Opportunity, motivation and ability to learn from failures and errors: Review, synthesis, and the way forward.



 
 

 
 
 
 

 
 
 
 
 
 
 
 
 
 
 

 

 

 
 

 

 

King’s Research Portal 
 

DOI:
10.5465/annals.2016.0049

Document Version
Peer reviewed version

Link to publication record in King's Research Portal

Citation for published version (APA):
Dahlin, K. B., Chuang, Y., & Roulet, T. J. (2018). Opportunity, motivation and ability to learn from failures and
errors: Review, synthesis, and the way forward. Academy of management annals, 12(1), 252-277.
https://doi.org/10.5465/annals.2016.0049

Citing this paper
Please note that where the full-text provided on King's Research Portal is the Author Accepted Manuscript or Post-Print version this may
differ from the final Published version. If citing, it is advised that you check and use the publisher's definitive version for pagination,
volume/issue, and date of publication details. And where the final published version is provided on the Research Portal, if citing you are
again advised to check the publisher's website for any subsequent corrections.

General rights
Copyright and moral rights for the publications made accessible in the Research Portal are retained by the authors and/or other copyright
owners and it is a condition of accessing publications that users recognize and abide by the legal requirements associated with these rights.

•Users may download and print one copy of any publication from the Research Portal for the purpose of private study or research.
•You may not further distribute the material or use it for any profit-making activity or commercial gain
•You may freely distribute the URL identifying the publication in the Research Portal

Take down policy
If you believe that this document breaches copyright please contact librarypure@kcl.ac.uk providing details, and we will remove access to
the work immediately and investigate your claim.

Download date: 06. Apr. 2021

https://doi.org/10.5465/annals.2016.0049
https://kclpure.kcl.ac.uk/portal/en/publications/opportunity-motivation-and-ability-to-learn-from-failures-and-errors(efe80078-ea36-4a8f-95c0-e2d8ae4877d4).html
https://doi.org/10.5465/annals.2016.0049


 

 

 

 

 

 

Opportunity, motivation and ability to learn from failures 

and errors: Review, synthesis, and the way forward.  
 

 

Journal: Academy of Management Annals 

Manuscript ID ANNALS-2016-0049.R4 

Document Type: Article 

Keywords: 
Organizational < COGNITION, Team < COGNITION, Organizational < 

LEARNING, Group < LEARNING, Individual < LEARNING 

  

 

 

Academy of Management Annals



 
1

Opportunity, motivation and ability to learn from failures and errors: 

Review, synthesis and ways to move forward 

 

Kristina B. Dahlin*, You-Ta Chuang** & Thomas J. Roulet*** 

* University of Oxford ** York University *** King’s College London 

 

Acknowledgment 

The authors are grateful for valuable comments and support from Elizabeth George, the 

editor, as well as from two reviewers. The authors have also benefited from discussions 

with Joelle Evans, Kristin Bondesson, Linda Argote, Matt Cronin, Ned Welch, Shitij Kapur 

and any health care worker we encountered while working on the manuscript. 

 

Abstract 

While organizations and individuals tend to focus on learning from success, research has 

shown that failure can yield crucial insights in various contexts that range from small 

mistakes and errors, product recalls, accidents, and medical errors, to large-scale disasters. 

This review of the literature identifies three mechanisms—opportunity, motivation, and 

ability—through which individuals, groups and organizations learn from failure, and it 

bridges the gaps between different levels of analysis. Opportunity to learn from failure mostly 

takes the shape of more information about errors and failures that are generated by one’s own 

and others’ prior failures or near-failures. Motivation to learn from failure is hindered by 

punitive leaders and organizations. Finally, ability to learn from failure partly relies on 

inherent attitudes and characteristics; but can be further developed through thoughtful 

analysis and transfers of successful routines. Our review leads us to distinguish between 

erroneous versus correct processes and adverse versus successful outcomes to better 

understand the full gamut of events that are faced by organizations. We identify the existence 

of noisy learning environment, where spurious successes (when erroneous processes still lead 

Page 1 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
2

to successful outcomes) and spurious failures (when correct processes are combined with 

adverse outcomes) lower the opportunity to learn. Considering noisy learning situations is 

helpful when understanding the differences between slow- and fast-learning environments. 

We conclude our review by identifying a number of unexplored areas we hope scholars will 

address to better our understanding of failure learning. 

INTRODUCTION 

 

Sometimes we may learn more from a man's errors, than from his virtues.  

