The social organization of errors and the manifestation of rework: learning from narratives of practice

Abstract Despite the considerable amount of research that has examined rework causation in construction, it remains an inherent problem that can potentially result in adverse project outcomes. This situation has arisen as studies have tended to ignore the social organization of errors (i.e. the pattern of relationships and social interactions between and among individuals and teams). Instead, studies have adopted a ‘reductionist view’ of rework causation by identifying its proximal and root causes rather than addressing the conditions resulting in its manifestation. This paper uses a case study approach with a sense-making lens to create a series of narratives of rework events that arose while constructing a transport mega-project. By making sense of the context surrounding the error events, it is revealed rework manifests from failures in ‘negotiated order’ which stems from role ambiguity, misunderstandings, misinterpretations, and break-downs in communications and interactions between people and organizations. As a consequence of these findings, their theoretical and practical implications arising from the research are discussed.


Introduction
'If you are not working, you can't make mistakes' (Riemer 1976, 257) Errors are a 'routine and inevitable part of work' in construction (Hughes 1951;Riemer 1976, 255;Reason 1990).They materialize in 'nearly every learning process' due to 'insufficient knowledge and skills' and as a consequence of 'inappropriate goals and plans, interruptions during the action process, and inaccurate interpretation of system feedback' (Heimbeck et al. 2003, 333).Generally, the adverse effects of errors are brought about by unintended 'mistakes at work' (Hughes 1951, 320).All forms of work during the construction of a physical asset, regardless of routines and activities, are faced with the probability of making an error 'in which the variables are skill and frequency' (Hughes 1951, 320).Thus, it follows statistically that 'the more times per day a [person] does a given operation, the greater chance of doing it wrong' (Hughes 1951, 320).
Without making errors, learning and innovation cannot occur (Frese and Keith 2015;Naveh and Lei 2019).While there are positive consequences of errors, there are equally negative effects, which can adversely impact the performance of organizations, projects, and people.The negative consequences of errors are abounding and include reductions in profit and productivity, losses in reputation, accidents, fatigue, and stress (Love, Matthews, and Fang 2020).Some errors are inconsequential and go unnoticed; others hinder the progress of construction and increase project costs requiring rework (i.e. after error occurrence).But it is those of a severe nature that go undetected, which can lead to 'unexpected future difficulties' (Riemer 1976, 256).Such difficulties can result in engineering failures causing considerable expense and disruption and, in the worst case, the collapse of structures and the loss of life.As a result, studies seeking to understand the consequence of errors manifesting as engineering failures (Kaminetzky 1991;Pfatteicher 2000;Gardner, Huh, and Chung 2002;Minor 2005;Ellingwood 2006;Ba zant and Verdure 2007;Garg, Chandra, and Kumar 2022) and accidents have been ubiquitous (Mitropoulos, Abdelhamid, and Howell 2005;Filho, Waterson, and Jun 2021;Mohandes et al. 2022).
Since the seminal works of Riemer (1976), research examining the social organization of errors in construction and their influence on quality has been limited (Atkinson 1999;Wantanakorn, Mawdesley, and Askew 1999;Atkinson 2002;Love and Josephson 2004;Love et al. 2022aLove et al. , 2022b)).More specifically, the only empirical-based study can be found in Love et al. (2022c), who examined how a water infrastructure alliance, when realizing that it had a rework problem, was able to transition its error culture (i.e.how errors and their consequences are dealt with) from a focus on 'error prevention' (i.e.errors can and need to be prevented) to 'error management' 1 (i.e.errors happen).
In this paper, the experience and practice with managing errors in a transport mega-project are drawn upon to help develop a body of knowledge about the social organization of errors in construction.Thus, the following research question is proposed: Why do errors and rework occur, and how are they managed?This question is addressed as construction organizations typically go to great lengths to prevent errors.They are viewed as unfavourable, and therefore a zero-tolerance towards them is often adopted, particularly errors relating to quality and safety issues (Love, Smith, et al. 2018;Love et al. 2022c).
Errors are an effect or symptom of an organization and project's conditions under which people work and are embedded into the social organization (i.e. the pattern of relationships and social interactions between and among individuals and teams) of practices (Riemer 1976;Reason 1990).To this end, the social organization forms the basis for examining how errors and rework, a product of their occurrence, are dealt with in projects.
The paper commences by briefly reviewing key concepts associated with human error theory (Section 2).A detailed review of the development of human error theory can be found in Read et al. (2021).The theoretical backdrop provides a basis for introducing the case study (Section 3) and a series of narratives that examine the nature and management of errors and how they manifest in rework (Section 4).Next, the insights gleaned from the case study are discussed with specific reference to its theoretical and practical implications (Section 5) before concluding the paper (Section 6).

Human error theory
Committing errors or 'mistakes at work' is common practice amongst all people, irrespective of their occupation, professional statute, level, and experience (Watson 1913;Greenwood and Woods 1919;Hughes 1951;Riemer 1976: Reason 1990).However, the word error tends to be associated with individual misgivings.The common phrase 'to see the errors in one's ways', for example, 'implies a person has engaged in some wrongdoing' (Armitage 2009, 194).Here the mainstay is the intention behind a person's (in)actions.Thus, Reason (1990) views an error to be a 'failure of a planned action to be completed as intended-without the intervention of some unforeseeable event; or the use of a wrong plan to achieve an aim' (9), aligns with the works of Rassmussen and Jensen (1974), Perrow (1984), Norman (1988) and Frese and Keith (2015).
Figure 1 provides a well-accepted nomenclature in the literature for error in its various guises, grouped into unintended and intended actions (Reason 1990;Read et al. 2021).Reason's (1990) error nomenclature builds on Rassmussen and Jensen's (1974) human performance classification (i.e.skill-based, rule-based, and knowledge-based) and Norman's (1988) research on errors in action, enabling an 'automatized cognitive phenomena which illuminate the usually covert, intrinsically human process of error' to be developed (Armitage 2009, 195).

