whether or not the PMI is formed is an example of the hospital's power in the PMI process, and (2) more informants in the hospital than the police sample perceived that the police's ability to influence wait times was an example of the police's power in the PMI process.

Post-Change Perceptions of Legitimacy
One year after the changes had been implemented, both partners in the PMI process still perceived that they and their partner had a legitimate role to play in the PMI process. Informants in both samples recognized that healthcare workers needed to access the PMI if they were brought into the hospital (Hospital: 80%; Police: 75%), and that the OMHA required the police to respond to 911 calls involving people in crisis and bring PMIs to the hospital for treatment (Hospital: 60%; Police: 100%). Only one between-organization difference in urgency was noted in the data: a higher proportion of our police informants than those in our hospital sample talked about the legitimacy of having the police attend mental health calls and bring PMIs to the hospital.

Post-Change Perceptions of Urgency
The data support the idea that, post-change, perceptions of urgency seemed to have shifted from the police to the hospital. More specifically, a majority of respondents in both the hospital (80%) and police (55%) samples stated that physicians needed to assess PMIs in a time sensitive manner. The remaining three actions that were deemed time sensitive at Time 1 were rarely mentioned during Time 2 interviews. Relatively few informants perceived physicians discharging PMIs (Hospital: 25%; Police: 0%) or police either leaving the hospital (Hospital: 10%; Police: 30%) or transporting PMIs to the hospital (Hospital: 15%; Police: 40%) as time sensitive actions.
Three between-organization differences were noted with respect to perceptions of urgency post-change: (1) a higher proportion of respondents in the hospital sample than the police sample perceived that the physicians needed to assess and discharge PMIs in a timely manner, (2) a higher proportion of those in the police sample than the hospital sample perceived that the roles they performed within the PMI process (i.e., transport PMIs to the hospital and leave the hospital in a timely manner) as urgent. These data support the idea that the partners differ in their perceptions of urgency post-change. More specifically, both partners were more likely to see urgency associated with their role in the PMI process but not their partners' role.

Summary of Perceptions of Agency Post-Change
Classification of the relationships between partners post-change using Mitchell et al.'s [15] stakeholder salience framework, provides us insights into perceptions of agency post change. Those in the hospital sample view themselves as the dependent partner in the cross-sector collaboration (i.e., low power, high legitimacy, high urgency), and the police as a dominant partner (i.e., moderate power, moderate legitimacy, low urgency). Those in the police sample, on the other hand, view themselves as a discretionary partner (i.e., low power, high legitimacy, moderate urgency) and the hospital as a definitive partner (i.e., moderate power, high legitimacy, low urgency).

Changes in Agency within a Police-Hospital Collaboration over Time
Two different analyses were used in this study to provide insights on how the planned change to the PMI process impacted perceptions of agency within our cross-sector collaboration over time. We began by undertaking meso-level analysis of the data to determine how organizational perceptions-as operationalized as the aggregate of informants in each sample's perceptions-changed over time. Meso-level analysis involved a comparison of the Time 1 results obtained using the entire police/hospital sample to the results obtained from the complete hospital/police sample at Time 2. The data used for these comparisons are provided in Table 2.
To help us interpret our data, we then used trajectory analysis to give us a micro-level perspective of the situation. This stage of the analysis focuses on the views of the individual informant rather than the perceptions of the group. For each change in perception (i.e., an informant made a comment at Time 1 but not at Time 2 or made a comment at Time 2 but not at Time 1), we reviewed our informants' responses in detail to help us understand why that person changed their view over time. The data used during trajectory analysis are shown in Table 3. In the section below, we weave together the findings from the meso-level and micro-level analyses to tell the story of how perceptions of agency changed concomitant with the introduction of changes into the system designed to make the collaboration more sustainable.

Power Meso-Level Analysis
With one exception, both police and hospital informants perceived that the hospital had less power in the cross-sector collaboration post-change (i.e., fewer respondents reported the doctor deciding the PMIs treatment and hospital staff deciding when police officers can leave the hospital as a display of power). Those in the hospital sample also perceived that, post-change, the police gained power/agency in the PMI process (i.e., police presence influences PMI process flow). These findings are consistent with the intent of the change from the police perspective and are likely due to the introduction of the "escalation policy".