Henry Wadsworth Longfellow 

 

Individuals and organizations repeatedly confront failures that range from small technical 

errors and mistakes to product breakdowns to large-scale disasters. Failure can stigmatize 

individual and organizational reputations, and it can be extremely costly for organizations and 

society. Failure is also more noticeable than success because negative information is more 

salient than positive information (Ito, Larsen, Smith & Cacioppo, 1998). As a consequence, 

individuals and organizations strongly prefer success, which makes learning from failure 

difficult because both the reporting of errors and other failures as well as the correct analysis 

and response are risky and emotionally fraught. However, learning from failures is critical for 

both operational performance and safety—failure learning is necessary for quality 

improvements and efficiency gains in production processes, and systematic failure reporting 

and analyses have been key for reductions in transportation accidents and adverse events in 

hospitals. Because of failure’s significance, the research on the topic spans many fields such 

as psychology (cf. Ellis & Davidi, 2005; Hofmann & Mark, 2006), organizational studies (cf. 

Reason, 1997; Zhao & Olivera, 2006), strategic management (cf. Muehlfeld, Sahib, & van 

Witteloostuijn, 2012), sociology (cf. Perrow, 1999), and health care management (cf. Hoff, 

Jameson, Hannan, & Flink, 2004; Kohn, Corrigan, Donaldson, 1999). Importantly, failures 

provide valuable learning opportunities: individuals and organizations modify their practices 

Page 2 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
3

to prevent similar future failures and to improve performance (Sitkin, 1992). Without 

examining failure learning, our understanding of success learning is also inherently biased 

(Baum & Dahlin, 2007; Denrell, 2003). Recognizing the potential of failure to improve 

performance, recent studies on failure have begun to shift their focus from why and how 

failure occurs in organizations to how individuals and organizations do (or do not) learn from 

failure. 

Failure learning has become a distinct area in the organizational learning literature, and 

it has attracted growing attention from scholars who seek to understand the phenomenon in 

various contexts such as product recalls (Haunschild & Rhee, 2004), project failure 

(Shepherd, Patzelt, & Wolfe, 2011), bankruptcies (Kim & Miner, 2007), healthcare errors and 

incidents (Chuang et al. 2007; Vogus & Sutcliffe, 2007), and accidents (Baum & Dahlin, 

2007; Haunschild & Sullivan, 2002). The studies on failure learning cover multiple levels of 

analysis and draw from a variety of theoretical frames to understand how actors do (or do not) 

learn from failure. While an array of factors that affect failure learning has been identified, 

there is a lack of systematic integration across levels of analysis and settings; hence, the 

collective wisdom about how to best learn from failure is limited and fragmented. 

Importantly, the studies on failure learning borrow much from traditional learning studies, but 

the links and dissimilarities are not clearly understood. Studies that have attempted to 

combine success and failure learning lack conclusive findings and theory exploring how these 

events are related. To that end, there is a greater divide between traditional learning studies 

and failure learning studies than is currently being theorized.  

We conduct a review of failure learning studies and synthesize them to better 

understand the underlying mechanisms that influence failure learning. We apply the 

framework of opportunity, motivation, and ability to integrate and discuss learning factors at 

the individual, group, and organizational levels. Specifically, opportunity represents a 

Page 3 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
4

mechanism that provides information or sufficient time to analyze the cause-effect of failures; 

motivation captures different actors’ willingness to act on failure information and to engage in 

failure learning activity; and ability represents actors’ skills or knowledge base to change their 

actions based on failure information (Argote, 2012; Reinholt, Pedersen & Foss, 2011). 

Mapping the studies of failure learning at multiple levels of analysis with this framework can 

generate new insights for future work. 

We argue that it is important to clearly separate between processes and outcomes and to 

acknowledge that bad processes do not always lead to failed outcomes and, conversely, that 

correct processes might still result in failed outcomes. A traditional focus on “successful 

processes—successful outcomes” versus “failed processes—failed outcomes” overlooks the 

fact that “successful processes—failed outcomes” and “failed processes—successful 

outcomes” are common. We call these last two process-outcome combinations spurious 

failures and spurious successes. When they are common, these combinations produce noise in 

traditional learning processes that negatively affects opportunity, motivation, and ability to 

learn. A complete understanding of process-outcome relationships in organizational learning 

helps to address a fundamental critical question of why we see a systematic decrease in 

failures in one setting, whereas in another we do not. While the risk of dying in a car accident 

has diminished by 50% over the last 25 years, and the risk of train accidents has been reduced 

by 70% (NCSA, 2015; FRA, 2016), the risk of dying from a hospital error has increased by 

350%
1
 (Binder, 2013; Kohn, Corrigan & Donaldson, 2000). 