Unintended actions: slips, lapses, and mistakes
In complex work environments, such as construction, cognitive failures (i.e.slips and lapses) (Carrigan and Barkus 2016) and mistakes, collectively referred to as active failures, are bound to happen (Reason 1990).Thus, it is impossible to implement mechanisms to prevent their occurrence.Understanding adverse events in construction require that distinctions are drawn between various error types.Each has different origins and demands different strategies for remediation (Parker and Lawton 2003, 453).
Slips are visible and unintended.They are errors in the human-automation process where there is no conscious control, and the individual's normal routine is disturbed, even though the original mental plan is correct.Workplace routines are mastered through repetition, as 'discrete actions are translated into one coherent and automatic procedure' (LaBerge and Samuels 1974;Armitage 2009:196).However, experienced people are more prone to slips than those new to a routine due to their cognitive loading of 'preprogrammed instructions (or schemata)' (Reason 1990;Armitage 2009, 196).
A lapse is another type of error and is simply forgetting, which involves a failure in memory retrieval and can happen in most healthy people.Stress, fatigue, depression, lack of sleep, an unhealthy diet, and dehydration are a few causes of people's forgetfulness.
In stark contrast to a lapse, a mistake arises when an 'action proceeds as planned but does not achieve its intended outcome as the plan was wrong' (Reason 1990;Armitage 2009, 196).Riemer's (1976) ethnographic study of errors on building sites led to four types of mistakes\being routinely identified during construction: 1. Miscalculations are 'probabilistic mistakes' which arise due to failures in individual perception, interpretation, or understanding and frequently transpire (Riemer 1976, 257).They are common in errors in judgement, with Riemer (1976) suggesting 'everyone makes them at one or another, and they are to be expected' (257).In essence, miscalculations are akin to cognitive failures.2. Hold-ups are failures in social organization occurring 'through blockages or break-downs in a communication network between essential persons or groups' (Riemer 1976, 257).Distortions, misinterpretations, and misunderstandings can materialize when work-orientated communication networks and information flows are ineffective and inefficient (Westrum 2014;Love, Smith, et al. 2018).
In this instance, a situation of social disorganization can prevail and hold-up a project's progress.3. Circumstantial errors occur due to the immediate and prevailing conditions surrounding work.Thus, people are likely to err when subjected to production pressure and competing demands, such as 'time vs. quality' (Love et al. 2022a).While circumstantial errors are akin to hold-ups and can be socially controlled, differences exist.Hold-ups result from a break-down in communication between parties.Contrastingly, circumstantial errors occur when clear and explicit communication expectations are contradictory (Riemer 1976, 260;Love et al. 2022a).4. Natural errors are unforeseen events for which no one person or team can be held liable or responsible or, in the legal and insurance vernacular, can be described as 'Acts of God' (i.e.inclement weather) (Riemer 1976, 257) Mistakes can also be categorized as rule or knowledgebased (Rassmussen and Jensen 1974;Reason 1990).In the case of rule-based errors, a practitioner may misapply a rule that worked in a previous situation (e.g.design) due to a changed condition.Relatedly, an imperfect rule may have remained uncorrected and formed part of a practitioner's problem-solving toolbox (Reason 1990).Similarly, knowledgebased mistakes emerge when a practitioner encounters a novel situation outside the range of their learnt problemsolving routines (Reason 1990).
Errors can also manifest at the team and organizational levels.Accordingly, team errors 'can occur due to the joint effect of antecedents across individual and team levels' (Lei, Naveh, andNovikov 2016, 1322).Several scenarios can result in team errors occurring in projects which include (Sasou and Reason 1999): Where the entire project team does not detect an error and work continues; An individual commits an error that goes undetected.The team all decides on a course of action, unaware of an error; and An individual error is detected, but the team decides not to correct it and continues their work.
In the case of organizational errors, they have been defined as the 'actions of multiple participants that deviate from organisationally specified rules that can potentially result in adverse' outcomes (e.g.accidents, litigation, and reputational loss) for an organization, especially in highstakes settings, such as construction (Goodman et al. 2011, 154).Hence, a rudimentary 'feature of an organizational error is that multiple individuals deviate from the expected organizational practice' (Goodman et al. 2011, 154).

Intended actions: violations
Contrastingly, violations are deliberate (or intentional) deviations from rules and established procedures (Frese and Keith 2015).Whether a violation is an error is debateable with such a discussion outside the scope of this paper.Violations are deliberate deviations from rules and procedures 'established to restrict self-interested behaviour and protect organizational members from the predations of others' (Busby and Iszatt-White 2016, 36).They originate from psychosocial factors, such as individual and organizational motives rather than cognitive ones (Reason, Parker, and Lawton 1998;Parker and Lawton 2003;Parker and Lawton 2006).The propensity for people to commit a violation increases when 'risks thought to be controllable (i.e.preventable by personal action) are likely to evoke unrealistic optimism about judgments in susceptibility' or, more colloquially described as the 'it won't happen to me' phenomenon (Weinstein 1984, 158;Armitage 2009) While organizations abhor deviant behaviour, it can also be considered to be 'pro-social' and 'a way of testing rules and the truces that surround them' (Mills and Murgatroyd 1991;Busby and Iszatt-White 2016, 36).For instance, breakthrough innovations may occur, particularly during design.However, this is seldom the case once construction has physically commenced.When there is an intention to cause damage, the violation is sabotage (Reason 1990).But, when there is some degree of intention but no desire to damage, then a violation can be categorized as routine (i.e.habitual, forming a part of an individual's behavioural repertoire) (Reason 1990, 195) or exceptional (i.e.singular violations occurring in a particular set of circumstances) (Reason 1990, 196).2.3.Models: a shift from an individual to system focus Reason (1990) replaced Norman's (1981Norman's ( , 1988) ) emphasis on the error proneness of individuals, 'who could be blamed and weeded out with a more sophisticated understanding of the complexities of the interactions between individuals and systems' (Parker and Lawton 2003, 453).Hence, there has been a shift from a person to system approach, which 'concentrates on the conditions under which individuals work and aims to build defences to avert errors or mitigate their effects' (Reason 2000, 768).The person approach marries with Dekker's (2006) 'bad apple theory' (i.e.'old view' where human error is the cause of accidents), which maintains that (1):