Power Micro-Level Analysis
The results from the trajectory analysis are consistent with and provide insights into why the hospital's power within the PMI process seems to have declined and police power increased post-change. Examination of the responses provided by respondents who perceived that the changes had resulted in a loss of power for the hospital (e.g., Doctor decides PMI treatment, Hospital staff decides when police officers can leave) provide additional support for the idea that the intervention and escalation policy had added more pressure to the healthcare workers at the hospital (doctors and staff) to assess PMIs more quickly than in the past: "I think it's maybe taken a little power from the Hospital because now there's that expectation that, barring anything unforeseen, the apprehended party is supposed to be processed within a couple of hours. Ultimately, the power still remains with the Hospital, but I think the Police have a little more because there is that expectation." (Police Constable) Trajectory analysis also provides insight into why ten of our police informants uniquely indicated during the Time 2 interview that the hospital displayed power in the cross-sector partnership by deciding whether or not to form the PMI (i.e., gained power over time). Examination of the responses provided by these ten individuals determined that all of the officers who formed this perception post-change (i.e., mentioned at Time 2 but not at Time 1) talked about how the requirement that all police officers complete background forms on the PMI and pass them along to healthcare workers at the hospital had increased communication between the police officers involved in the PMI process and hospital staff. The officers went on to talk about how this increase in communication had led them to expect that the doctors they talked to would be more likely to take their views into consideration when the physician decided whether to form the PMI (i.e., expected an increase in agency). When this did not happen, the officers interpreted the doctors' actions (i.e., ignoring their opinion and forming the PMI anyway) as a display of power: "Sometimes you'll go to a doctor and you'll explain to the doctor what's going on.... the doctor will have a conversation with them after they've talked to you, and they're, like, no, they're not forming them, they just need to go home and take a rest or whatever. If they aren't going to listen to us, why are we even talking?" (Police Constable) Finally, trajectory analysis helps us understand why post-change there was an increase in the number of informants in the health care sample that perceived that the police displayed power by exerting influence over PMI process flow. Examination of the responses given by the seven healthcare informants whose perceptions of police power had shifted over time indicated that, in all cases, the informants felt the new intervention and escalation policy provided police officers with more power within the collaboration by giving them explicit permission to be more vocal when they were waiting with a PMI in the ED about how much time they had been there. For example: "It's empowered the police... they are a little bit more likely to be more vocal on expediting themselves through the department." (ED Resource Nurse) "I can say it changed who has power because the police are now demanding that this is the process that we have to follow... they have more power in terms of handing the person over." (ED Nurse)

Legitimacy Meso-Level Analysis
Overall, meso-level data support the idea that the changes introduced to the PMI process in an attempt to make the system more sustainable over time had little impact on perceptions of legitimacy. Healthcare workers were still perceived by the police and the health care providers to have high levels of legitimacy. Nor do those in the police sample acknowledge any change in their levels of legitimacy over time. We do, however, note that fewer of the informants in our hospital sample acknowledge that the police are a legitimate player in the PMI post-change as compared to before the changes were introduced.

Legitimacy Micro-Level Analysis
In some ways, the trajectory analysis reinforced findings from the meso-level analysis (i.e., both sets of analysis showed that informants in the health care sample perceived a decline in police legitimacy in the PMI process post-change). In other ways, however, the trajectory analysis uncovered differences that were masked when aggregating findings to the organizational level. More specifically, the trajectory analysis revealed two subgroups of healthcare workers who had very different views of how the planned change impacted the legitimacy of the hospital in the PMI process. In the Time 1 interviews, one group of 7 healthcare workers stated that the hospital had a legitimate role to play in the PMI process as they were the group that assessed PMIs. None of these individuals gave this response in the Time 2 interviews, stating instead that in their opinion the rush to assess PMIs (a by-product of the escalation process) sometimes led to inappropriate forming of PMIs.
"I think in the rush to assess the people sometimes you see people who wouldn't normally get formed get placed on a form and this could be to do with the decline in the amount of time staff time to observe their behaviors. Going from a six-hour observation, not that it was good for the Police, down to a two-hour observation, there might be a lot more question and uncertainty whether or not the patient really needs to be here and at that point might be placed on a form and directed to psychiatry just because they are showing some symptoms." (Hospital Resource Nurse) "I think that... the doctors are simply forming the patient to expedite the Police to get out." (Hospital Nurse) A second group of 6 healthcare informants identified the healthcare workers' assessment of the PMIs as a source of legitimacy in the Time 2 (i.e., post-change) interviews but not at Time 1. These six informants all spoke positively about the speed at which doctors assessed PMIs, equating the reduction in PMI wait-times to better patient care. As one triage nurse commented: "We are providing the help they need quicker." This analysis links these differing views of Hospital legitimacy to intended (i.e., better patient care) and unintended (i.e., inappropriate forming) outcomes of implementing the intervention and escalation policy. It also supports the idea the introduction of changes to the PMI process contributed to confusion within the health care sample as to their own legitimacy in the process.

