Authority gradients between rail network controllers, train crew and track workers in Australia and New Zealand: motivations and behaviours

The underlying relationship dynamics leading to authority gradients between frontline rail workers are being overlooked despite rail incident investigation reports pointing to the contrary. The aim of this study was to understand how the power disparities across multidisciplinary rail frontline workers can lead to authority gradients from the perspective of network controllers—the group at the apex of the operational hierarchy. A scenario generation technique was used to scaffold interviews with 55 rail network controllers from eight organisations across Australasia and identify how they perceived risky situations with all data thematically analysed. Authority gradient generation was found to be defined through: (1) motivations for network controllers to retain a position of authoritarianism toward train crew and track workers; and (2) behaviours that reinforced a power differential by curtailing their empowerment. Network controllers feared the probability and consequence of error and mistrusted in the capabilities of train crew and track workers, questioning their honesty, levels of competence, and believing they wanted to inherently cut corners. These motivations created a contemptuous regard that bred hostility and disparaging language, and engendered intimidation tactics where network controllers acted punitively and pressured train crew and track workers into compliance. The results from the scenarios point to perceptions of risk as the catalyst for fear and mistrust, with heightened perceptions of risk associated with increased vigilance and hostility across groups. This insight into how authority gradients are generated across operational teams in rail provides a new dimension to understanding teamwork error.


Introduction
In hierarchically structured organisations, some workers have less power than others (Tuckey et al. 2009). This can be attributed to their role within the structure, or interpersonal relational factors between individuals (Magee & Galinsky 2008). A disparity in allocation of power is known as an 'authority gradient' (Sasou & Reason 1999). In the context of rail organisations, authority gradients between rail drivers and co-located workers (i.e. co-drivers, guards) are known to contribute to incidents on the network (Australian Transport Safety Bureau 2014; Office of Transport Safety Investigations 2012). For example, the 2003 Waterfall rail accident in New South Wales, Australia, was partly attributed to an action failure from the train guard for not intervening when the speed of the train was excessive enough to alarm passengers (McInerney 2005). One conclusion drawn was that rail authorities 'failed to ensure that authority gradients did not exist between train drivers and guards, so that train guards were not reluctant to take action to stop a train in an emergency situation' (McInerney 2005, p. xxxvi).
In the safety critical environment of rail (Krehl & Balfe 2014), drivers and guards share responsibility for safely moving the train; however network controllers and track maintenance workers also form part of the multidisciplinary team. Collectively, these groups ensure network operability (Dorrian et al. 2011). Network controllers, also called train 1 3 controllers, signallers, planners or dispatchers (Wilson & Norris 2005), ostensibly sit at the top of a power hierarchy within the operational sphere of railway activities (Naweed 2020). By ensuring the safe separation of rail traffic and working groups, managing incidents and accidents, and applying standard procedures and organisational rules for safety and efficiency (Siegel & Schraagen 2017), network controllers play a pivotal role in providing oversight of activities on the network (Cheng & Tsai 2011). They provide direction to other workers, which includes provision of authority to train drivers for rail vehicle movements, and track maintenance workers for access and length of time permitted on track (Baldry & Ellison 2006). These dynamics, while critical for safe working, also have potential to create an authority gradient between the network controller and other workers (Luva & Naweed 2021).
Controlling the network safely relies on the integrity of information exchanged between teams (Naweed 2020), and can be easily impeded if the teams are dispersed geographically (Waring et al. 2020). While rail investigation reports reference authority gradients in the aetiology of incidents, they do not provide a framework for understanding how they are generated, or to what extent power disparities disrupt communication. The Australian Transport Safety Bureau defines an authority gradient as a 'term used to describe the influence in crew communications where there is a significant difference in experience, expertise or authority' (2014, p. 15). As the adverse impact of authority gradients on communication is recognised by industry, the paucity of rail research investigating the phenomena is surprising.
The need for closer examination of authority gradients and their impacts is implicated by compelling accounts in rail incidents (e.g. Australian Transport Safety Bureau 2015, 2016; Office of Transport Safety Investigations 2012). In a recent review of existing literature and examination of tools, frameworks, methods and theories, Luva and Naweed (2021) identified a critical research gap around authority gradients within rail, with a need to better understand authority gradient generation in this industry.

Towards a better understanding of authority gradient generation in rail
One way the rail industry tries to deal with authority gradients is to understand 'team-work skills' (Cooper et al. 2010, p. 14). This equates to an application of 'non-technical skills,' defined by Flin and O'Connor as 'the cognitive, social and personal resource skills that complement technical skills, and contribute to safe and efficient task performance' (2008, p. 1). In a prevalently used rail-centric non-technical skills framework (Rail Safety and Standards Board 2016), various categorisations of non-technical skill are made that evoke authority gradients. These include 'assertiveness,' placed under a 'communication' category, 'confidence and initiative,' placed under a 'self-management' category, and the skills of 'treating others with respect' and 'dealing with conflict or aggressive behaviour,' which are both placed under a 'co-operation and working with others' category. Thus, the current focus in rail is placed on training individuals in behaviours that improve team operations, rather than understanding why team operations are affected by behaviours synonymous with authority gradients. Beyond the focus on non-technical skills, studies on team communication breakdown have a tendency to focus on procedures, protocols, and phraseology as root cause of communication error (Turner et al. 2017). Such a focus does not consider the influence of the operational environment (i.e. the roles and responsibilities of the team members within an organisation), nor the influence of relationships between team members/operational groups. For example, when examining communication error between trackside personnel and signallers, Gibson et al. (2006, p.64) suggest 'a key question is the underlying mechanisms which lead to these types of error. This requires an understanding of the context in which personnel are trained, the culture in which they work, and also basic human processes.' From this, one may surmise that authority gradients are an underlying mechanism of breakdowns within teams, meaning that communication error may be better understood as a symptom rather than a cause. In error taxonomies, underlying causes of communication tend to be considered from a cognitive analytical perspective through error analysis and classifications tools (Gibson et al. 2006). Tools such as Technique for the Retrospective and Predictive Analysis of Cognitive Errors (TRACEr) (Shorrock 2006) or Incident Factor Classification System (IFCS) (Dixon et al. 2017) tend to broadly categorise such breakdowns as 'communication error' and applications of non-technical skills frameworks treat them as 'non-technical skills failure.' Approaches to human reliability analysis that categorise performance shaping factors do not adequately address 'attitudinal factors' (Houghton et al. 2016, p. 375). The context of communication between individuals is not recognised, that is, the relationship between the individuals communicating. Figure 1 illustrates the nature of the communication relationship between operational roles that are the focus of this study.
In aviation, power disparities impacting on communication within and across teams have been understood for some time and originated from the study of power dynamics or a 'trans-cockpit authority gradient' between junior and senior pilots in the aircraft cockpit (Alkov et al. 1992). The aviation industry has sought to address this over the years through crew resource management (CRM) training with the aim of giving individuals the tools to improve their team interactions, i.e. non-technical skills targeting the symptoms of team dysfunction. These principles have been adopted 1 3 across similar hierarchically organised, safety-critical systems including fire and emergency services (Griffith et al. 2015), the offshore oil industry (Flin 1995), nuclear power plants (Juhász & Soós 2011), surgical teams (Flin et al. 2010) and health care more broadly (Gordon et al. 2012). While attempting to address one of the symptoms of an authority gradient, such as the perceived inability to challenge someone in a more powerful position, training team members in non-technical skills such as 'assertiveness' does not address the underlying issues that lead to authority gradient generation.
While the 'trans-cockpit authority gradient' is an example of the authority gradient between individuals in a direct supervisory relationship, authority gradients also manifest between teams/professional groups in hierarchically structured systems, without supervisory relationships. For example, in a medical context of cancer care, Oborn and Dawson found that while the contributions of nurses were 'de-valued' (2010, p. 849) by surgeon teams, pathologists had lower standing than any other team, including nurses, oncologists, radiologists, and surgeons. The inequality in these perceived power distributions engendered institutional silencing where pathologists and nurses did not challenge surgeons. Thus, the context of error in communication across the disciplines is the power disparity that inhibits knowledge sharing and equal participation in multidisciplinary settings.