Rework
the workplace of complex systems, such as construction would perform well were it not for the erratic behaviour of some unreliable people (i.e.bad apples); people cause accidents: humans are primary contributors to more than 66% of them; failures are unpleasant surprises.They are unexpected and do not belong in the system.Failures are introduced to the system only through the inherent unreliability of people.
The paradigm shift from a person to a systems approach led to the creation of Reason's (1990Reason's ( , 2000) ) Swiss Cheese Model, which forms the foundations of human error theory (Parker and Lawton 2003;Armitage 2009).According to Reason's (1990Reason's ( , 2000) ) Swiss Cheese Model, a series of barriers prevent hazards from causing human losses.Each barrier contains holes akin to unintended weaknesses-hence the Swiss cheese metaphor.The holes (weaknesses) in the barriers open and close at random.When all holes are aligned by chance, the hazard reaches a person and causes harm.But it goes without saying that the Swiss Cheese Model has become the dominant paradigm for managing safety risks and analyzing errors resulting in accidents in safety-critical industries (Mitropoulos, Abdelhamid, and Howell 2005;Perneger 2005;Read et al. 2021).
Despite the popularity of the Swiss Cheese Model, its inbuilt assumption of linear causality (or sequential causality) of events has been criticized as it oversimplifies the dynamic and inter-relatedness of multiple events that often result in accidents (Leveson 2020).More often than not, errors are rarely caused by single factors but instead 'elements in a concatenation' (Armitage 2009, 197).Additionally, the Swiss Cheese Model emphasizes preventing human error rather than hazards and changing a system's design (Leveson 2020).In this instance, human error is viewed as a symptom of a system that needs to be redesigned.So, Leveson (2020) argues that behaviour changes can only occur when a system's design is altered.
Other safety approaches based on resilience, systems theory, and complexity science s have also evolved (Martinetti et al. 2019;Leveson 2020;Read et al. 2021).For example, the notion of Safety-I, referred to as 'simplistic systems thinking', considers system influences on behaviour but from a mechanistic standpoint where errors can be predicted and the reliability of human failure calculated (Mannion and Braithwaite 2017, 686;Read et al. 2021).More specifically, Safety-I assumes that 'things go wrong because of identifiable failures or malfunctions of specific components: technology, procedures, the human workers and the organizations in which they are embedded.Therefore, humans-acting alone or collectively-are viewed predominantly as a liability or hazard, principally because they are the most variable of these components' (Hollnagel, Wears, and Braithwaite 2015, 1).
Like Dekker's (2006) advocacy for a move from an old to a 'new view' of human error (i.e.where it is a symptom of trouble deeper inside a system), there has been a shift in thinking and practice from Safety-I (i.e.why things go wrong) to Safety-II (i.e.why things go right) (Hollnagel 2014;Hollnagel, Wears, and Braithwaite 2015).Under the umbrella of resilience, Safety-II 'assumes that everyday performance variability provides the adaptations that are needed to respond to varying conditions, and hence is the reason why things go right.
Humans are consequently seen as a resource necessary for system flexibility and resilience' (Hollnagel, Wears, and Braithwaite 2015, 2).A system is therefore deemed to be resilient if 'it can adjust its functioning prior to, during, or following events (changes, disturbances, and opportunities), and thereby sustain required operations under both expected and unexpected conditions' (Hollnagel 2016).Leveson (2020) has been critical of Hollnagel's Safety-II, suggesting it follows a linear-chain-of-events model and ignores a systems design.Moreover, Leveson (2020) considers Safety-II to be the 'opposite of a systems approach (or a sociotechnical approach) as the technical seems to be excluded from playing an important role in safety.In reality, human flexibility and resilience are always limited by the overall system design system in which humans exist and work' (104).
As an alternative to Safety-II, Leveson (2020) advocates Safety-III, based on Systems Theory, which emphasizes designing a system to prevent and control hazards.Initially published in 2002, Leveson introduced the Systems-Theoretic Accident Model and Processes (STAMP) to analyze accidents, particularly system accidents.In Leveson's (2002) model of safety, 'accidents occur when external disturbances, component failures, or dysfunctional interactions among system components are not adequately handled by the control system' (250).A detailed examination of Safety-III and its underpinning theory, STAMP, can be found in Leveson (2004Leveson ( , 2012Leveson ( , 2020)).

Moving beyond human error and safety: emergence of Quality-I to Quality II
A cursory glance at the literature reveals that human error theory has been predominately used to analyze accident causation and manage safety risks (Read et al. 2021).Equally, human theory can analyze quality deviations, failures, rework, and associated risks.But there has been a proclivity to overlook this line of inquiry even though a symbiotic relationship exists between quality and safety (Wanberg et al. 2013;Love et al. 2022aLove et al. , 2022b)).
From both quantitative and qualitative perspectives, research has shown that when errors and violations materialize and necessitate rework, then the likelihood of a safety incident occurring significantly increases (Wanberg et al. 2013;Love, Teo, andMorrison 2018, Love, Teo, et al. 2018).For example, Love, Teo, and Morrison (2018) study revealed an association between injuries and rework was significantly strong (q ¼ 0.631), indicating that '63% of the variance in injuries can be attributable to changes due to rework' (275).To this end, when rework is reduced, safety performance will improve.Building on Dekker's 'Old and New View' ( 2006) and Hollnagel's Safety-I and Safety-II (2014), Love et al. (2022c) developed a conceptualization of human error and quality through the lens of Quality-I and Quality-II.Table 1 aligns Quality-I with error prevention, the dominant mindset in construction projects and Quality-II with an error management orientation, which has been observed in Alliances (Love et al. 2022b;Matthews, Love, Ika, et al. 2022).Alliancing promotes a 'no blame, no fault' culture and aims to engender 'collaboration, knowledge, sharing and organizational learning' (Lloyd-walker, Mills, and Walker 2014, 229).

Research approach
Due to the research context's novelty, an exploratory case study approach is adopted to examine the 'how' and 'what' of the social organization of errors (Stake 1995).To recap, this research addresses the following question: Why do errors and rework occur, and how are they managed?In this research, rework is defined as 'the total direct cost of redoing work in the field regardless of the initiating cause', which excludes explicitly change orders (variations) and errors caused by off-site manufacture (1078).
A sensemaking lens focussing on the individual within an organizational setting (i.e. program alliance) is adopted to garner an understanding of the nature and management of errors.It takes a hermeneutic approach to understand an individual's experiences with errors and rework and how they are managed (Dervin, Foreman-Wernet, and Lauterbach 2003).As a result, the research focuses on understanding and promoting agency, enabling the individual and the researcher to engage in an unfettered communicative process.

Context: transport mega-project
The case study setting in this paper is a transport mega-project that comprises a program of works initiated by an Australian State Government to remove existing and construct new road and rail infrastructure throughout the metropolitan area of a major Australian city.The project, commencing works in 2015 with an original completion date of 2022 and now revised to 2025 due to additional required works, is being delivered using four program alliances.For reasons of confidentially and political sensitivity, we cannot provide any more detail about the nature of the project.
Notably, the rationale for the case's selection for this study is based on the authors' ongoing involvement in an alliance's continuous improvement initiative, seeking to reduce its waste (i.e.non-value adding activities).When the research was undertaken, the alliance had completed four projects, and four were being constructed.
Design changes and errors in the Issue for Construction (IFC) drawings and non-conformances have resulted in the need for rework during construction, which the alliance seeks to mitigate.There is a mindfulness by the Alliance Leadership/Management Teams (ALT/AMT) and project teams that many errors requiring rework are not being captured.Apart from non-conformances requiring rework, no measurement of it, per se, is undertaken due to the absence of standard terminology and an integrated information system to record and document its costs and causes (Matthews, Love, Ika, et al. 2022).Additionally, as the Alliance had implemented a 'no blame, no fault' culture, it had unconsciously engaged with and enacted organizational practices that marry with an error management orientation (Love and Matthews 2022).This research delves deeper, moving beyond identifying and describing a project's error culture using narratives to unearth and interpret the 'why and how' of error making and its consequences.