Urgency Meso-Level Analysis
Our longitudinal analysis of the data determined that both partners' perceptions of what actions were urgent were impacted by the change. A greater number of informants in both the police and health care sample perceived that after the changes were introduced physicians needed to act with urgency with respect to assessing the PMIs in a timely manner. This finding is consistent with the intent of the escalation policy. While one in four of those in the health care sample also noted in the Time 2 interviews that physicians now feel urgency around the PMI discharge decision (an increase in urgency over time), none of the police officers in the sample shared this view. This suggests that unfortunately, at least from the police's perspective, the changes had little impact on the physicians' sense of urgency with respect to discharging patients in a timely manner. Finally, the data showing that post-change the number of police and health care informants who mentioned examples of police urgency (e.g., need to leave and bring PMI to hospital in a timely manner) was negligible reinforces the effectiveness of the escalation policy in terms of reducing wait time for offers and PMIs.

Urgency Micro-Level Analysis
The results to our trajectory analysis are consistent with the findings from our meso-analysis with respect to changes to perceptions of urgency over time: both partners agree that post change physicians are acting with more urgency and police officers have less urgency in the PMI process. Examination of the findings from our micro-level analysis provide insights into why, post-change, physicians felt more urgency with respect to assessing PMIs in a timely manner. Examination of the responses given by 10 of the health care informants who uniquely talked about physician urgency in the Time 2 interview made the connection between the implementation of the intervention and escalation policy and the idea that physicians needed to assess PMIs brought in by police in a time sensitive manner. For example: "It's [The intervention and escalation policy is] working because the physicians are seeing that patient quicker so instead of that patient sitting there and being sick and not knowing what's going on, they're being assessed. They're being able to be medicated sooner." (ED Nurse) These perceptions are reinforced by the views of the 10 police informants who also uniquely talked about physician urgency in the Time 2 interviews: "I think they're [doctors] recognizing that there is some time sensitivity when we have all these officers tied up. So they've done their due diligence and they've continued to, not always but for the most part, ensure that they assess the individual as soon as possible." (Police Constable) Findings from the trajectory analysis also suggest two reasons why the majority of those in both the hospital (16 informants) and police (14 informants) samples perceive that the polices' sense of urgency around the need to transport PMIs to the hospital and leave the ED in a timely manner has declined over time. Some respondents linked a decline in police urgency with respect to leaving the hospital to the success of the intervention and escalation policy. For example: "They don't have to wait in the hospital the hours that they used to wait, which was maybe between four to six hours sometimes. Now they wait two." (ED Nurse) "We can see there has been a large reduction in wait times for the police." (Hospital Manager) There is also evidence to suggest that the increased communication between healthcare workers and police officers mentioned earlier had reduced some of the uncertainty, as well as urgency relating to how long police officers would be waiting. For example: "There are times when we go up there and there are the longer wait times, but the hospital's at least communicating with us a bit more... giving us a reason. Like, they could be short-staffed or they're just buried up there... and at least knowing that, it at least gives us a bit of an explanation as to what's going on. It helps us, especially us supervisors. If we know it's going to be a couple of more hours, then we can redeploy people around and make things work on our end of... it makes things easier for us." (Staff Sergeant) Findings from our trajectory analysis hint at one other way that the planned change may have decreased police urgency. We note 12 police informants mentioned in pre-change interviews that police officers acted with a sense of urgency when they transported PMIs to the Hospital but did not make similar comments post-change. This group of officers all talked about how the fact that their job provided constant exposure to citizens with mental health problems along with feedback on the new forms they were filling out meant that, at least in their view, they were getting better at assessing PMIs when they responded to a 911 call and less worried about how these situations would escalate. The reduced worry may have led to less feelings of urgency to rush to the hospital. For example: "I think we're getting better as far as recognizing a lot of these mental health crises and I think that a lot of our officers, because we deal with it so frequently, it's allowing them... to better recognize that sort of thing." (Police Constable)

Discussion
Three conclusions are supported by this research. First, organizations with limited power (i.e., the police service) but high levels of legitimacy and urgency (i.e., dependent stakeholders) can have agency in cross-sector collaborations. Second, perceptions of how planned change impacts agency (operationalized in terms of perceived power, legitimacy, and urgency) are influenced by the organizational context of the perceiver. Third, different levels of analysis (i.e., meso versus micro) can lead to different interpretations of data associated with agency. In the section below, we explore each of these conclusions.