Below, we begin our review by defining failures and failure learning as well as errors 

and error learning, followed by a review of the failure and error learning literatures at three 

levels of analysis, clustering mechanisms under the opportunity-motivation-ability headings. 

We highlight the key constructs and mechanisms that can be identified as influencing learning 

                                                        
1
 Part of the increase is due to new definitions and better measurement of preventable hospital errors. The 

numbers are also challenged; however, overall it is clear that there is little to no improvement, which still 

demonstrates the contrast with transportation accidents. 

Page 4 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
5

processes and outcomes. From there, we discuss how the four process-outcome combinations 

influence the opportunity, motivation, and ability that are associated with failure learning. 

Finally, synthesizing and assessing the literature as a whole, we identify the research 

challenges in failure learning and discuss the promising research opportunities that may 

advance our understanding of failure learning. 

 

DEFINING SCOPE: FAILURE, ERRORS, AND FAILURE LEARNING  

There is a literature on errors and another literature on failures in organizations. They are 

related, and many of the mechanisms and findings overlap. In fact, they often use the same  

definition for errors and failures, that they “deviate from expected and desired goals” 

(Rasmussen, 1982; Reason, 1990; Leape, 1974; Zhao & Olivera, 2003; Sitkin, 1992). The 

literatures do also differ since errors are incorrectly executed tasks or routines (such as a train 

engineer who drives a train over the speed limit or a nurse who gives the incorrect medication 

to a patient), while failures are undesired performance outcomes (a train accident occurring 

instead of the train getting from point A to point B as planned; a patient who dies after 

surgery instead of leaving the hospital healthier than before entering it). 

With regard to errors, they have been classified by Rasmussen (1982) into rule-based 

errors (breaking a known rule), skill-based errors (making a mistake or forgetting) and 

knowledge-based errors (not knowing enough). Another error typology is whether the error is 

action-based or related to decision-making (Lei, Naveh, & Novikov, 2016; Zhao & Olivera, 

2006). Nevertheless, not all errors, mistakes or incidents necessarily lead to failure. Some 

errors and mistakes can even produce positive outcomes, such as the discovery of new 

organizational processes and innovation, or be too insignificant have any impact on an event’s 

eventual success or failure (Cannon & Edmondson, 2005). 

Page 5 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
6

Failures may be caused by a combination of errors such as incorrectly executed routines 

and tasks, violations, risks, or chance factors (Frese & Keith, 2015; Hofmann & Frese, 2011). 

It can be avoidable or unavoidable, intentional or unintentional. It can involve human action 

and organizational processes and arrangements (e.g., Ramanujam & Goodman, 2003; Reason, 

1997).  

We define error and failure learning as the process by which individuals, groups, or 

organizations identify error or failure events, analyze such events to find their causes, and 

search for and implement solutions to prevent similar errors or failures in the future. This 

definition is consistent with the definitions of learning in the organizational learning literature 

(Argote, 2012). The outcomes of error and failure learning can therefore include changes in 

understanding (Huber, 1991), behaviors (Chuang & Baum, 2003; Ginsburg et al., 2009; 

Shepherd, Patzelt, & Wolfe, 2011) or performance improvement (Cannon & Edmondson, 

2001; Dahlin & Baum, 2007; Heimbeck, et al., 2003; Zhao, 2011). 

We aimed to be inclusive, if not exhaustive, in identifying studies on error and failure 

learning and searched key management, health management and safety journals for relevant 

studies, focusing mostly but not exclusively on the period from year 2000. All identified 

articles were sorted by their level(s) of analysis. We reviewed the learning mechanisms in the 

articles and categorized them according to failure learning triggers, clustering them under 

three headings: opportunity to learn, motivation to learn, and ability to learn.  

 

OPPORTUNITY TO LEARN FROM FAILURE 

Opportunity to learn refers to the scope of information and the time that allows actors to 

learn from failure events. Information-based opportunity refers to the amount of information 

that is available about similar failure events because such events can provide information 

about failure causes (Argote, 2012). Time-based opportunity refers to the time that is given to 

Page 6 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
7

actors to reflect upon failure events and to analyze the information that can be derived from 

the events to learn from them and the time in which to execute an action that is related to a 

failure learning activity (cf. Carroll, 1963). Information-based opportunities are usually 

studied by quantifying the amount of available information about similar failure events 

(number, frequency, recency), information access owing to group composition, organizational 

members’ networks and information diffusion inside or between organizations. By contrast, 

time-based opportunity refers to how much time that is available to  process information 

and/or carrying out a task (Carroll, 1963).  