Data collection
Case study research typically draws on a myriad of data sources to examine phenomena (Stake 1995).Unstructured interviews form the primary data sources to determine the 'how' and 'why' of errors and rework events.Notably, some leading questions were introduced to safeguard the ensuing dialogue that remained within the bounds of the study.The interviews aimed to stimulate a process of 'spontaneous communication' and interrupt the hegemony of power assumed within the context of doing research by encouraging self-reflective communication (Dervin and Naumer 2009, 879).
Members of the alliance (i.e.Owner Participant, OP, and Non-owner Participants, NoPs) and its subcontractors were randomly selected and invited to participate in the research by being interviewed.Three months were set aside to conduct the interviews.A total of 30 invitations were distributed via email to members performing various roles within the Alliance, with all agreeing to be interviewed.A total of 19 interviews were conducted within the time frame as potential interviewees could not find time in their busy schedules to accommodate an interview.
A list of participants interviewed juxtaposed with examples of rework identified are present in the Supplementary Appendix of this paper (Supplementary Table A).Notably, interviews with an OP representative could not be obtained.Two researchers jointly conducted the interviews between March and May 2021.Due to the Coronavirus 2019 (COVID-19) outbreak, travel restrictions resulted in the interviews being performed using Microsoft Teams.Each varied in length from 26 to 60 min and (%12 h in total) were digitally recorded and transcribed (%110k words in total) (Table 1).
Copies of the interview transcripts were made available for comment by the interviewees.
Documentary sources, such as non-conformance reports (NCRs), site diaries, (internal) requests for information (i/RFI), and punch lists were also made available.Researchers are seldom provided access to such documentation inasmuch as it is typically treated as confidential.But in this case the Alliance had recognized the need to address rework and its likely adverse impact on safety performance due to the Barwon Water Alliance's (BWA) experiences (Love et al. 2016(Love et al. , 2022c)).
Each interview commenced by asking the interviewee to identify an event where an error occurred and led to rework having to be performed.Then, from the interviewee's perspective, they were asked to describe why the events materialized.Several interviewees identified the same event.As perceptions varied, the researchers attempted to make sense of events juxtaposed with the supporting documentation to craft a series of rework narratives using the process presented in Figure 2.
While interviewees identified numerous instances of rework, only those mentioned repeatedly (i.e.having significance) provided a means to make sense of their insights and people's (in)actions.Each narrative is framed by the type of error that arose during construction.Noteworthy, error types often interact with one another and shoehorning an event according to the nomenclature presented in Figure 1 may be considered reductionist (Love et al. 2022d).Thus, Riemer's (1976) classification is used though 'miscalculation' is replaced with 'cognitive failure', and the generic term 'mistake' is used.Thus, the reification of error renders the phenomena of rework causation easier to comprehend in this case, particularly in the context of team and organizational errors.

Research findings
The Alliance was procuring its projects in an environment characterized by volatility (e.g.changing stakeholder demands and needs), uncertainty (e.g.difficulties in being able to predict and plan work), complexity (e.g. the number of interfaces), and ambiguity (e.g.incomplete information) (VUCA).Amongst many of the challenges the Alliance faces while delivering its projects (e.g.meeting design and construction deliverables, ensuring safety, minimizing congestion, and disruptions to rail services), rework was a problem that remained difficult to address.Put simply, there was limited understanding of its causation and how to capture such knowledge and embed it within the Alliance's continuous improvement initiative.
Questions about errors and rework were asked to help interpret and experience the interviewee's world rather than try to explain or predict that world.This way, understanding experiences provides researchers with an 'insider view' and a more profound understanding of the participant's perspective.This 'insider view' does not reveal objective 'truths' but allows the researcher to make sense of the interviewees' subjective' realities unearthed within their social context (i.e. the Alliance).
While people were encouraged to communicate and share their experiences with errors and rework, this tended to occur in silos within their respective teams rather than across the portfolio of the Alliance's projects.The corollary is the same mistakes that were repeatedly made.Reinforcing this point, an engineering coordinator who possessed a broad overview of issues in each of the Alliance's projects as they were privy to all package information through management of constructability inputs to design, assessment of scope growth, construction RFI's and resolution of technical issues, remarked: 'I coordinate the RFIs across projects.We often see repeated issues being raised across projects.Stakeholders often want to rework items [re-design], so we engage them in our design reviews.Repeatedly, the same design and coordination issues are common, but we are not learning.[the coordinator then went on to say].You know concrete cracking is a major rework area, and if you can learn from an NCR, design can be changed, but you can't find this information out as we don't have a database to search for keywords.'When the Alliance commenced its works in 2017, the likelihood of having to perform rework was not an issue that had been initially considered and accommodated for in its cost contingency.Yet, after the construction of their first project and several NCRs requiring rework (e.g.damaged railhead), it was realized that the risk of such events needed to be accommodated in a project's cost and time contingency.
Affirming the Alliance's proactivity, a planning manager asserted: 'Actually, we manage the risk of rework when we formulate a schedule.Just take a concrete activity where there is likely to be rework.We benchmark previous work [projects], allow a reasonable time, and resource the project accordingly'.
Even though it was understood and known that concrete works were prone to rework, the Alliance encountered difficulties in learning to mitigate its occurrence (Supplementary Table A).Indeed, the problems with concrete should be expected as it is the primary material used in constructing footings and piles, bridge structures, retaining walls and the like in each of the projects delivered by the Alliance.Hampering the Alliance's ability to learn was its inability to routinely monitor rework causes and consequences.It had no system to capture, besides the documentation of NCRs, and consolidate all such data for decision-making purposes (Matthews, Love, Porter et al. 2022).
Generally, the errors identified were unintentional.They were products of the environment where projects were procured as production pressure and resource constraints prevailed.While production pressure is an ever-present reality in brownfield rail projects requiring meticulous planning to ensure minimal disruptions to passenger and freight services and the network is kept to a minimum, political forces were at play.
Production pressure emanated from a drive from the client to deliver election promises focussing on alleviating congestion and improving road safety.Furthermore, projects were subjected to benchmarking by the client's administrative arm in key result areas (KRA), pushing to consistently improve project performance (e.g.reduce time and cost).
Resource constraints materialized from an overheated market and international and state border closures due to COVID-19 reducing access to a skilled and experienced workforce.