Legitimacy and Urgency Can Influence Sustainability Transitions
While agency and power are closely linked in the sustainability transitions literature [8], our study demonstrates that an organization with little power (i.e., the police) can still implement changes designed to make a cross-sector collaboration more sustainable as long as their partner perceives their participation in the collaboration is legitimate and their need for change is urgent (i.e., they are viewed as dependent stakeholders). While such a conclusion runs counter to much of the sustainability transitions literature, it is consistent with the stakeholder literature [15]. It also aligned with research linking influence and legitimacy [66,67], as well as urgency and readiness to change [68]. This study adds, therefore, to the growing body of literature which reports that groups with little power can influence larger systems [9,10,12,17,18] and supports the need for transitions researchers to utilize broader conceptualizations of agency when studying sustainability transitions.

Organizational Context Impacts Perceptions of Agency
While much of the sustainability transitions literature investigates how agency impacts sustainability transitions [34,69], this study examines how sustainability transitions impact agency. Specifically, we examine how two planned changes (i.e., a police change, a hospital change) designed to improve the sustainability of a police-hospital collaboration resulted in changes to perceptions of police and hospital agency (operationalized as power, legitimacy, and urgency) over time. As may be seen in Figure 1, the extent to which each of the three dimensions of agency included in our analysis are perceived to have been affected by these planned changes is largely dependent on the organizational context of the perceiver. Accordingly, we explain these findings below by drawing on research involving perceptual bias, which refers to errors that disrupt and distort the perceptual process, thus leading to faulty judgements and errors in thinking [70].

Organizational Context and Perceptions of Power
Meso-level data suggest organizational members perceive the planned changes provide a greater increase in power for their partner than themselves. To explain why informants may be more likely to perceive gains in their partner's power, we draw from the contrast effects literature. A contrast effect is the enhancement (or diminishment) of a perception as a result of successive exposure to a stimulus of lesser or greater value [71]. We suspect that contrast effects may impact informants' views related to their partner's power. For example, while healthcare workers are able to contrast the PMI's behavior with and without the presence of police officers, police officers are only able to perceive PMIs in situations where they are present. Given that police officers are not able to view PMIs in both contexts (i.e., when police officers are and are not present), their understanding of their impact on PMIs may be limited. As a result, police officers may not be entirely aware of their power within the cross-sector collaboration. Future research on agency within sustainability transitions may benefit from further consideration of contrast effects and successive exposure. Insights into how people form cognitions and how they make judgements can help researchers with the interpretation of their data and managers with the implementation of planned changes into social systems.

Organizational Context and Perceptions of Legitimacy
The meso-level analysis revealed that perceptions of legitimacy may, to some extent, be in the eyes of the beholder (i.e., the police officers in our sample were more likely than their counterparts in the health care sample to view their role in the cross-sector collaboration as legitimate). Perceptual bias may again play a role with respect to perceptions of legitimacy. We reason that individuals' perceptions related to legitimacy may be affected by selective perception and/or confirmation bias. Selective perception is the process by which individuals perceive what they want to perceive while ignoring opposing viewpoints (i.e., interpret information in a way that is congruent with our existing values and beliefs), whereas confirmation bias is the tendency to gravitate to and remember facts that confirm what is already believed [70]. Specifically, the results suggest that each organization is paying attention to information that makes them feel that their involvement in the PMI process is legitimate, and less attention to information related to their partner's involvement in the PMI process. This reasoning would explain why organizational members were more likely to discuss the importance of their own role when asked about police officer and healthcare worker involvement in the PMI process.
The legitimacy literature can also shine light on why this perceptual bias may have occurred [72]. Drori and Honig [60] differentiate between two types of legitimacy: internal legitimacy (i.e., organizational legitimacy as viewed by insiders/employees) and external legitimacy (i.e., organizational legitimacy as viewed by outsiders). They also posit that the process of building