Information-based learning opportunities are often measured as one’s own or others’ 

prior experiences with similar failure events (counted as the of number of events or the 

number of cumulative events in previous time periods) and as how organizational structures 

and routines influence the diffusion of failure-learning-related information. Experience and its 

effect on learning are the most common empirical approach, and while this was first studied 

in production settings that range from classic cases of airplane construction to that of liberty 

ship builders (Wright 1936; Thompson, 2001) and learning curves, which refers to the idea 

that cumulative experience affects performance at a decreasing rate, the experience and its 

effect on learning have come to be applied across levels of analysis, and used across a large 

number of settings—among them failure learning.  

Two different processes convert experiences into better performance: learning-by-doing 

and analytical learning. Learning-by-doing is mostly automatic and tacit, while analytical 

learning involves active decision-making that uses information about a prior event to reshape 

future routines (Thomson, 2012; Reason, 1990). Failure learning theories are much more 

concerned with the active decision-making theory, especially on the individual and group 

levels, while few organizational learning curve studies separate empirical and theoretically 

Page 7 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
8

between learning by doing and analytical learning (see Lapré, Mukherjee & Van Wassenhove, 

2000; or Sinclair, Klepper & Cohen, 2000 for exceptions).
2
   

Time-based, or temporal, learning opportunities are concerned with the amount of time 

that an actor has to execute a routine or to process information about a routine that has gone 

wrong. Outside of the realm of student learning, relatively few studies have examined 

temporal mechanisms. Exceptions in failure learning a small number of studies that analyze 

the impact of workload and slack (Malone et al. 2007; Lawton, et al. 2012) and how the speed 

of assimilating and analyzing relevant information affects work processes and routines 

(Edmondson, et al. 2001). In addition, autonomy provides the mental and operational space 

that can allow individuals to prioritize their tasks, allowing them the time that they need to 

learn from failure (Kerr, 2009; Stern et al., 2008).  

 

Individual and group level opportunities to learn from failure 

A wide range of studies have examined how information-based and time-based opportunities 

lead to failure learning. While some learning opportunities lead to automatic reduction of 

errors and failures, this concerns mainly errors such as slips and mistakes. In contrast 

knowledge-based errors and failures require more deliberate reflection to reduce the 

likelihood of repeating them (Iedema et al. 2006). A study examining the effect cardiac 

surgeons’ prior experience had on learning, demonstrated that past surgery failures improved 

a surgeon’s future surgery outcomes (KC, Staats, & Gino, 2013). Moreover, other cardiac 

surgeons’ failures interacted with the surgeon’s own prior failures to further improve the 

surgeon’s surgery outcomes. In other words, one’s own and others’ failure experiences can 

have a joint effect on learning: any related failure provides an opportunity to reflect on what 

                                                        
2
  

Page 8 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
9

has gone wrong and how to improve procedures. Interestingly, the positive interaction effect 

suggest that information-based experience can have an increasing return on failure learning.  

        Temporal opportunities such as working conditions affect how experiences are converted 

into lower error rates. Residents with greater work autonomy, that is, they “perceive that they 

have the freedom and discretion to plan, schedule and carry out their jobs” were found to have 

higher error reduction (Stern et al., 2008, p. 1554). Task autonomy allowed the residents to 

reflect on errors and learn from them by making procedural changes. When the organizational 

climate was such that learning was encouraged, the residents took more time to reflect on 

work processes and errors and further reduced their own errors, thus, motivation (climate) and 

opportunity (time) interacted to accelerate learning (ibid.). 

Teams play an important role in error learning (Edmondson et al. 2001); however, when 

the environment is ambiguous and changing, team information processing becomes 

complicated, which hampers learning. Individual and group levels of learning are intertwined 

as individual characteristics and team composition jointly determine teams’ error reporting 

(Edmondson, 1996). Good member coordination (Baker et al. 2006) and common goals 

(Tjosvold, et al. 2004) enable teams to benefit from the full potential of each of their 

members. In the same way, team stability and work processes enable the group to better 

process information (Edmondson, 1996). When error information gathering activity is not a 

part of existing team routines (Lawton, et al. 2012), or the team does not have enough 

autonomy to allow it to collect critical information (Kerr, 2009), learning might not occur. In 

other words, having neither established nor improvised ways to gather failure information 

reduces opportunities to learn from failure experiences.  

