*1.1. Collaboration*

Collaboration is here defined as a horizontal process where stakeholders, both at the organizational and individual level, share competences, unite resources, and unprestigiously work together towards a common goal. [10]. Compared with coordination and cooperation, collaboration calls for more frequent interaction, higher embeddedness, and a larger will of risk sharing [11]. There are multiple motives for engaging in collaborative processes, but possible explanations include an overall assessment of advantages vs. disadvantages [3], a desire for individual or social benefits [12], and the sharing of risks [13]. As a working form, it has over the years become popular across multiple fields and branches and received considerable attention, especially in managemen<sup>t</sup> literature [14–17]. Well-functioning collaboration processes are often presented as a solution to task allocation and regulatory fragmentation [18], thus they represent the leading perspectives within fields such as team-development, coaching, and integration [19]. In the field of crisis management, collaboration is viewed as a key success criterion [20–22] and has been found to positively a ffect the overall outcome of a crisis [23]. We define a crisis as a situation or incident that outsources available resources [24] and is not confined within administrative, geographical, or physical boundaries [25]. Collaboration, or more specifically cross-sector collaboration, which is the main focus of this study, is viewed in the crisis literature as a core concern as it helps both crisis managers and societies to e ffectively deal with adverse consequences [23] and meet societal expectations [26]. On the contrary, if managers fail in their quest, a lack of collaboration may lead to less resilience, flexibility, and e fficiency [27]. While there are examples of sudden, informal collaboration processes during major disasters such as Hurricane Katrina [28] and the 2011 terrorist attacks in Norway [29], collaboration is something that, in most cases, needs to be learned, developed, and exercised. Thus, there exists an assumption that cross-sector collaboration exercises develop, train, and test joint preparedness e fforts and response [8]. The problem, however, is that recent studies have found that such exercises tend to have limited perceived levels of learning and utility [30–35]. Sources today are conflicting as to why this perceived limitation occurs, but cited reasons include a lack of focus on variation [36], lack of trust [37], and insu fficient focus on collaboration learning and utility enhancing elements [3,6,7,38]. Collaborative strategies can, however, di ffer depending on the current situation. On a scale of less to more collaboration sequential, parallel, and synchronous types of collaboration have been identified. Sequential strategies are often used when it is optimal to go through o fficial channels and stick to routines. Parallel routines are when tasks are carried out simultaneously, while acting "on their own". Such a subtle type of collaboration is used when members do not go in and help each other across. It is characterized by the standardization of developed roles and established procedures. Synchronous collaboration means stepping over the boundary into the unfamiliar and flexibly covering for others where needed, even if this does not lie within a specific area of competence. Synchronous collaboration is the idealized seamless form of collaboration referred to when governing bodies stress their ability to interact, but it is also a challenging and exhausting type of collaboration. These types of collaboration have been identified as useful depending on the current situation. The synchronous type of collaboration is seldom used in everyday practice, but merely when there is a lack of resources such as during mass casualty scenarios [3].

#### *1.2. Learning and Utility*

The goal of learning is to acquire new knowledge [39]. The idea of collaboration learning during exercises is, in this study rooted in, and limited to, Johan Stein's [40] and Klabbers [41] perspectives on how institutions learn, hence the differences between first- and second-order learning. First-order learning is when participants learn new things during the exercise but are unable or unwilling to transfer knowledge to practice. Second-order learning, on the other hand, is when participants manage to acquire new knowledge and apply that knowledge in real situations [31]. While the goal of obtaining new collaboration skills and understanding during exercises may seem obvious, Berlin and Carlström [3] found that while it was considered theoretically and socially correct to support collaboration engaging processes, participants, in practice, prefer their everyday, standardized working patterns, which results in a decoupling between theoretical structures and practice [42]. To increase exercise utility, planners should focus more on integrating collaboration-learning elements, hence creating new configurations of thoughts that bridge exercise learning and real life practice [43]. Bourgeois et al. [44] discovered that the promotion of collaboration enhancing factors increased the sense of team-belonging and the ability to learn, thus substantiating Borell and Eriksson's [36] later argumen<sup>t</sup> on how perceived crisis learning is greatly dependent on the design and applied exercise model. When designing an exercise, there are multiple ways to enhance both collaboration learning and exercise utility. Firstly, the exercise needs to have a purpose and primary objective [45], hence the exercise needs a clearly defined training content with associated learning outcomes [46]. Participants must be informed that the main goal is collaboration development, not only complex scenario solving. Secondly, the organizers need to provide clear collaborative instructions and ensure that the exercise participation feels relevant and is free from long or unnecessary waiting periods [31]. Thirdly, it is important that all participants at all times have an overview of the ongoing scenario development [45] and feel that their opinions matter. Moreover, they must be included in formative assessment and collective reflection processes throughout all stages of the exercise [39]. On that note, organizers and managers also need to take into consideration that collaboration learning processes are not always either black or white. Some professional boundaries are, and will always be non-negotiable due to e.g., jurisdiction or complexity of task [47].