What does an authority gradient look like?
An authority gradient exists as a hierarchy of power, separate from leadership or authority that impedes communication by affecting how 'psychologically safe' team members feel about speaking up or challenging other team members (Eppich 2015). An authority gradient is 'the difference in control of decision making' (Acquaviva et al. 2013) and can be tied to professional identity (Friedman et al. 2015), role status (Gilardi et al. 2014), occur across professions and disciplines (Eppich 2015) and between junior and senior members of the same profession (Gluyas 2015). The inherent risk of an authority gradient is that safety-compromising behaviours will be overlooked, ignored, or disregarded. Walrath et al. 2015, p. 176) evaluated experiences of authority gradients with nursing students and asked them to describe scenarios where an authority gradient had been observed to obstruct team operations. Behaviours reported included, 'disregard for the professional input of other team members, verbal aggression, condescending language, and power plays by the individual perceived by the student to be of higher status. ' Okuyama et al. (2014, p. 6) pointed to 'fear of reprisal' and 'concerns of appearing incompetent' for medical staff failing to 'speak up' and Gluyas (2015, p. 53) found staff were unlikely to speak up if they had experienced 'previous rudeness or intimidation from other staff' or were fearful of the response of others.
Rail operations rely on the collaboration of multidisciplinary teams (Roth et al. 2020). In the same way that power dispersals impact on team dynamics and communication in a health care setting, power may become imbalanced across operational groups and between individuals in rail. To understand how the authority gradient is generated, it is necessary to identify what factors motivate network controllers, positioned at the top of the power hierarchy (Cheng 2019), to work in ways that reinforce an unequal distribution in power. Better understanding of how power becomes imbalanced in Fig. 1 Diagram showing the direction and methods of communication between the network controller, train crew, and track maintenance workers. Radiobased communication is bidirectional for all parties, whereas rail signal-based communication for authorising movement is unidirectional. Train crew and track maintenance workers generally do not communicate directly 1 3 rail, may in turn inform better ways of tackling authority gradients in this setting.

Aims and research questions
The aim of this exploratory study was to understand how the power hierarchy across multidisciplinary rail operational teams, comprising network controllers, train crew and track maintenance workers, results in the generation of authority gradients between workers. This was undertaken with the following research questions: RQ 1: What factors motivate network controllers to work in ways that can lead to the generation of an authority gradient? RQ 2: How is an authority gradient generated between network controllers and workers within other operational groups?

Study design
This study adopted a qualitative action research design to collect insights directly from those managing safe separation of trains on the rail network. In doing so, it collected their perspectives, opinions and experiences to gain a deeper understanding of the attitudes as well as the behaviours informing how they work and interact with others (Curry et al. 2009). Given a lack of prior research on this topic, inductive thematic analysis (e.g. Penney et al. 2022) was employed to identify core attitudes and assumptions, and formed the basis of the key themes (Braun & Clarke 2006). As the research sought to reveal meaning in language and action, the approach drew on principles from constructivist grounded theory for understanding data (Charmaz 2017a).
Data were collected in a semi-structured one-to-one format. Interviewing individual participants overcame rostering constraints and recognised the operational needs of the control room environment. In order to obtain a 'thick description' of attitudes and beliefs of work and interactions (Ponterotto 2006), including the motivations and behaviours linked to authority gradients, the Scenario Invention Task Technique (SITT) was used as a mechanism to scaffold data collection (Naweed 2015). The SITT is a retrospective scenario-based technique that uses storytelling in conjunction with the Rich Picture Data method (Monk & Howard 1998) to externalise knowledge. It involves generating challenging scenarios using drawings and is particularly useful for eliciting knowledge from content experts (Roth et al. 2020). The SITT has been used to elicit knowledge from various industries, including health care , aviation (Kingshott & Naweed 2018), and rail (Naweed 2013).

Participants and recruitment
A total of 55 network controllers (four female, 51 male) ranging from 22 to 67 years of age (M age = 45.5, SD = 11.1) were interviewed. Participants were recruited from a total of eight rail organisations across Australia and New Zealand, using a maximum variation sampling approach (Palinkas et al. 2015). Most had > 10 years of experience in a controller role (M exp = 15.83, SD = 12.18).