Unintentional errors: non-conformances
Two repeatedly identified examples of errors and rework by interviewees epitomizing the Alliance's work environment and used to address the paper's research question are the: (1) architectural screens (i.e.hold-ups); and (2) non-compliant pile footings (i.e.circumstantial).Other events consistently identified were the damaged railhead (i.e.circumstantial) and missing ligatures (i.e.circumstantial) identified in Supplementary Table A. Before examining the two specific events, and in addition to examples briefly identified in Supplementary Table A, several samples of NCRs requiring rework are now presented: Incorrect installation of a pile reinforcement cage.A cage with an 890mm diameter, N12 spiral@150 pitch, and 8 x N32 vertical bars had been incorrectly installed instead of N12 spiral@125 pitch with 12 x N32 bars.Previously 900mm piles had been specified, but issues during their installation had occurred.So, an 890mm cage had been specified, but the initial review of the shop drawings was not completed as there was an overwhelming need for it to be fabricated due to time constraints.However, previous shop drawings specifying a 900mm cage were accidentally used to order the cage.Thus, in accord with this action can be attributable to a lapse and the omission of a planned item.Upstand/plinth low starter bars: Upon completion of a pour for a reinforced concrete slab, it was revealed that a mechanical plinth and adjacent upstand starter bars had been set too low.The plinths had been set with a laser, but the low instead of the high point had been used, Why and how did errors and rework occur?How are they managed?
Identify events (i.e., repeated) Identify events in documentary sources

9
(1) (2) Making sense of the social context resulting in bars being 30mm below their required height.
Additionally, the upstands collapsed as somebody had stood on them during the pour.Surprisingly, inspections to check the reduced levels had not been undertaken before the pour.The pressure surrounding the schedule meant staff were stretched, and there was an explicit assumption that laser results were correct (i.e.mistake).
The cost of rectification in this instance was AU$1500.
Starter bars not installed on the slab: A subcontractor had initially forgotten to install five starter bars in the slab for an underpass ramp's retention wall.Then, after realizing this oversight, the subcontractor drilled and grouted the bars into the slab (i.e.mistake).This perceived remedial action was non-compliant with the IFC drawing.In this instance, an error resulted in another error occurring.An iRFI was raised and then followed up with the issue of an NCR.The estimated cost of rectification was AU$500.
Documentation, specifically NCR and corrective action forms, were checked to obtain additional information about events raised (e.g.root cause, severity, and '5 Why Process').However, many NCRs raised contained scant information, making it impossible to garner detailed information about costs and remedial actions.In the case of the '5 Why Process,' it was often observed that the response for an NCR was 'to be filled in by the engineer'.The raising of an NCR occurs at a specific time, with engineers being required to complete it as work is remedied and closed out.Despite this requirement, NCRs were seldom updated, albeit the form itself. Engineers perceive this to be a time-consuming and non-value-adding activity, as they should focus their immediate attention on ensuring construction activities adhere to the plan and conform to standards.The absence of such data stymies an organization's decision-making, risk management, benchmarking, continuous improvement, and learning ability.

Architectural screen rectification
In the architectural screen example to be examined here, the issue of interface management (IM) was brought to the fore and repeatedly branded as a source of frustration, often exacerbating the error-making and treatment process.In short, an interface is a point of interaction or a common boundary between systems, people, and parties involved in the project's delivery.Thus, IM is a systematic approach to managing risks that exist at an interface (e.g.architecture and structural design and civil and landscape design. The Alliance comprises differentiated design teams, requiring high degrees of integration.Identifying and managing the numerous physical and system interfaces in the Alliance is arduous due to its size, the complexity of work, and the number of interfaces (e.g.stakeholders) it must deal with daily.
Cognisant of the challenges with IM, the Alliance developed and implemented a process to address issues, such as interface requirements specification, detail interface requirements, interface actions and requirements, alignment to design reviews, design configurations, and defining responsibilities so that risks could be managed and controlled at their source.Potential design and engineering interfaces problems recognized by the Alliance were: (1) gaps in scope, which can result in the detailed design being comprised; (2) physical clashes between design elements (e.g.services and structures); (3) operational clashes between design elements; and (4) errors and omissions, which often materialize due to staff's inexperience.
Despite the Alliance having a formal IM process in place and the NoPs aware of its importance, a significant error still occurred while designing an architectural screen for a bridge.In sum, the architectural screens had to be dismantled.The existing frame, brackets, and connections had to be strengthened to meet the revised specification at an estimated cost of %AU$616,000 (i.e.%AU$28,530 and AU$587,480 design and construction costs, respectively) requiring an additional three months of off-site work.But why was this the case?
The architectural screens had been designed, manufactured, and installed on a bridge over a rail line in accordance with the IFC drawings.After their installation, they began to deflect profusely during high winds raising concerns about their structural integrity.The senior project engineer (SEP), noticing and reporting this problem, agreed to be interviewed, describing, in their opinion, why this happened and how they were rectified.Yet, before identifying and explaining the whys and wherefores of the event, the SEP made the following comment, not wanting to cast blame, stating: 'I suppose the biggest rework event we've been dealing with here at [xxx] and without throwing people under the bus is the architectural screens, which we constructed as per the IFC drawings'.
The SEP emphasized that a 'no-blame' mindset was entrenched into their site management team's mindset, pointing out that 'no one wants to go finger-pointing' and 'we treated it as a mistake and were diplomatic'.But the SEP expressed their concerns about the prevailing production pressure in the project, describing it as 'intense' and suggesting it was impacting the performance (i.e.ability complete tasks effectively) of their immediate project team and subcontractors.Besides the SEP, others reflecting on the architectural screens event were a quality manager, an engineering coordinator, and a project engineer.While the aforementioned interviewees were only indirectly involved with the event, they possessed explicit knowledge of the case as they had roles in coordinating responses to i/RFIs and the NCR.Their insights are thus drawn upon to help make sense of what transpired.
Unfortunately, the architects engaged in the design of screens were unavailable; they had a minor role in the Alliance, acting as a sub-consultant as most works were of a civil nature.Furthermore, the structural engineers were also not interviewed.Accordingly, the views of the presented rework event may be construed as one-sided, only representing part of the story.But this should not be seen as a pretext to detract attention away from the quest to garner an understanding of the rework that manifested.Reference to the NCR and iRFI's raised and documented were used to verify the accounts of those interviewed.
There was a perception that the architect had experience and knowledge in designing architectural screens on bridge structures.Consequently, when designing the screens, they did not actively engage with the structural engineer regarding the steel and aluminium frame grade, the bolt and rivet, and the welding specification.In doing so, the architect assumed sole responsibility for their design, eschewing rigorous interface design and coordination (IDC) meetings and reviews.Bolstering this point, the SEP remarked: 'We did some digging, and it turns out some of the design details were lacking and hadn't been checked, resulting in us having to pull down the screens and send them back to the workshop'.
The SEP then went on to make the following comment: 'We brought it [the problem with the screens] to the attention of the designers.The designers then checked their details drawings.They found that a particular detail hadn't been checked, which was inadequate'.
For the site management team, it was 'frustrating' having to do something twice as they had constructed and installed the screens as per the IFC drawings.The SEP stated, 'we don't dwell on mistakes.When stuff-ups occur, we try to find solutions and keep moving forward'.In the case of the screens, the connections were 'blatantly wrong'.William Blake's wise words come to mind here as he stated, 'hindsight is a wonderful thing, but foresight is better'.Bearing this in mind, should the site management team have identified the errors before installing the screens?If truth be told, the SEP and site management team had no idea whatsoever that there were problems with screen connections, bolts, and brackets.
The SEP could only assume that the architect did not think there was any need for the structural engineers' input and that they had relied on a standardized design solution drawn from projects of a similar ilk to respond to production pressure.It must be stressed here that the SEP's proposed explanation is speculative as the architect's perspective has not been heard.But putting this aside, the lack of design coordination, communication, and reviews before the IFC resulted in the quality manager implying that both the architect and structural engineer had placed themselves in a position where the 'right hand didn't know what the left hand was doing'.It was suggested by both the SEP and engineering coordinator that if the reviews before IFC had been undertaken, the errors in the design would have been identified.
When the steel fabricator produced the shop drawings based on the IFC, they should have been checked and approved by the structural engineer.But instead, the architect approved them, as they corresponded with the dimensions and specifications of the IFC drawings.The blocking of communicative action with the structural engineer in this instance, whether deliberate or in haste, to make their work efficient and meet schedule demands further exacerbates the impact of the original mistake (i.e.application of a bad rule).
This condition of social disorganization resulted in a hold-up being experienced.
The hold-up, in this case, portrays a distorted outcome arising out of a failed 'negotiated order' 2 where work was treated separately rather than being part of an integrated package.While normative rules, particularly the requirement for design checks and reviews, existed, the fluid and often ambiguous nature of design coordination requiring adaptation and modification was distorted as the architect seemingly pursued their own goals, which resulted in a distortion of a 'negotiated reality' and a break-down in communication.