Group characteristics can also influence information-based learning. Group diversity 

and intergroup linkages are two mechanisms that provide teams with access to a wide range 

of information which, in turn, reduce failure rates (Chuang, et al., 2007; Tucker & 

Page 9 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
10

Edmondson, 2003). Member rotation is another way to expose teams to a greater flow of 

information that can enable them to better analyze problems (Argote & Todorova, 2007). In 

addition, the time aspect matters here—being exposed to many different but related 

experiences in a short amount of time benefits learning. Experience also transfers across 

levels: group-level success experience has been found to help individual-level failure 

learning, which ultimately benefits the learning rate of the organization as a whole (Zheng et 

al., 2013).  

 

Organization-level opportunity to learn from failure 

In our review of the literature, it emerged that information-based opportunity is the most 

studied organizational learning mechanism. To summarize across the organizational 

opportunity studies: experience matters in most cases and across a wide array of settings; 

however, it does so somewhat differently in different settings and organizations. The nature of 

an experience event, its outcome, rareness, and complexity, influence its learning impact. A 

trend in organizational-level studies is to separate one’s own versus others’ experiences and to 

focus on in which case one matters more than the other and when (Baum & Dahlin, 2007; 

Chuang & Baum, 2003; Kim & Miner, 2007; Madsen & Desai, 2010; Madsen, Dillon & 

Tinsley, 2016). 

Prior failure events provide opportunities to learn. Over time, as many problems are 

resolved or better managed, we would expect a diminishing return to experience, which yields 

a learning curve that is similar to that in production learning. In fact, there is rich empirical 

support that shows that learning curves have a similar shape with regard to failure reduction. 

Train, mining, and airline accidents have been found to reduce future accident propensity on 

the industry and firm levels (Baum & Dahlin, 2007; Desai, 2016; Madsen, 2009; Haunschild 

& Rhee, 2004). The failure-reducing effect is strongest for recent accidents (Haunschild, 

Page 10 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
11

Polidoro & Chandler, 2015), accidents of larger magnitude, which are measured in terms of 

accident cost and level of injuries (Madsen, 2009), and events of high visibility (Desai, 2011), 

which are measured in terms of the media scrutiny of the accidents. There are two 

mechanisms that can induce organizations to learn from highly visible failures: in such cases, 

the information about causes and remedies is more available; however, the motivation to 

counteract accidents may also increase when outsiders pay more attention and reputations are 

at stake. When there is more press attention to an accident or a product recall, organizations 

invest more in activities that can reduce the risk of future accidents, such as a train line 

installing new track (Desai, 2011). 

The effect of recency, that more recent events have greater impact than older events on 

failure reduction, can also have multiple explanations. For instance, there is the proposition 

that new routines and practices make past experiences less relevant as time passes, or, as 

suggested by Haunschild et al. (2015), that an adverse event attracts an organization’s 

attention and motivates the organization to reduce the risk of future accidents; however, this 

motivation weakens over time as other important organizational goals take precedence. 

          Complex problems provide a greater opportunity to learn. A notable study supporting 

the argument that complex challenges trigger faster and more efficient learning is an 

investigation of British IVF clinics, which demonstrates how opportunity affects learning 

(Stan & Vermeulen, 2013). The key performance metric of IVF clinics is live births per 

treatment cycle, and a key aim is to lower the number of failed cycles. Whereas private clinics 

could choose to accept only patients with good prognoses, which meant fewer failures, public 

clinics were not allowed to screen patients and therefore had higher failure rates at the start of 

their activities. However, working with more difficult cases enhanced the information-based 

learning of the public clinics, and they increased their ability to successfully treat any patient, 

which resulted in higher learning rates than those of private clinics. Working on complex 

Page 11 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
12

problems provided greater opportunity to find solutions to difficult problems and led to faster 

failure reduction across problems. The concept that complex failure situations provide richer 

information is in line with the finding that airlines learned more from complex accident 

causes than from simpler ones (Haunschild and Sullivan, 2002) and that this was especially 

true for specialist airlines. Complex problems offered more venues for learning and 

counteracted simplified cause-effect analyses (e.g., simplifying failure attributions to factors 

that are beyond the organization’s control such as patient age in hospital cases or to pilot error 

in the case of airlines). 

          Whose experiences matter. Not only do failures that are experienced by an individual 

organization provide opportunities to learn, but the failures that are experienced by other 

organizations provide information for learning. Airlines learn both from their own and other 

airlines’ accidents (Haunschild & Sullivan, 2002). Based on the behavioral theory of firm 

logic, which states that organizations are more likely to look further for better solutions when 

their peers outperform them, train companies that had more accidents compared to their peers 

were found to learn more from such peers rather than from their own accidents (Baum & 

Dahlin, 2007). Ontario nursing home chains learned from both their own and others' 

businesses with respect to naming their units. The learning effect, or willingness to change 

was dampened when the chain had followed a strategy for a long time (Chuang & Baum, 

2003).  