Organizational Context and Perceptions of Power
Meso-level data suggest organizational members perceive the planned changes provide a greater increase in power for their partner than themselves. To explain why informants may be more likely to perceive gains in their partner's power, we draw from the contrast effects literature. A contrast effect is the enhancement (or diminishment) of a perception as a result of successive exposure to a stimulus of lesser or greater value [71]. We suspect that contrast effects may impact informants' views related to their partner's power. For example, while healthcare workers are able to contrast the PMI's behavior with and without the presence of police officers, police officers are only able to perceive PMIs in situations where they are present. Given that police officers are not able to view PMIs in both contexts (i.e., when police officers are and are not present), their understanding of their impact on PMIs may be limited. As a result, police officers may not be entirely aware of their power within the cross-sector collaboration. Future research on agency within sustainability transitions may benefit from further consideration of contrast effects and successive exposure. Insights into how people form cognitions and how they make judgements can help researchers with the interpretation of their data and managers with the implementation of planned changes into social systems.

Organizational Context and Perceptions of Legitimacy
The meso-level analysis revealed that perceptions of legitimacy may, to some extent, be in the eyes of the beholder (i.e., the police officers in our sample were more likely than their counterparts in the health care sample to view their role in the cross-sector collaboration as legitimate). Perceptual bias may again play a role with respect to perceptions of legitimacy. We reason that individuals' perceptions related to legitimacy may be affected by selective perception and/or confirmation bias. Selective perception is the process by which individuals perceive what they want to perceive while ignoring opposing viewpoints (i.e., interpret information in a way that is congruent with our existing values and beliefs), whereas confirmation bias is the tendency to gravitate to and remember facts that confirm what is already believed [70]. Specifically, the results suggest that each organization is paying attention to information that makes them feel that their involvement in the PMI process is legitimate, and less attention to information related to their partner's involvement in the PMI process. This reasoning would explain why organizational members were more likely to discuss the importance of their own role when asked about police officer and healthcare worker involvement in the PMI process.
The legitimacy literature can also shine light on why this perceptual bias may have occurred [72]. Drori and Honig [60] differentiate between two types of legitimacy: internal legitimacy (i.e., organizational legitimacy as viewed by insiders/employees) and external legitimacy (i.e., organizational legitimacy as viewed by outsiders). They also posit that the process of building legitimacy can have four stages: (1) developing internal legitimacy, (2) recognizing the need for external legitimacy, (3) searching for external legitimacy, and (4) initiating external legitimacy. Findings from this case study suggest that the police service has strong internal legitimacy (i.e., officers view police apprehensions as a legitimate behavior) but little external legitimacy (i.e., healthcare workers do not view police actions as legitimate). It may be that the change police implemented by the police (i.e., mandatory completion of background forms) was done with the goal of increasing the external legitimacy of the police within the PMI process over time. Unfortunately, we do not know if the introduction of these background forms ultimately contributed to a gain in external legitimacy for the police from their partners at the hospital. Such a change was not, however, observed one year after the changes had been introduced. We recommend that future researchers use longer time frames or more data collection periods when examining changes in agency during sustainability transitions.

Organizational Context and Perceptions of Urgency
Findings from the meso-level analysis with respect to urgency are somewhat opposite to what we observed when considering changes in power. More specifically, we note that each of the partners involved in the PMI process were more likely to describe the actions taken by their organization as time sensitive and less likely to perceive that what their partner did was urgent. These results would suggest that, from a perceptual bias perspective, each organization may be paying attention to information that makes them feel their role in the PMI process is time sensitive and critical, and paying less attention to information about their partner's role in the PMI process. The urgency literature may provide insight as to why this perceptual bias may be occurring. Notably, as urgency is largely based on the criticality of organizational behaviors [73], employees within an organization may be better suited to identify activities that are critical in nature to their operation than outsiders. Future research is needed to better understand urgency as an explanation of agency within sustainable transitions, given the link between urgency and readiness for change [68].