Learning from hold-ups
Consequently, a thorough review of the design and fabrication process was undertaken due to this error and subsequent rework.In the true spirit of the Alliance's ethos, no party was blamed for the event.However, it was recognized that there was a need to improve collaboration and better manager interfaces.Thus, to support the existing IM process, an interface matrix is now required to be established before a project commences, and interfaces are explicitly identified in Package Action Plans.
Additionally, regular IDC meetings between the architecture and structures teams must be undertaken to ensure scope and integration across packages, determine accountability, and identify reviewers.Here the aim was to increase the IDC process's rigour and improve its sub-consultants technical management.In conjunction with these technicallyoriented improvements, the Alliance actively shared knowledge about this event through its lesson-learned workshops.It also reinforced the collaboration needed to deliver such a challenging and complex set of projects.There is a general belief within the Alliance that engaging in the process of collective learning can help create an environment of 'getting it right the first time'.
The Alliance has learned from the rework that emerged from the screens, with interviewees acknowledging it was due to a break-down in communication between parties and defining accountabilities.Helping to facilitate communication, the Alliance has increasingly focussed on facilitating psychological safety by fostering open reporting (e.g.nonconformances and uncooperative and unsafe behaviours), active questioning (e.g. to understand why a course of action is being undertaken), speaking up, and sharing of insights and concerns.

Non-compliant pile footing
The causal context of the non-compliant pile footing example presented in Figure 3 differs significantly from what transpired in the architectural screens.Still, it is representative of many of the NCRs that have been raised.The error is circumstantial, resulting in a rework cost of AU$500.This type of NCR is a repeat, though raised on another project.Its cost was relatively minor, and schedule lessons had not been shared across projects.As a matter of fact, there was no way of knowing when the piles were being constructed that an NCR for such work had been previously raised, as no system was in place to search for or notify site management of potential quality problems.
During an inspection of the site by a quality manager, it was observed there was a problem with the pile.Unbeknownst to the quality manager, no NCR had been raised for the pile.The quality manager could not fathom why an NCR had not been raised but put it down to the project engineers on-site having limited experience as they were recent graduates.Adding to the quality manager's reflexions, the engineering manager pointed out that project engineers on the job may not have been given the required supervision.Expanding on this point, it was noted that the project was subjected to resource constraints due to shortages in workforce supply, which contributed to the circumstances within which the errors occurred.
In this example, the project engineers were likely to have been under pressure and, when the error occurred, found themselves in a role conflict situation; that is, whether to ignore and ensure works were carried out or report the work as a non-conformance potentially delaying progress.There was a view that reporting the work as a non-conformance would have brought attention to them, with questions being raised about their suitability to do the job at hand.Thus, the reluctance of the project engineers to ask questions and seek out answers to problems masked their ignorance and inexperience.Yet, Alliance staff were encouraged to report non-conformances and openly discuss problems, as it was essential for learning and mitigating errors and rework.Lamenting their frustration with the project engineers onsite, particularly as an NCR had not been raised, an engineering manager made the following remark: An iRFI was subsequently raised along with photographs of the pile presented in Figure 3.As shown in Figure 3, the pile was oval rather than circular.The bolts were out of position, raising concerns about whether the steel structure forming part of the overhead wiring system could be installed on top of them.Here the pile footing acts as an anchor and is loaded to 35Kn.The concrete cover requirements for the pile's reinforcement cage were also too low.
It turned out that the formwork had been incorrectly setout, and its bracing had been wrongly applied.So, when the Self-compacting Concrete (SSC, with a slump flow of 650 mm) was poured into the pile's reinforcement cage, it began to leak outside the formwork.Recognizing the need to address the error quickly and minimize its adverse consequences (e.g.delaying works), the structural engineer responded to the iRFI the next day, recommending that an NCR be issued.
On reviewing the documentation provided, the structural engineer recommended no corrective action on the pile's shape even though it did not adhere to the IFC drawings.But, the modifications to the pile base plate (i.e.chamfering of corners) were required to enable the grout pad to sit on the top of the pile.

Learning from circumstantial errors
The pile example presented here raised concerns as it was a repeated error.The immediate response was to recommend additional considerations to be included in the pile's construction methodology to avoid future re-occurrence, which were: (1) access the location of the pile and seek to understand the constraints affecting the integrity of the formwork (e.g. level of the surface); (2) evaluate the amount of bracing required to ensure the integrity of the pile's shape; (3) apply adequate pressure to the bracing to minimize the deformation of the Formatube (i.e. is a wax-lined, cardboard tube specifically designed to form round concrete columns); and (4) use high workable concrete instead of the selected SSC as it was too difficult to handle.
Putting these technically-focussed recommendations aside, the error was repeated, but it had not been communicated effectively to site management and identified as a likely problem a priori.Increased emphasis has been placed on sharing NCRs between projects by sending alerts of events to site management staff.In addition, a knowledgebased engineering system (KBES) is being developed and used to assess rework risks (Matthews, Love, Porter, et al. 2022).However, addressing issues associated with acquiring experienced engineers and having the capacity to supervise all aspects of their work pose a challenge for the Alliance, considering the prevailing production pressure and overheated market. 3 Indeed, the project engineers had failed to report the error.The action of not reporting the error can be viewed as a violation.Setting aside the project engineers' inexperience, procedural drift (a product of adaptive behaviour and risk perception) occurred; a mismatch between required quality assurance procedures and actual work practice occurred (Dekker 2006).Undoubtedly, this mismatch primarily arose due to the conflicting priorities and demands that confronted the project engineers.Thus, given the circumstances, procedural drift most likely came about due to the absence of supervision rather than issues associated with prescriptive or badly crafted procedures and overconfidence.During morning pre-start meetings, quality-related topics were introduced and discussed to heighten the awareness of all likely rework-related risks.The active reporting of non-conformances was reinforced.
Additionally, supervisors would be required to be present during pour pre-checks to ensure there were no mistakes in future; otherwise, a concrete pour could not progress.This response somewhat contradicts Love et al.'s (2022c) Quality-II principle in aligning with Quality-I (see Table 1).Still, supervisory control is needed, at least in the short term, to compensate for the inexperience of project engineers and facilitate communication between parties.