         Learning from near misses. It is not only failure events that provide information-based 

opportunities to learn, but such information can also be gleaned from events that are neither 

purely successful nor wholly failures (Rerup, 2006). Most attention has been given to near 

misses—when there is almost a failure but there are no direct negative consequences 

(Kessels-Habraken, et al. 2010). The opportunity to learn from near misses (especially those 

of others), or what is also known as latent errors (Reason, 1997; Ramanujam, 2003), is only 

Page 12 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
13

possible if such actions are recorded or easily observed. Kim and Miner (2007) used the 

ratings of banks to capture whether they were close to failing and found that near failures 

affected learning more than actual failures did. Building on industry insiders’ quotes, they 

found that near failures have greater informational content—whereas failed firms disappear 

from the industry and are forgotten, near failures remain and can tell their story and at the 

same time they also remind others of their survival. Perhaps more importantly, near failures 

combine bad performance with remedies for restoring performance after a period of trouble. 

Rather than telling a story of how to fail, they tell a story of redemption when one is close to 

failure. The greater informational content of the near failures means that there is a greater 

opportunity to learn from them. However, the reporting of near-misses and thus the 

information gathering process is more difficult because of the challenges in defining the scope 

of such events (Kessels-Habraken, et al. 2010). 

 

MOTIVATION TO LEARN FROM FAILURE 

Motivation is the desire or willingness to act in a certain way and, in the context of error and 

failure learning, the desire or willingness to invest in reducing adverse event frequency. 

Motivation to learn from failure therefore refers to the resource levels that are devoted by 

individuals and organizations to failure learning activity; such resources include attention and 

operational investments (Kanfer and Ackerman, 1989). Effective learning processes require 

individuals and organizations to allocate cognitive resources to (i) correctly identify and 

analyze error and failure causes and (ii) search for and implement solutions that prevent 

similar errors or failures in the future. Failure learning studies concerning motivation address 

contextual factors such as safety climate, psychological safety, leadership style and attitudes; 

cognitive and emotional barriers such as attribution errors (internal versus external); 

information processing in groups and, for organizational-level studies, factors that trigger 

Page 13 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
14

motivation such as high visibility events and public attention. Other organization-level 

motivation triggers include threats to ones’ reputation, social comparisons (doing worse than 

the competition), the recency of an event, and climate variables.  

 

Individual and group level motivation to learn from failure. 

Most individual-level studies on error and failure learning study motivational factors. In two 

laboratory studies, using a computer simulation task, Zhao (2011) found a positive association 

between participants’ self-reported motivation to learn from errors and their actual failure 

learning. In a field study of an Australian hospital, employees’ safety motivation had a 

positive effect on their safety participation, which was measured as behaviors that promote a 

safety-oriented environment (Neal & Griffin, 2006). While individual motivation affects 

learning behavior in both the lab and inside organizations, organizational factors affect 

individual motivation to learn from failure. For example, safety climate, which is defined as 

“perceptions of policies, procedures, and practices related to safety” (Neal & Griffin, 2006; 

956), plays an important role in motivating individual learning. Group-level safety promotion 

is positively associated with individual safety motivation and safety participation in hospitals 

(Buljac-Samardzic et al., 2012). There are both bottom-up and top-down processes between 

individuals and groups that help to produce a safety climate (Neal & Griffin, 2006). Group 

climate has been found to affect both individual motivation and individual behavior, and in a 

study of 33 organizations, individuals were more likely to report accidents when supervisors 

enforced safety policies (Probst, 2015). Adding the organizational level of analysis, individual 

attitudes to safety affects organizational learning by promoting a safety climate at the group 

level (Zohar & Luria, 2005). Individual safety motivation and engagement in failure learning 

activities both affect and are affected by group, managerial and organizational attitudes.  

Page 14 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
15

Motivation to learn from failure is also driven by individuals’ causal attributions 

concerning failures. In attribution theory, individuals cope with outcomes by making causal 

attributions to guide their future behaviors (Ilgen and Davis, 2000; Nisbett and Ross, 1980). 