The Importance of Level of Analysis in Understanding Agency
Our study responded to calls for more research that explores sustainability transitions from multiple levels [8] in general and at the micro-level [19][20][21] in particular. Our use of both meso and micro analysis in our study demonstrated the utility of this approach. While there was very little difference in our findings with respect to changes in perceptions of power and urgency over time associated with the level of analysis, our findings relating to changes in perceptions of legitimacy over time varied depending on the type of analysis we did. Most notably, the individualized data used in the micro analysis allowed for the identification of two groups of healthcare workers that had competing views related to their own legitimacy over time. The meso analysis, which used data that were aggregated at the organizational level, did not, however, uncover these differences.
The fact that meso and micro analyses revealed different outcomes is important from a methodological perspective. This study uncovered two ways in which micro-level analyses add value: (1) they can help researchers better understand changes observed in the meso-level analysis, and (2) they can help researchers identify findings that may not be apparent at macro-and meso-levels of analysis. Moreover, the fact that our findings varied depending on the type of analysis used further highlights the need to conduct studies involving multiple levels of analysis [8], and develop clear analytical guidelines for connecting meso-and micro-levels of analysis [74].

Conclusions
This paper explored agency in cross-sector collaborations by investigating the implementation of a planned change designed to improve the sustainability of a police-hospital collaboration. Scholars have predominantly conceptualized agency in terms of power and explored the relationship between agency and sustainability transitions by examining how power impacts transitions. This research takes a different approach, however, and conceptualizes agency in terms of perceptions of power, legitimacy, and urgency and investigates the impact of a transition on perceptions of agency within the cross-sector collaboration over time. Three conclusions are drawn from our analysis: (1) organizations with limited power but high levels of legitimacy and urgency (i.e., dependent stakeholders) can have agency in cross-sector collaborations, (2) perceptions of how planned change impacts agency are influenced by organizational context, and (3) the use of different levels of analysis (i.e., meso vs. micro) can lead to different interpretations of the data.
While pre-change there was a high degree of consensus within both samples that the hospital was the dominant stakeholder and the police were dependent stakeholders, post-change perceptions of agency depended on who was asked. More specifically, employees in the hospital sample viewed themselves as the dependent stakeholder and the police as the dominant partner in the cross-sector collaboration, while those in the police sample saw themselves as the discretionary stakeholder and the hospital as a definitive partner (see Figure 1). These findings also provide support for the idea that planned change can impact perceptions of stakeholder agency within a system and that perceptual biases mean that people may draw quite different inferences to explain what has objectively changed.
The findings from this study have numerous implications for managers who are currently enacting (or planning to enact) sustainability transitions within a cross-sector collaboration. First, managers should be aware that their organization does not necessarily need organizational power to affect sustainability transitions. The findings of this paper suggest organizations with legitimacy and urgency may be able to successfully bring about cross-sector change designed to improve sustainability of the system. We recommend managers either attempt to increase their external legitimacy [60] or adopt emergent practices that are widely accepted by their partners [75]. Second, managers should also be cognizant of the notion that, even though they believe their organization has the power (or legitimacy and urgency) needed to successfully transition their system to a sustainable end-state, their cross-sector partners may have differing views. Thus, practitioners seeking to introduce changes into a system need to focus not just on how they view the situation but consider how their partner might construe what is going on, as well. It is often said that "perception is the lens through which we view reality." Practitioners (and researchers who seek to help them) who desire successful sustainable transitions need to be aware that there is a difference in how they perceive reality, how their partners perceive reality, and what reality truly is. The findings from this study support this conclusion and provide a valuable caution for both researchers and practitioners moving forward. This paper is not without its limitations. First, one of the strengths of this study (longitudinal data from the same individuals collected pre-change and one-year post-change) is also one of its limitations. It is very likely that changes in agency continued after our study ended. Future research in the area should allow an increase in the number of data collection periods and the time frame of the study to three or more years to give researchers and practitioners a more complete picture of how the planned changed unfolded over time.
Second, participants in this study are from two public sector organizations. Future studies should focus on undertaking this type of study in private organizations where collaborations may be quite different. Third, responses from organizations in multiple sectors (i.e., policing, healthcare) were used to assemble the dataset. While a strength of this study, the fact that we included organizations from multiple sectors in the same study might also introduce a number of confounds into the analysis. Future researchers could remove this limitation by focusing on organizations from a single sector.
Finally, our conceptualization of the interorganizational relationships is limited to Mitchell et al.'s [15] stakeholder framework. While the decision to use this framework is consistent with many other researchers [49,76], it is a rather narrow definition of relationships. Future research in this area could benefit from a more comprehensive outlook of interorganizational relationships.
Author Contributions: Both authors contributed significantly to this research. M.H. collected and analyzed the data. He also wrote the first draft of this manuscript. L.D. revised the manuscript and rewrote many portions of the manuscript. All authors have read and agreed to the published version of the manuscript.

Conflicts of Interest:
There were no conflicts of interest.