Discussion
The narratives presented recognize 'human experiences as dynamic entities that are in a constant state of flux' (Wang and Geale 2015, 196).They provide means for understanding the dynamics and social organization of errors enabling the 'implicit explicit, the hidden seen, the unformed formed, and the confusing clear' (Wang and Geale 2015, 196).Thus, interviewees spoke freely and openly about their experiences and views without externally imposed constraints.As a result, the narratives illuminate the intricacy of error-making and rework, which have helped the researchers create meaning in a story form.Indeed, specific perspectives of a selected event are conveyed, emphasizing the meaning of the individual's experience and interpretations, enabling the 'truth of their experience, not an objective decontextualized truth' to be garnered (Bailey and Tilley 2002, 581).
In Supplementary Table A, examples of non-conformances due to hold-ups, circumstances, and mistakes can be seen.Violations were not explicitly recognized as contributors to a rework event.But considering the resourcing and time constraints that confront the Alliance and the tendency for people to routinely look for ways to make their work more efficient when subjected to such conditions, violations should not be ruled out or considered an aberration.
Unsafe acts are associated with violations and have been found to generally manifest when rework is being performed (Love, Teo, et al. 2018).No safety incidents were identified while a rework was performed, which is attributable to site management (e.g.supervisors) having increased awareness, due to BWA's experiences, of the risks people will take when subjected to resource and time constraints.Daily pre-start meetings reinforced the need to report an error immediately and implement a risk management plan to rectify the works.
Over the Alliance's seven years of operation, it has transitioned from a 'no blame culture' to establishing an 'error management culture' (Love et al. 2022b;Matthews, Love, Ika et al. 2022).The Alliance has made significant strides to containing 4 and reducing 5 errors and rework by unconsciously implementing organizational error management practices and adopting a Quality-II mindset. 6But as Love and Matthews (2022) have observed, the Alliance can do more to manage errors better and reduce rework.
Accordingly, Love and Matthews (2022) have suggested that the incorporation of resilience principles, such as awareness (i.e.acquisition of data about rework causes, consequences, and effects), preparedness (i.e.anticipating and preparing to cope with errors), flexibility (i.e.ability to respond to change and variability) and opacity (i.e.visibility of rework events) complement error management and can further support the Alliance's quest to mitigate its rework.While the Alliance recognized rework was a problem, its lack of awareness and preparedness has been the motivation for conducting this research.Garnering an understanding of the social organization of errors that result in the manifestation of rework provides a basis to raise awareness and communicate its dynamics enabling mechanisms to be identified and used to anticipate and manage their occurrence (Matthews, Love, Porter, et al. 2022).
Communication is critical to raising awareness and sharing knowledge and experiences, with information flow indicative of an 'organization's functioning' (Westrum 2014, 58).Yet, in projects, each organization differs in how it transfers, uses, and negotiates its information needs and requirements for decision-making.In the case of the architectural screens, the 'primary negotiation process' was overlooked, hindering the effectiveness of the IM process and contributing to hold-ups and mistakes (Riemer 1976, 265).Upon identifying the mistake, the parties involved sought to quickly mitigate its adverse consequences by engaging in a 'secondary process of negotiation' to ensure the work was rectified (Riemer 1976, 265).Indeed, almost all errors requiring rectification 'regardless of their social basis [e.g.hold-ups, circumstantial, and miscalculations] are managed through a re-negotiation' (Riemer 1976, 265).If the process of re-negotiation is not attended to promptly and handled effectively, then error cascades may materialize.The upshot, in this case, could be increased costs, delays, and adverse safety and environmental impacts.
Circumstantial errors, such as the non-compliant pile examples and those identified in Supplementary Table A are products of the client's demands and requirements and the subsequent contractual negotiations with the Alliance.Naturally, the Alliance's management would like to control the conditions of work and how they impact their project teams.While the Alliance cannot prevent the production pressure driven by political aspirations and skills shortages due to an overheated market, it has been able to embrace, adapt, and respond to the conditions and meet its KRAs (Love and Matthews 2022).
Of note, quality was not identified as KRA and linked to the 'gain-share, pain-share' regime.Thus, if quality is not given the same credence as safety and other KRAs (e.g.environment and community relations), tensions will arise, resulting in the orchestration of trade-offs in resources and time (Love et al. 2022a).This is not to say, the Alliance did not value quality; quite the contrary.In this case, the client is responsible for specifying quality as a KRA and linking it to the 'gain-share, pain-share' regime.However, if the Alliance is not measured on its quality, its importance will naturally be downplayed compared to other KRAs, such as safety.As can be seen in non-compliant piles, procedural drift was at play.If such behaviour is not addressed, it can become normalized and part of working conditions.What is more, a disconnect between what management thinks is happening on-site and what is actually going on can occur.Thus, the Alliance cannot afford such a misalignment to fester as it can erode its established culture, which it has worked tirelessly to develop and seeks to preserve and build upon in its future projects.

Theoretical implications
The theoretical implications of this research are two-fold.Firstly, while the research focussed solely on errors that led to rework, it nevertheless provides insights about how they manifested and were subsequently managed during construction when they could have been corrected before adverse outcomes emerged.The narratives produced for the architectural screen and non-compliant piles demonstrate that error management varies amongst error types (Zhao and Olivera 2006;Ingardi, Meyer, and Verdin 2021).For example, in the architectural screens' rectification case, the response was to provide agency and empower individuals to speak up and share their knowledge and experience about perceived error risks.
Contrastingly, the non-compliant pile was a repeat event, but site management was unaware that this was the case.However, moving forward and learning from this event, the Alliance strove to raise awareness of all closed-off NCRs by providing all project teams with information on why they occurred and were resolved.Relevant non-conformances were then discussed at daily pre-start meetings enabling appropriate risk mitigation to be put in place.Thus, in line with Ingardi, Meyer, and Verdin (2021), the research indicates that there are varying 'conventions through which, and the outcomes for which different types of errors' are communicated in practice (502).Consequently, several mediums are needed to communicate knowledge and experiences about the causes of errors and rework in projects to raise awareness amongst site management and their subcontractors to help them anticipate and plan for 'what might go wrong'.
And secondly, the research reveals that an understanding of context 7 matters when determining the whys and wherefores of errors and rework causation.The context provides the initial 'conditions' without which it might not be possible to adequately or fully understand how rework manifests.Yet previous lines of inquiry examining rework causation have tended to eschew the context of error-making and their treatment and instead adopted a 'reductionist' perspective focussing on determining proximal and root causes and proposing 'one-size fits all' type prevention strategies (e.g.Josephson, Larsson, and Li 2002;Ye et al. 2015;Asadi, Wilkinson, and Rotimi 2021;Garg and Misra 2021;Asadi, Rotimi, and Wilkinson 2022).
By making sense of the context surrounding error events, as presented in this paper, it becomes apparent that rework manifests from social disorganization, emerging from role ambiguity, misunderstandings, misinterpretations, and breakdowns in communications and interactions between people and organizations.Moreover, various error types evoke different responses suggesting that strategies to address rework must be tailor-made to the context in which they occur.