Causal attribution involves how an individual allocates causes for a certain outcome, which, 

in turn, influence his or her motivation to learn. Specifically, individuals tend to attribute 

success to internal causes such as ability and effort, and failure to external causes such as 

environmental factors and luck (cf. Jones & Harris, 1967). At the same time, individuals often 

turn this around and attribute others’ successes to external causes and others’ failures to 

internal causes. Such attributions affect individual motivation to engage in failure learning 

activity (Chuang et al., 2007; Zhao and Olivera, 2006). Attribution theory helps us to better 

understand the previously discussed result that surgeons’ cumulative number of successful 

cardiac procedures had a greater impact on their future success rate than failed procedures 

had, while other surgeons’ cumulative number of failed procedures significantly helped to 

improve a cardiac surgeon’s subsequent success rate (KC et al., 2013). One can argue that 

surgeons attributed the prior successes to their own effort and actions, while the prior failures 

were attributed to uncontrollable factors. Thus, the surgeons assumed that there was little to 

learn from own failed procedures. In contrast, other surgeons’ failures were attributed to their 

efforts and actions, which made the focal surgeon more willing to review and learn from 

others’ failure causes. In a similar vein, individuals who work in teams attributed failure to 

actions that are taken by other individual team members rather than by the team when they 

experience poor team performance in a laboratory setting (Naquin and Tynan, 2003). 

However, this tendency decreased as subjects’ knowledge of teamwork increased. Another 

sign that attribution biases can be alleviated is a study in which individuals were better able to 

make internal attributions after failing in a task when they received after-event reviews that 

helped them to understand what contributed to their performance (Ellis, Mendel, and Nir, 

Page 15 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
16

2006). Individuals who made more internal than external attributions improved their 

performance more. In a study of CEO attributions and organizational performance, Salancik 

and Meindl (1984) found an even stronger learning effect from internal attributions: when 

CEOs attributed poor firm performance to internal causes even when the low performance 

was clearly caused by an external factor, their firms performed better in subsequent periods. 

Internal attributions are clearly important for behavioral change. 

Failures often generate strong negative emotions (Paget, 1988) as individuals 

experience feelings of guilt, embarrassment or fear when they are involved in failures or make 

errors (Edmondson, 1996; Paget, 1988). Negative emotions prompt individuals to become 

more risk averse (Loewenstein et al., 2001), affect judgment (Forgas, 1995), and lower 

engagement in failure learning. In general, failure-induced negative emotions should lower 

individuals’ motivation to learn. However, in a laboratory study, negative emotions were 

instead positively associated with the motivation to learn (Zhao, 2011). The author explains 

this effect by suggesting that the strength of emotions matters, not just whether an emotion is 

negative, and negative emotion only affects learning above a relatively high threshold. If the 

strength of emotion matters, the finding that managerial error intolerance increases negative 

emotions suggests that managers’ attitudes can push staff into non-learning (Keith & Frese, 

2005).  

There are individual differences in the emotional response to failure and individual 

differences in coping orientations that help to explain the different emotions that are 

generated by failure as well as differences in the motivation to learn from failure: Individual 

affective organizational commitment—how loyal they feel towards their organizations—

decreased with negative emotions about failure (Shepherd et al., 2011); however, this link 

between negative emotions and failure was moderated by coping orientations when dealing 

Page 16 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
17

with failure. Negative emotions also decrease when individuals perceive that failure is a 

normal occurrence in their work environment (Shepherd et al. 2011). 

Individuals’ motivation to learn from failure is affected by psychological safety. Not to 

be confused with safety climate, psychological safety refers to the perception that it is safe to 

take interpersonal and professional risks in the workplace (Edmonson, 1999; Edmonson and 

Lei, 2014). Psychological safety has been found to positively influence failure learning by 

increasing an individual’s motivation to engage in failure learning activity (reporting failures 

and errors; willingness to discuss possible solutions) because the fear of negative 

consequences to self-image, status, or career are lower when psychological safety is high 

(Edmondson, 1996; Tjosvold, et al. 2004). Surveying three firms in the software, electronics 

and finance industries, Carmeli and Gittell (2009) found that psychological safety was 

positively associated with failure-based learning behaviors. 

At the group level, group norms shape learning from failure: compliance with norms 

motivates individuals to identify and record failures and to act to prevent them (e.g., Katz-

Navon et al., 2009; Vogus & Sutcliffe, 2007; Zohar, 2002). When studying groups in different 

functions in the same organization, sharing the same beliefs about how to cope with failure 

exerted great influence on failure learning (Cannon & Edmondson, 2001).  