Procedure
After obtaining consent, interviews ran from 90 to 120 minutes. Table 1 provides an overview of the interview protocol. The protocol was designed using various question types, from those that were closed, to those more open ended and cued to encourage explanation and presentation of new ideas. The first part set the scene for the SITT which probed how network controllers perceived and managed risky situations. The SITT commenced with an instruction to: 'Invent a scenario where even the most experienced train controller inadvertently decreases rail safety.' Signal passed at danger (SPAD) incidents where a train passes a STOP signal were given as examples to contextualise their scenarios; however, to decrease impositions on creativity, any scenarios were encouraged and scenarios that did not feature SPADs were also created. Participants were asked to walk the researcher through the scenario after constructing it. Scenarios were then probed to generate insights around the research questions (see Table 1). All scenarios were created using felt markers and A3-sized paper. Figure 2 provides an example scenario created during the study.

Research team
The research team comprised two investigators, one female and one male. One was a senior university researcher with a background in psychology specialising in rail human factors. The second member was a PhD candidate with 5 years' prior experience as a network controller. Both researchers discussed their own assumptions and possible biases on the research topic at length (Creswell 2013).

Data analysis
All interviews were transcribed, and scenario illustrations were de-identified. This also involved scanning and replacing handwriting with digital typescript and substituting real names (e.g. of stations) and other potential identifiers (e.g. train head codes) with generic tags. Data analysis was conducted at a 'latent level,' meaning analysis sought to understand and interpret the underlying attitudes and assumptions (Braun & Clarke 2006) held by participants. In contrast to coding explicit ideas or concepts to derive semantic themes (i.e. directly referencing what has been reported as an idea), themes developed at the latent level encapsulate more implicit ideas or concepts that underpin what is explicitly expressed (Braun et al. 2016). In practice, this meant revealing common views about what things network controllers said about other workers, how they spoke about other workers, and identifying the salient power/status dimensions between them. The analytical framework corresponded with the constant comparative method, a tool typically used for grounded theory, involving opening coding for initial key concepts, refining codes continuously through axial and selective coding (Blair 2015;Curry et al. 2009). The six-step process used to analyse the data and extract themes is given in Table 2.

Results
Analysis revealed two overarching themes: (1) motivations for maintaining a power differential, reflecting factors that were found to motivate network controllers to retain an authoritarian position in relation to other workers; and (2) behaviours that reinforce a power differential, reflecting factors that curtailed the empowerment of other workers. As shown in Fig. 3, each of these themes were associated with two subthemes. Motivations for maintaining a power differential were fear associated with the risk of error, and mistrust in the capabilities of other workers. Behaviours that reinforced a power differential were intimidation of other workers, and contemptuous regard for other workers. Figure 2 also shows the minor categories associated with each subtheme. The results for both themes are presented next, along with excerpts and  Table 2 Process used to analysis study data and identify themes (Braun and Clarke 2006) No Step Activities 1 Understanding the data Active reading of the interview transcripts collected across eight organisations from Australia and New Zealand, from 55 network controllers. Interview transcripts read and re-read, patterns within responses were identified and noted 2 Devise initial codes Identification of commonality in attitudes and behaviours, organised into groupings that identify a feature of the data. Codes developed based on indicators of attitudes that illustrate and/or evidence a power disparity within the operational team 3 Identify themes Grouping of codes into overarching themes that recognise features of a status/power differential 4 Review of coding Data set reviewed and codes evaluated as to relevance of themes and subthemes 5 Refine themes Use the themes to structure the data in such a way that a narrative can be built around the 'story' that the data tells 6 Write-up Finalise analysis and write-up of the study findings Fig. 3 Overview of the themes (inner circle) and subthemes (outer circle), reflecting the factors found to motivate and reinforce behaviours associated with a power differential in network control each minor category are also included with each minor category subheading.

Fear associated with the risk of error
Fear was revealed as one of two primary motivators, driving behaviours and attitudes that may lead to the generation of an authority gradient between network controllers and other workers. The level of fear was commensurate with the level of perceived risk in working with different operational groupings (track workers or train crew). As fear was associated with risk, the main categories here were probability (that an incident would occur) and consequence (should an incident occur). The authority gradient became steeper when risk was perceived to be higher, due to the coping mechanisms employed by individuals in effort to minimise the risk.

Probability: "If we're going to make a serious error and I'm talking a safety error, it's probably more so after the peak because what happens? All the track workers come out."
Authorisation from network control was required for track workers to access and occupy sections of track. Network control was responsible for excluding rail traffic from a portion of the rail network and a period of time agreed with the protection officer. To achieve this, the network controller and protection officer communicated via radio or mobile telephone. Participants viewed the time taken to issue a track warrant and/or instate the electronic systems for authorising track possessions as a risk which placed additional pressure upon them, for example: If you go and look on the graph yesterday, we had eleven sets of track work, eleven sets! What that's doing is taking an awful lot of time to sort out, an awful lot of pressure on the train controller but more importantly, it's not safe! (Ppt46_Org6) Participants felt apprehensive when track workers requested unscheduled work, and experienced additional pressure to facilitate time on track, partly because it required immediate adjustments to their plan, leaving some feeling "worried" (Ppt9_Org1). The potential for diverting attention from other tasks, while generating and sustaining the necessary written or electronic authorisation, was another aspect of track work that was viewed as increasing the probability for error: In managing the competing priorities, participants indicated that they turned the volume on their desk telephone down, ignored non-essential radio calls, or took their telephone 'off the hook'. The expectation was that workers attempting to contact the controller would "be patient" (Ppt41_Org6) but that this was "not always the case" (Ppt41_Org6). Implied frictions between workers in one of the organisations led to the introduction of a 'Take 5' system for helping network controllers cope with demanding periods, for "controllers to get away from their desk and come back feeling refreshed" (Ppt41_Org6). The implication was that the heightened vigilance associated with peak periods influenced resultant behaviour.
In comparison to track workers, who worked between rail traffic movements and with high levels of interaction with the network controller, train crew typically received authority for movement of scheduled rail traffic via signalling systems. To prevent a SPAD, participants needed to remain "vigilant" (Ppt2_Org1; Ppt29_Org5) in their interactions with train crew. In some rail systems, older forms of train control such as verbal track warrants or Train Orders were used in unsignalised territory, but in track circuited control areas, train movements were authorised by signal aspects. Participants perceived that the probability of error increased if train crew were complacent in their interactions with the signalling system:

Most of the SPADs that I've known, or my experience has been is the driver's looked up thinking he's had a proceed aspect in whatever colour not really seen what he saw and then as he's-oh, shit [sic]. (Ppt8_Org1)
Train Orders utilised radio-based transmission of standardised phrases between network control and train crew with both maintaining a permanent written record. These orders were viewed as "dangerous" (Ppt6_Org1; Ppt7_Org1) with one organisation referring to them euphemistically as "suicide notes," (Ppt5_Org1) describing them as "the most likely thing to get a train controller in trouble and have got some very technical and obscure rules based around them" (Ppt5_ Org1). Facilitating these methods of rail movement was viewed as high risk and therefore associated with heightened levels of vigilance. Figure 4 illustrates a collision scenario associated with multiple Train Orders being issued in the context of single track occupancy and demonstrates the perceived danger inherent in the issuing of verbal authorities.
Older train control systems were perceived to present an element of risk as there was "no secondary system, no-one checks that I've done the right thing" (Ppt5_Org1). Communication between the network controller and train crew typically happened via the signalling system which passed on a message in the field to the driver. Rail movements relying on verbal authorisation (with an accompanying written record) were utilised when a signal could not display a proceed aspect. Sometimes called a 'Special Proceed Authority' or an 'Alternative Proceed Authority', this was the method through which rail traffic was authorised to pass a signal displaying a stop aspect and held potential for high risk such as a collision (if there was opposing rail traffic), or derailment (if points were set incorrectly or there was damage to the rail). Increased exchanges of communication associated with provision of verbal authorities to train crew were viewed to increase the probability of error occurrence, which had the effect of enhancing the fear associated with the risk.

Consequence: "If we make a mistake the consequences can be real, very real for people."
Participants linked the consequence of a potential error with serious injury or death, "you've got people on track and you don't want them getting run over by a train" (Ppt45_Org6), but also with personal consequence to themselves:

I had to go in for drug and alcohol testing for a statement […] I was stood down from duties for two weeks while I was investigated. (Ppt52_Org7)
Facing disciplinary action as a result of error attributed to the controller was a significant concern. For example, in terms of the consequences of error, the Train Order system was described as "the most likely thing to get a train controller stood down" (Ppt5_Org1). Consequences for error when engaging with track workers were recounted, for example, when a track worker communicated a wrong station, "we nearly ran him over with a coal train" (Ppt20_Org4). This also included fatalities: A poor guy lost his life in the last 2 or 3 months, struck by a train on a track, one of our track workers. (Ppt25_ Org4) Participants perceived a higher level of risk with track workers, "they are in a more dangerous situation with more problems should you do something wrong," (Ppt42_Org6) with significant consequences associated with dangerous situations: We've given a ganger 2 some track time and we've accidentally forgotten to put blocking on and then we give a train a light to go right through the section.
The ease with which such situations could occur was a recurring point: The ganger didn't ask for the blocking and the controller just gave him foul time 3 and the next thing, of course, the train comes along here and that guy's suddenly completely unprotected. (Ppt 37_Org6) Although the consequence for error with track workers was acknowledged, participants also perceived personal consequences by adopting an overly cautious approach brought on by organisational pressures and mixed messaging: Station A is issued a Train Order to proceed to Station C (denote by green arrow from the left towards the right), while a second train at Station C is issued a Train Order to proceed simultaneously to Station B (green arrow from the right towards the left). Note: handwriting has been replaced with typescript. (Ppt7_Org1) is a major project, and we need to get this done and why aren't you letting these guys on? (Ppt 50_Org7) Organisation-imposed consequences were also expressed in relation to decisions or actions that negatively impacted train scheduling, "I've held a train because I was distracted […] so, then you've got to give a 'please explain' as to why you've done that" (Ppt9_Org1). Failure to communicate conditions affecting the network was also perceived to result in disciplinary action: During the middle of the peak, a ganger might decide to put a speed restriction on a certain area so you're having to tell the trains of that restriction and if we miss a train then we get stood down and drug tested if the train doesn't get the restriction. (Ppt45_Org6) The consequence of a potential incident or accident impacted on the level of anxiety participants felt in their interactions. Being stood down, drug tested, and addressing a "please explain" (Ppt9_Org1) request from those higher up were all identified as consequences that carried personal risk and induced fear.
To minimise the probability of a risky situation occurring, participants were focussed on a "state of preparedness" (Ppt13_Org3) needed to prevent becoming "flustered" (Ppt15_Org3), pointing to the potential for unfavourable interactions between staff in "stressful" (Ppt20_Org4) periods. These were likely catalysts for "very assertive" (Ppt16_Org2) exchanges, and efforts towards "getting that balance right between being rude and curt, and just to the point" (Ppt15_Org3). Participants advocated a need to "explain yourself clearly" (Ppt13_Org3) and acknowledged "tone will definitely change" (Ppt31_Org5) when giving an instruction. Terse exchanges may be interpreted as verbal sanctions, steepening the authority gradient between actors. Participants acknowledged the importance of not coming "across as accusatory in my tone or my manner, if it's not the driver's fault" (Ppt50_Org7). This raises the question of how behaviour may manifest or be interpreted if the driver were perceived to be at fault.

Mistrust in the capabilities of other workers
A lack of trust in the competence of other workers and their abilities to adhere to critical procedures and regulations influenced how participants perceived risk and worked with different operational groups. Trust between network controllers and other workers was viewed to be an important aspect of teamwork. Mistrust in other workers stemmed from misgivings around their level of competence, perceptions of dishonesty, and a belief that they wanted to cut corners.