Practical implications
Despite the significant amount of research that has sought to avert rework in construction, it remains an innate and pervasive aspect of practice.However, its impact on project performance can vary considerably (Love, Ika, et al. 2019).This situation arises as there is limited understanding and knowledge about how human error, in its various guises, results in the need for rework.The narratives uncover the nuances of diverse error contexts and provide meaning to the dialogue and reflection conveyed about the rework events that unfolded.Drawing on the experiences of these events, construction organizations can take heed from them and accordingly learn.
So, what can construction organizations practically take away from the findings presented?Firstly, they need to acknowledge that their projects comprise dynamic processes characterized by the continuous interaction of human thought and behaviour that will continuously 'bump' into differing personal, social and organizational environments, rendering errors an inevitable feature of practice and failing 'negotiated orders'.Put simply, errors and rework come about due to social disorganization in projects.Accordingly, management must focus on enabling communicative action, collaboration, and building trust.Effectively addressing these tenets of practice is easier said than done-nurturing them in projects is a managerial challenge in environments that are not procured under the auspices of collaborative delivery strategies, especially when risks are not shared.However, when trust is present, people feel that they can speak up without punishment.If psychological safety promotes speaking up and being heard, then information flow can be used as an indicator of trust (Westrum 2014;Love, Smith, et al. 2019).
Secondly, construction organizations need to adopt a mindset where there is an acceptance to err.Indeed, such a mindset is counterintuitive but is necessary for effective learning.Thus, management must create an environment where reporting errors (e.g.non-conformances), for example, is encouraged.Rather than apportioning blame for people's (in)actions and judgements, construction organizations should seek to understand why their assessments and actions made sense at the time, given the circumstances surrounding them (Table 1).However, in this case, the challenge for construction organizations will be to identify, initiate and enact the change needed to ensure errors are not repeated.
And finally, construction organizations need to focus on raising awareness, communicating via various mediums, and engaging in shared-learning of rework events.However, organizations need to measure their rework before awareness can effectively occur.As a result, developing a dedicated system (e.g.KBES) to bring together rework data (i.e.not only from NCRs) is needed to enable them to identify risks (i.e.anticipate) and bolster their continuous improvement strategies.

Conclusion
To sum up, the research presented in this paper addressed the following question: Why do errors and rework occur, and how are they managed?Previous studies addressing the rework phenomena have ignored the role that errors play and the conditions within which it manifests.The result is that the efforts to remedy rework have been futile as the context contributing to its occurrence has gone unheeded.
Unintentional errors are artefacts of practice in construction.Indeed, there are both positive and negative consequences of error.Having to perform rework is a negative error consequence that can result in adverse project outcomes.However, the error's effect and learning depend on its management.This research draws on the experiences and insights of practitioners about specific rework events that occurred during the construction of a transport mega-project procured using an Alliance delivery strategy.The interviews revealed that the Alliance's projects were subjected to considerable production pressure and resource constraints due to a prevailing skills shortage.Additionally, the Alliance's ability to capture rework information was restricted to non-conformances events, which has been the focus of this research.
Interviewees repeatedly mentioned two non-conformances requiring rework: (1) the architectural screens; and (2) a noncompliant pile footing.Narratives were crafted and presented to understand the dynamics and social organization of errors.Thus, meaning in a story form enabled a truth from interviewees' experiences and insights to be represented.The narratives reveal rework manifests from failures in 'negotiated order' which stems from role ambiguity, misunderstandings, misinterpretations, and break-downs in communications and interactions between people and organizations.While these findings provide new insights to rework causation, we need to point out that our sample was only limited to 19 interviews, which restricted our ability to delve deeper and provide a richer account of events presented.
From a theoretical perspective, the need for understanding context and the dynamics of negotiated order have come to the fore in this paper.But this paper's exploration of the social organization of errors is just the 'tip of an iceberg' as further research is needed to develop a much-needed understanding and awareness of rework causation in construction.Similarly practical perspective, there is a need to accept 'errors happen' and encourage people to communicate and discuss those arising in projects to learn openly.For this to happen, an environment of psychological safety needs to be established, but this is not straightforward as it appears to be, as people will naturally avoid interpersonal risk-taking.Thus, construction organizations need to develop and build the capabilities to engender and augment psychological safety as part of a strategy to improve their management of errors and mitigate rework.
Notes 1. Error management can be described as a perspective that pledges for a 'useful approach to errors with the goal of reducing future errors, of avoiding negative error consequences and of dealing quickly with error consequences once they occur' (Frese 1995, 113).2. A negotiated order is the pattern of activities emerging over time as an outcome of the interplay of the variety of interests, understandings, reactions, and initiatives of the individuals and groups involved in an organisation [or project] (Strauss et al. 1963).Accordingly, Strauss (1978) asserts that 'the negotiated order on any given day could be conceived of as the sum total of the organisation's rules and policies, along with whatever agreements, understandings, pacts, contracts, and other working arrangements currently obtained.These include agreements at every level of organisation, of every clique and coalition, and include covert as well as overt agreements' (5-6).3.At the time this research undertaken the industry had shown signs it was overheating.A year on in April 2022 Australia several mega-projects were put on-hold due to an overheated industry.The Australian government announced it would delay several mega-projects such as the Beaches Link Highway and the Parramatta rail line (Winston 2022).4. Measures designed to enhance the detection and recovery of errors, as well as seeking to minimize their adverse consequences.5. Measures designed to limit the occurrence of errors.6. Organisational error management practices comprise (van Dyck et al. 2005(van Dyck et al. , 1230)): (1) communicating about errors; (2) sharing error knowledge; (3) helping in error situations; (4) quick error detection and damage control; (5) analyzing errors; (6) coordinating error handling; and (7) effective error handling.7. Context is 'defined as the formal and informal setting in which a situation occurs', which includes its economic, political, legal, geographical, historical, socio-cultural, environmental, institutional or managerial circumstances and/or the continual unfolding interaction between the situation and the setting (Brown 2010, 7).

Figure 2 .
Figure 2. The process used to create the rework narrative.

Table 1 .
From Quality-I to Quality-II.