#### **2. Materials and Methods**

This study reports on data collected from a 2018, two-day, full-scale, wildland-fire collaboration exercise in southeastern Norway. The main goal of the exercise was collaboration development, but included also technical, logistical and managerial elements. Participants included the Norwegian Civil Defense, County Wildland-Fire troops, local fire and rescue personnel, the Norwegian Directorate for Civil Protection, and local fire planes and helicopters. The exercise scenario ran from alarm to completion and included air- and land-based fire extinguishing, controlling of fires, and establishment of fire ditches. Logistical exercise elements included crew reception, parking, and housing/catering establishment. The exercise planning and directing staff was composed of senior representatives from the participating organizations. There were no performed joint evaluations following the exercise. The exercise was conducted in May, and data was collected in early fall. The exercise had 184 participants (*N* = 184) representing local full-time and part-time fire and rescue services, regional wildland-fire troops, civil defense, and wildland-fire planes and helicopter personnel.

The sample included both the operational and tactical level personnel. The selection of sample participants was based on an assumption that relevant personnel in need of collaboration exercise participated, and that the exercise had a relevant and clearly defined collaborative purpose and primary objective.

#### *2.1. Data Collection and Procedures*

A G\*Power 3.13 analysis [48] calculated the appropriate sample size to be 82 participants. A *t*-test, linear bivariate regression—one group, two-tailed, with an alpha significance level of 0.05 [49], a statistical power of 0.80, and an effect size of 0.3—was applied. The collection of data occurred through the use of a validated online survey instrument. The collaboration, learning, and utility scale (CLU-scale) [31] became the instrument of choice as it is especially designed to measure the perceived effects of collaboration exercises, with an emphasis on learning and usefulness (Table 1). Also, as the CLU has been applied in similar studies [7,31,32], a comparison with other collaboration exercises are made possible. CLU is a Swedish developed survey tool. An expert group of five academic instrument-developing experts together with three emergency practitioners, representing blue-light response organizations, developed it. The instrument was developed in different stages based on Meyer and Rowans 1977 decoupling theory [42], Berlin and Carlströms theories on sequential, parallel, and synchronous collaboration [50], and Steins [40] learning theories, which have their outspring from Klabbers [41] perspectives on how institutions learn, hence the differences between first- and second-order learning. Before completion, the CLU-scale was tested in multiple pilot-studies. The final product consisted of 17 items measuring the three dimensions collaboration (C), learning (L), and utility (U). The C dimension measures the perceived collaboration characteristics, the L dimension emphasizes collaboration related lessons, and the U concerns transfer of value to real-life scenarios. The CLU-scale is based on a 5-point Likert scale with the values 1 (strongly disagree), 2 (mildly disagree), 3 (neutral), 4 (mildly agree), and 5 (strongly agree). The instrument's homogeneity was tested through a calculation of Cronbach's alpha, showing an alpha of 0.88 [51]. Analysis stems from descriptive data and bivariate and multiple regressions. Means and standard deviations were included for descriptive purposes. The instrument has earlier been applied in multiple, similar studies [7,30–33,52]. To ensure appliance with ethical research standards, the researchers sought permission from the Norwegian Center for Research Data (NSD) prior to data collection.


**Table 1.** The collaboration, learning, utility scale (CLU-scale).

Dimensions: Collaboration (C), Learning (L), Usefulness (U). Source: Berlin and Carlström [31].

Prior to contacting the defined sample population, permission was sought from the participating organizations. The email contained information about the study, methods, instrument, and purpose. The letter emphasized volunteerism and assurance of anonymity. After obtaining written permissions and email addresses from the organizations, an invitation to participate was sent out to the population sample. Apart from a hyperlink to the survey designed in the online survey platform "Nettskjema" [online-form] hosted by the University in Oslo, the invitation contained information about the study,

data handling, and how to contact the researchers. Volunteerism and anonymity were highlighted, as well as the option to at any time withdraw from the study without facing consequences. To ensure further anonymity, only demographical questions related to gender, age, professional experience, and professional a ffiliation were asked. Age and experience were divided into groupings, and a ffiliation was limited to public or NGO sectors. Age groupings were 1 = 18–26, 2 = 27–35, 3 = 36–44, 4 = 45–53, 5 = 54–62, and 6 = 62+. Choice of a ffiliation was 1 = private, 2 = public, and 3 = volunteer. Years of professional experience was divided into 1 = 0–5, 2 = 6–11, 3 = 12–15, 4 = 16–20, and 5 = 21+. Number of collaboration exercises attended over the last five years were divided into 1 = 1–3, 2 = 4–7, 3 = 8–11, and 4 = 11+. The participants were asked to complete the survey within three weeks. During this period, two reminders were sent out. After collection, all data was uploaded into Statistical Package for the Social Sciences (SPSS) and analyzed. Identifiable information in the data set was removed and replaced with a number. The scrambling key was stored on a safe drive at the University of South-Eastern Norway accessible only to the research team. The key and identifiable information were deleted after the completion of this research project.