Competence: "[Other workers] have all been ticked off as competent […] but we still find that they call quite a lot of times to ask us about what they need to do."
The system and protocols in place to increase safety and regulate operations, such as Train Orders and track warrants, were perceived to be a "a very clever system" (Ppt28_Org4) but "only as clever as the person taking the warrant" (Ppt46_ Org6). Confidence in the competence of other operational staff influenced how much trust was placed in them. Some participants disclosed resentment and wariness in their dealings with track workers, with references to "leaving yourself open" (Ppt8_Org1) and "looking at it from my own bum," (Ppt9_Org1) which alluded to the potential for reprimand, and a desire to take precautions to guard against blame or punishment. Participants perceived poor reliability in the information conveyed to them, for example from track workers who "just simply don't have any idea where they are," (Ppt13_Org3) and from train crew who "might phone you up and they don't know where they are" (Ppt11_Org2). Mistrust of the information shared created a requirement for vigilance and close oversight under the belief that, should their guard be down, an incident was likely, "it's so important to be alert, because we have to watch what the drivers do" (Ppt11_Org2); "drivers think that they all remain competent, they don't do their job properly" (Ppt7_Org1); and, "the only thing between [track workers] doing something crazy is their POs 4 " (Ppt36_Org5). Traincrew could be "hassled" (Ppt48_Org7) for delays to their scheduled run-time and participants were conscious of slowing down track workers to prevent error, "whether I have to be rude to do it or not" (Ppt24_Org4). The perceived lack of competency of other team members resulted in rudeness in verbal exchanges.

Cutting corners: "He was taking a shortcut; he was telling me that the train had cleared it's location and the train wasn't [clear]."
Participants perceived that track workers were prone to "taking a short cut" (Ppt52_Org7) with the implication that their task had not been completed safely or correctly "because they just want to get out there and work" (Ppt8_Org1). This view also extended to train crew: I was a bit nervous with certain drivers. If I knew they were on a certain run, like if they were on the last train and they're going home, you'd want to make sure you give them greens all the way because they're just worried about getting home, so they might go too fast for the area. (Ppt19_Org2) Participants attributed "laziness" (Ppt30_Org5; Ppt37_ Org6) to track workers, indicating "they can't be bothered to go and check the signal numbers" (Ppt13_Org3), and a "blasé" (Ppt46_Org 6) attitude, which impacted their perception about how diligent and trustworthy they behaved as a group: We've had the odd occasion in the past where we've had to chase gangers up at the pub because they've finished work for the day and forgotten to call clear.
A perceived lack of diligence or respect for potential safety risks by track workers led to feelings of hostility, alluding to manifestations of behaviours that serve to steepen the authority gradient:

If you fuck it up, you're dead, this train will run over you and I've got to live with this at train control knowing full well that you weren't listening or whatever. (Ppt46_Org6)
Participants were also wary of "uncooperative" (Ppt9_ Org1) train crew, which affected perceptions of their reliability and working as part of a team. This included failing to report degraded conditions on the network, or doing so in a timely manner, "[the driver] tells me an hour and a half later, well, what else could have happened in the meantime?" (Ppt9_Org1); "all they want to do is get from A to B […] And very often we ask them to do things and whatever and it's an argument" (Ppt7_Org1). A disregard for the professional judgement of train crew was reflected in the perceived lack of conscientiousness in the discharge of their duties.

Dishonesty: "He's giving me false information."
Participants perceived a culture of dishonesty, evasion and deceit towards them from other workers when they came seeking authority. Examples were given of train crew calling to request signals to proceed, but not being ready to depart, inducing a tendency for participants to feel "impatient" (Ppt12_Org3) with them. From the train crew perspective, this was thought to be motivated by a desire to retain their train's scheduled path; however, in practice, this had the effect of causing delays to the network controller's plan, forcing them to make revisions. Similar accounts were given of track workers deliberately breaking safe-working rules, such as attempts to obtain authorisation for track work while still in the car driving to location, "people will try and book on before they're actually at the location" (Ppt13_Org3). This was considered manipulative and relied on participants' "knowledge of the rules and the network" (Ppt13_Org3) to ensure correct procedures were being followed. The belief that track workers misidentified a signal or station in their haste to obtain a work on track authority was ubiquitous, and generalised feelings of mistrust for this group: We've had it in the past where I've put the limits here and his worksite's out here! Next thing a train comes up and nearly runs him over. (Ppt20_Org4) When managing instances where trains exceeded their limits of authority (i.e. a SPAD), participants recounted instances of deliberate obstruction from train crew:

I'll ask where [the train crew] are, if they've gone past the signal, I'll ask them what colour it was. I've had all sorts of answers. Most will say, red. Some will swear that it was green, even though it wasn't and some just won't tell you. (Ppt13_Org3)
Train crew were also perceived as trying to "weasel their way out" (Ppt35_Org5) of a SPAD occurrence by attempting to "down-play" (Ppt35_Org5) their culpability by conveying incorrect information:

[The driver] was asked-are you on those points? He said-no, nowhere near the points […] So, they gave him a form to set back behind that signal. He was sitting on those points. Went back and derailed. (Ppt27_Org4)
The evasiveness of other professional groupings in relation to error, such as in the train crew response to SPADs, or track workers instituting their worksite protections before arrival on site, may be an indicator of an authority gradient manifestation, the fear of reprisal.

Intimidation of other workers
Intimidation of other workers emerged as a behaviour that contributed to an environment in which workers were discouraged from challenging a decision or an action taken by a network controller because they "don't like you questioning their moves" (Ppt1_Org1). Disempowering other workers through intimidation therefore generated an authority gradient: [You could] have somebody withholding information that they could very easily wish to share, but they feel suddenly like they shouldn't, because apparently you're talking and they should listen and not tell you anything. (Ppt36_Org5)

Punitive action: "In the old days, we used to have what we called the 'black book…"
Punitive action taken by network controllers towards track workers included withholding access to the track, delaying commencement of track possessions, and reducing the track time that was available to them. Participants purportedly took this action in response to infractions that negatively impacted the train plan (i.e. the scheduled operation of the day). A delay to train running/services was treated as an infraction and attracted punitive action: [The track worker] delays the train by 3 minutes. Then straight away I'm saying-boh-bong, you've just blown your chance! This bloke, the first time he's already buggered that up I'm straight away going to reduce his time slot (note: "boh-bong" is a reference to the sound effect in the quiz show Family Feud which denotes an incorrect answer). (Ppt20_Org4) Punitive action towards train crew took the form of setting the train up on a longer route, having it traverse loops unnecessarily, or withholding information about the planned route, which made it more difficult for train crew to plan for their journey. Power plays that steepened the authority gradient reduced the likelihood that instructions or directives would be challenged. Any "questioning" (Ppt1_Org1) of network controller decisions were perceived to be tantamount to infractions that attracted punitive action, engendering fear of reprisal. Figure 5 illustrates a SPAD scenario where the train crew (on Train 1) challenged the network controller about a delayed departure, resulting in a heated exchange where train crew became "pissed off" (Ppt43_Org6) with the network controller:

[The driver] was being quite arrogant and he was going on at me for not giving him an initial [signal] out […] after that he was pretty much bitching and moaning to me on the whole journey. (Ppt43_Org6)
Train 1 was on approach to a section of track at C Station that was closed to rail traffic as track workers were on the line. Having completed their work, the track workers cleared the track with the network controller over the radio. In this scenario, the exchange was perceived to have been overheard by train crew; "he can hear over the radio, me getting those workers off and clear" (Ppt43_Org6). The next signal (Signal 1) had not been set to proceed, although the controller was aware that the train crew would assume that it would be:

[The driver] had EVERY (emphasis added based on intonation of voice) chance to be able to slow down in time, but he chose not to thinking that because he approached the signal, I was eventually going to turn it to a proceed. (Ppt43_Org6)
The description of the scenario given (see Fig. 5) reasoned that the signal was not changed due to an emergency call about "something stupid" (Ppt43_Org6) in a "completely different area" (Ppt43_Org6), which diverted the attention of the network controller. One interpretation of this scenario is that relationship dynamics predisposed punitive action in the form of a deliberately delayed signal clearance, potentially with full awareness of the safety criticality in the situation and the risk of a SPAD.

Pressuring compliance: "If you don't want to listen to what I'm saying, well then I'm not going to let you on my track."
For track workers, compliance was achieved by leveraging their need to access the track to complete their work. Similarly, train crew were pressured into undertaking activities that they may have felt uncomfortable with. Following a SPAD, railway procedure dictates that the network controller check "are you alright to proceed" (Ppt49_Org7) with the train crew. More than one instance of pressuring "nervous" (Ppt7_Org1; Ppt13_Org3) train crew into continuing following a SPAD was recounted by participants. According to one, it was acknowledged that a driver was gravely affected by a SPAD, "he was absolutely petrified, from the sound of his voice" (Ppt5_Org1); however, they were instructed to proceed to the next station. Participants alluded to emotional reactions to responses provide by train crew in such situations:

You ask them-are you alright to proceed? Some say, yes, and some say, no. That's okay. Some controllers get beefed up about it. (Ppt49_Org7)
The impact of the authority gradient between network controller and train crew influencing compliance was illustrated by a scenario in which a driver agreed to a network controller's request to restore a signal, even though the controller must have known the driver was at risk of passing the signal at stop: A controller asked the driver if he could take a signal back and the driver said-yeah, yeah, you can take it back. So, he took the signal back, put it back to red in other words, and the driver went past that signal at stop. (Ppt13_Org3) Another example was given of a driver agreeing to continue twice before a manager intervened, recognising the driver was not fit to continue: A driver-had a TPWS 5 overspeed intervention […] by our current rules they can continue, so we told her to continue and we notified her manager. Less than 5 minutes later she had another TPWS intervention and it was an overspeed, right? When we went back and listened to the tapes she sounded really shaken, really like, she didn't sound right. Again, the train controller said-are you right to continue? And reported it to the manager. The manager said stop the train, she's not going any further. (Ppt1_Org1)

Contemptuous regard for other workers
The ways in which participants spoke about other workers provided an insight into the attitudinal culture and demeanour when interacting with these groups: I might get more gangers calling me because I'll give them a small amount of time if they want it whereas there would be certain controllers that they'll hear them on the radio and they'll avoid them, they'll be like-oh, that person's working today so I'm not going to bother, I'll go somewhere where they're not controlling. And I have spoken to gangers and they do do that. They will avoid being in one area because it's a controller they just don't want to deal with because they'll be made to feel like a nuisance. (Ppt45_Org6)

Disparaging language: "Stop being a sook and get going!"
Participants described having to resist the need to "baby" (Ppt42_Org6) track workers who, according to one participant, "look to train controllers to protect them rather than looking to [protect] themselves" (Ppt42_Org6). This perception was broadly shared, with scenario descriptions that evoked an unruly team: Track workers were described as being not "the sharpest tools" (Ppt46_Org6) and not having the "level of education that other people do," (Ppt55_Org8) and as "dickheads" (Ppt8_Org1) and "a real idiot" (Ppt20_Org4). Participants held perceptions that they had to "educate the below rail workers" (Ppt5_Org1) using language and actions synonymous with tough love to protect track workers from themselves. This disparaging commentary illustrated a lack of professional respect.

[Track workers] don't see things and understand things as we do and I think it's important to be sort of empathetic about it. (Ppt55_Org8)
Participants characterised a busy period as "getting smashed with track workers" where track workers were considered a "thorn in their sides" (Ppt38_Org6) while acknowledging that allowing them onto the track was "a necessary evil" (Ppt38_Org6). A blue-collar, white-collar distinction between role perceptions was clear with track workers referred to as "tradies" (Ppt 5_Org1) who were "out there to fix that bit of hole in the road or clunk in the road" (Ppt20_Org4) or only "good for digging" (Ppt46_Org6). In terms of train crew, the language used to discuss them in scenarios illustrated perceptions of train crew egocentricity, "they don't see what's happening around them" (Ppt9_ Org1) and "they just think they're the only ones out there" (Ppt47_Org6). Train crew were perceived to be "arrogant" (Ppt10_Org2; Ppt43_Org6) or "don't listen" (Ppt46_Org6; Ppt44_Org6) because they were "used to their own little world" (Ppt46_Org6). The complexity of train driving and the skill inherent in this job were minimised with condescending language-a clear indicator of the perceived unequal status, "they can physically drive the train but a monkey can do that" (Ppt11_Org2), "he's following signals like he should be getting paid to do" (Ppt20_Org4) and:

Ultimately [train drivers are] paid to follow signals. That's the job of a train driver. That's their ultimate job-follow signals. (Ppt46_Org6)
When making mistakes or having incidents, such as having a SPAD, some participants perceived that train crew "should have judged it better" (Ppt27_Org4) and "that's why they get paid the big bucks, to stop at those red signals" (Ppt20_Org4). For some, the network controller relationship with train crew amounted little more than to "tell [train crew] where to go and they do as you say and keep the trains moving" (Ppt16_Org2).

Hostility: "I'll just tell them to bugger off if they keep harassing me!"
"Negative feeling" (Ppt48_Org7) in participants was generated towards track workers from the pressures associated with unplanned track work. Not receiving advanced notice of track maintenance work, when the time to provide it was perceived to be adequate, engendered feelings of hostility. This was because controllers would need to devise a new plan to facilitate track work between scheduled train services, increasing their workload and the level of risk associated with facilitation of unplanned work in the process. Participants expressed a view that track workers needed to check if their planned work would "fit in with what's happening" (Ppt48_Org7) rather than "just turn up on site" (Ppt46_Org6):

Me and my ten workers are here now and […] they just expect things to happen. (Ppt48_Org7)
"Bargaining for time" (Ppt42_Org6) was described as part of the negotiation process that occurred with track worker parties. One participant illustrated the power play by indicating that "sometimes [track workers will] call you late and you have to tell them off" (Ppt42_Org6).
In participant scenarios, network controllers were irritated by "the driver always challenging and questioning" (Ppt1_Org1) decisions being made if "they think it's the wrong move," (Ppt1_Org1) and therefore felt defensive if their decisions needed to be justified. References were given to "wear the crap" (Ppt3_Org1) in network controller dealings with train crew, with some relationships perceived to be adversarial, "an 'us and them' situation" (Ppt26_Org4). Train crew were frequently portrayed in situations of conflict from a network controller perspective-having a "fight" (Ppt6_Org1) with them, getting into "arguments" (Ppt4_ Org1; Ppt7_Org1; Ppt15_Org3), with them, and encountering "outright aggression with tone of voice" (Ppt13_Org3). Train crew were described as a "law unto themselves" (Ppt46_Org6): [Train crew] feel they're in the right and it was a technical SPAD, 6 they're usually pissed off, they'll be-what the fuck are you doing to me, type reaction.
While participants recounted situations in which train crew appeared more confident in challenging directives, the hostility with which these instances were met is an indicator of the authority gradient manifestations.

Discussion
The results of this study reveal rich insights into network controller dynamics with train crew and track workers, and the attitudes and behaviours associated with controller perceptions of maintaining safety and access across the network. Fear around the potential for error bred mistrust in the capabilities of other frontline workers, for example the efficacy with which their duties were discharged (RQ1). A heightened state of alertness towards risk was in turn linked to authority gradient generation, where compliance from train crew or track workers was imposed in communication by intimidation and positional power in both language and tone (RQ2). Importantly, these opaque facets of communication influenced or shaped how information was delivered and/or exchanged, including any phraseological and protocol-based aspects.
While the application of non-technical skills or human factors based tools do not reveal the underlying nature of relationships between frontline workers in rail, this study unearthed a number of relationship facets that confirm and inform the 'us versus them' mentality reported to exist amongst drivers and network controllers by Naweed (2020, p.10) and, moreover, exposed large power disparities between network controllers and track workers that significantly impacted communication. The findings associated with driver welfare checks following SPADs and pressure to continue driving are also supported by research from the driver perspective . Uncovering and understanding the reasons for the opposition in the interactions between network controllers and other workers, invariably enables deconstruction of authority gradient generation. It also helps build a better understanding of system-wide behaviours. For network controllers, risk perception is at the core of any fear and mistrust participants associated with failures in their system of safe-working and risk profiles associated with different frontline workers, inherent to their roles, amplifies authority gradients by influencing attitudes and behaviours. Dadashi et al. (2021) point to an assessment of risk informing how operators in rail handle decision making. Risk has been discussed as a 'dominating factor in accounting for attitude' (Sjöberg et al. 2004, p. 30) and is a defining feature of the relationship dynamic across the frontline operational team. In this study, risk is regarded as the intersection of probability and consequence (International Electrotechnical Commission 2002), a construct that has also been acknowledged to influence train crew in the ways they undertake their roles (Rainbird & Naweed 2016). Similarly, network controller attitudes and behaviours are characterised by assessment of risk for each professional group, effecting a willingness to acquiesce, diffusion of responsibility and silencing amongst team members, all of which are symptoms of an authority gradient.
The risk profile reflected by network controllers towards train crew was based on high error consequence, but low error probability. Verbal exchange with train crew is minimal with communication and 'authorisation' primarily occurring through electronic signalling systems; however, when verbal exchanges increase with Train Orders or Special Process Authorities, so too does risk perception and level of caution. In comparison, the risk profile reflected towards track workers is based not only on high error consequence, but also on high error probability. In practice, the risk here is greater because increased verbal exchanges coincide with the network controller having to perform more concurrent tasks, which increases the chances of error. As the network controller is more actively engaged in communication with track workers, with an imperative for shared understanding in the absence of being co-located, they are also held accountable for any errors that may occur. Heightened risk is managed through heightened levels of vigilance and wariness (Sanne 2008), and this likely influenced some of the behaviours reported in the scenarios, such as reducing track worker track time as a form of punitive action. This serves to widen the relationship gap between network controllers and other working groups and reinforce power hierarchy, rather than foster concepts of multidisciplinary teamwork.
Increased verbal exchanges create opportunity for interpersonal communication to be influenced by power disparities with corresponding impact on relationships. In comparison to the signal-based authorisation delivered to train crew, track maintenance workers have no choice but to make verbal requests for 'authorisation' to work on track. There are two contributions here to the development of the authority gradients. The first is the hierarchical necessity of an organisational structure in which the network controller provides complete oversight of activities on the network to maintain safety (Cheng & Tsai 2011). Frontline workers must seek permission for their activities from network control, and whether this is to move the train or undertake track repairs, there is an entrenched power disparity between roles. However, while this institutional hierarchy of roles goes some way to explain power imbalances, it does not explain why individuals who perceived themselves to hold less power feel unable to challenge a directive they know to be erroneous, or to volunteer information even when they are sure they are correct. This is covered by the second contribution, which is based on increased verbal communication and reflects the communication act itself. Exposing track workers to greater interpersonal interactions with network controllers creates greater opportunities for prosody to influence interpretation. In their study of workplace bullying, Dzurec, Kennison and Albataineh describe prosody as the way the intended meaning is conveyed through 'rhythm, stress, and intonation, word choice and physical presence ' (2014, p.282). While physical presence is not a relevant factor influencing the relationship, network controller prosody appears to be an influencing factor on whether other workers feel able to speak up, and it may be that track workers are particularly susceptible due to their increased instances of verbal communication. In aviation studies, where the influence of authority gradients has been established for some time (Alkov et al. 1992), flight crew were reluctant to seek clarification or query air traffic controllers if they sounded busy, annoyed, or hurried (Connell 1996). This was noted in the results, with participants reflecting on situations where track workers avoided working in areas under the authority of network controllers with which they had negative interactions.
Participants also asserted that track workers expected controllers to protect them, rather than take responsibility for their own safety. This is in line with Chionis and Karanikas's (2018) study of trainees and professionals in aviation maintenance where the level of risk tolerance in trainees was based on feelings of security and being protected by more experienced professional staff. This point illustrates the power disparity between network controllers and track workers, with participants perceiving that track workers deferred their individual responsibility for safety to network controllers.
The interactions between controllers and other workers depicted in scenarios illustrates how attitudes likely manifest into subsequent behaviours (Hillson and Murray-Webster 2004). The results of this study indicate that when network controllers perceive heightened risk in engaging with a particular group, it likely influences the ways in which they communicate with those groups. More authoritarian exchanges generate steeper authority gradients; thus the more collegiate an interaction, the 'flatter' the authority gradient will be, and the more likely that both parties would feel empowered to challenge an instruction or directive believed to be erroneous. The interpersonal risk associated with fear of ridicule or being perceived as incompetent in organisational environments (Edmondson 2002) impact on feelings of 'psychological safety', i.e. the perceived ability to express oneself without fear of negative consequences (Appelbaum et al. 2016). This psychological safety is integral to an organisational culture seeking to flatten hierarchies and prioritise inclusivity. This study has provided insights into the culture of various work groups, potential deficiencies in psychological safety and impacts of weaknesses in communication, attributable to authority gradients. Further work could consider organisational, and critically, broader systems influences as they relate to a culture of safety and failure.

Strengths, limitations, and future research directions
Focussing on the behaviours and risk perceptions of codependant roles, the SITT method elicited tacit knowledge and expertise that would otherwise be difficult to garner (Naweed 2015), with rich concrete scenarios reflecting working dynamics and relationships across operational teams. The hypothetical nature of this activity allowed participants the safety of self-reflection in dealing with direct experiences, opportunities for imagining potentialities in familiar situations (Naweed 2020), and encouraged consideration of alternative perspectives when evaluating team contribution to incidents, providing insights for the context of error across network control, train crew and track maintenance team groupings. However, as a qualitative study, results are subjective and nuanced, and not all participants displayed the same attitudes and behaviours, or to the same degree. Sjöberg et al. (2004) discuss variations in risk perceptions between individuals as indicative of individuals assigning differing characteristics to the same risks. Instances of network controllers viewing risk characteristics in a dissimilar way may therefore be a function of age, experience, personality, and organisational culture (Chionis & Karanikas 2018). Employing constructivist grounded theory, the role of the researcher in making decisions when categorising the data reflects the researcher's own values and experiences and is therefore inherently subjective (Charmaz 2017b). As one researcher was a former network controller, the experience in this role was a strength of this study. Although some effort was made in "bracketing" and managing bias (Creswell 2013), the study nevertheless carries an inherent limitation and concern for bias.
As the data collected were specific to the Australia and New Zealand context, this created limitation for its application to rail settings beyond this region. As this study did encompass different types of network control operations, which varied from state to state and across countries, findings may be generalisable, but this must be treated with caution, given different cultural contexts.
Although a range of experiences and opinions were presented by network controllers across participating organisations, the attitudinal similarities were prevalent enough to suggest a pervasive culture that transcends individual organisations to extend across the industry. This is suggestive of a hierarchical cultural contribution to the generation of an authority gradient that warrants future research focus. As a key dimension of cultural relativity, power distance theory, devised by Hofstede (1983) and applied in similar systems such as aviation, is a relevant theoretical tool for providing further understanding on these relationships (Luva & Naweed 2021). Further research into team interdependencies with a focus on track worker and train crew groups and the cultural factors that serve to construct relationships across teams is a key direction for future research.

Conclusion
Unequal distributions of power across multidisciplinary rail operational teams can result in authority gradients that obstruct lines of communication, preventing team members from providing input or challenging decisions. While 'communication error' is usually the prognosis in instances of teamwork failures, power relationships between individuals are a significant yet under-acknowledged, causal factor in breakdowns in communication. In rail, the breakdowns in communication that cause errors are risky for train crew and track workers (derailment, collision, fatality), but they are also risky for network controllers operating remotely. Network controllers perceive operational risk in terms of actual harm to frontline workers, but also the risk associated with organisational imperatives, such as deviating from schedules or breaching safe-working rules or regulations. While authority gradients prevent clear lines of communication, network controllers utilise their more powerful position as a tool for closely monitoring and managing the operations of those groups they perceive to be high-risk workers, such as track maintenance workers. Perceptions of mistrust and fear influence attitudes of network controllers when interacting with other members of the operational team and may motivate heightened vigilance and closer monitoring of high-risk worker activity. Equally, in view of the risk associated with relevant work groups, network controllers may adopt behaviours that have the effect of disempowering other work groups, punishing missteps, and achieving compliance through coercive control. These motivations and behaviours are integral to understanding authority gradient manifestation in rail. Further perspectives on the organisational contribution to the generation of authority gradients and the influence of team culture are important future research directions for addressing influences on communication error in rail.
Funding Open Access funding enabled and organized by CAUL and its Member Institutions. The research leading to these results received funding from the Australian Research Council (ARC) in the Discovery Early Career Research Award (DECRA) scheme under Grant Agreement No. DE160101137.

Declarations
Competing interests The authors declare no competing interests.

Conflict of interest
The authors declare that they have no financial interests.

Ethical approval
The study was approved by the Human Research in Ethics Committee of the authors' institution under Approval No.

0000021909.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.