Information sharing and interpersonal relationships within groups is another key 

element impacting how motivated teams’ are to learn from failures. Teams with cooperative 

rather than competitive goals learn more (Tjosvold, et al. 2004). Trust between the members 

of a team affects learning motivation (Carmeli et al. 2012). Comparing hospitals with similar 

characteristics but different failure learning rates, Edmondson et al., (2001) found that team 

member error information sharing affected the learning rate. In a similar fashion, managerial 

safety practices also affect the motivation to learn—managers who demonstrate that safety is 

important positively affect error reduction (Katz-Navon et al. 2009). Lack of information 

Page 17 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
18

sharing depends on the routines, awareness of others’ knowledge and status dynamics within 

groups: a study found that slower-learning groups had team members that did not bring 

attention to errors since they assumed, often incorrectly, that other team members are already 

aware of them (Cannon & Edmondson, 2001; Edmondson, et al. 2001).  

According to Gersick & Hackman (1990), routines are double-edged swords: on the one 

hand, they can prevent team members from addressing failure by reducing their motivation to 

learn because change would upset the routine; however, on the other hand, routines create 

comfort within a group and lead to a climate where group members are more comfortable 

sharing what went wrong.  

A high team workload not only decreases the opportunity to learn, it also considerably 

reduces team motivation to learn from failure (Lawton et al. 2012), and it may ultimately 

affect the ability to change because the team might be unaware of the need to improve 

existing routines (Edmonton, et al. 2001). Team’s ability to manage its workload is also 

important, and teams that do better at his were found to improve their error rate faster 

(Lawton, et al. 2012). To reduce errors and failures routines often need replacing, but 

renewing routines is cumbersome, with preexisting routines being obstacles to change; thus, 

unless there is a clear rationale for why routines must change, groups are likely to resist 

change and may even implement defensive strategies with respect to learning processes, for 

example, by shifting responsibility and finding arguments to defend failing mechanisms 

(Hodgkinson & Wright, 2002). Workload is a time-dependent opportunity but also impacts 

motivation to learn since decisions that are made under time pressure may require individuals 

and groups to focus on a shorter-term horizon (Malone et al. 2007).  

 

Organization-level motivation to learn from failure.  

Page 18 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
19

A majority of the studies on how motivation affects error and failure are conducted in 

the health care context. There is a very real concern that hospitals are not reducing their error 

rates and are causing patient harm at a high rate (Leape, 1994; Kohn et al., 2000). The focus is 

often on how individual motivation to report errors is affected by organizational factors. The 

problem of non-reporting is high (Zhaou & Olivera, 2006; Ramanujam, 2003), which renders 

the analysis of error causes incomplete. The reason for an individual to not report an error that 

he or she made, or which was made by someone close, is often a lack of trust that responsible 

managers in the organization will conduct proper analyses to determine the true causes. It is a 

common and simple solution for organizations to blame an error on an individual (Hofmann 

& Stetzer, 1998; Rathert & May, 2007). The belief that organizations prefer to blame 

individuals rather than to find organizational causes is supported by data: Perrow (1999) 

reported that a re-examination of industrial accident causes where 70% to 80% were 

attributed to operator errors, in contrast found that operators had caused less than half of the 

accidents they were blamed for, or that 30% to 40% of accidents were caused by operator 

errors. 

In line with the tendency to blame the person who is closest to an error, factors that 

impede error reporting and analysis at the organizational level can be linked to the culture of 

blaming individuals rather than exploring other error causes (Khatri et al., 2009), ward 

climate (Lawton et al. 2012), whether health care work units are patient-centered (work is 

planned around the patient’s needs) (Rathert & May, 2007) and that when greater distance 

between professional groups in health care reduces the willingness of orderlies and nurses to 

admit to or discuss errors with doctors (Khatri et al., 2009). Organizational climate is a 

determinant that trickles down to affect both group and individual-level learning from failure. 

Motivational factors that have been studied outside health care are different. Based on 

the behavioral theory of firms, which states that organizations are more motivated to act and 

Page 19 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
20

change when they perform worse than they did in the past or when they perform worse than 

their competitors or peers (Cyert & March, 1963), Baum and Dahlin (2007) found that train 

companies that have higher accident costs than their peers learn faster but mainly from 

others’ experiences. Learning from others’ and not one’s own experiences might be due to 

poor failure performance: the organization might need external ideas of how to change 

behaviors. Another motivation study found that when an automobile manufacturer were 

forced to recall products and thus its reputation was challenged, the likelihood of future 

recalls was lowered (Haunschild & Rhee, 2004; Rhee, 2009). The effect in the auto recall case 

was curvilinear: low- and high-reputation car companies reduced recalls less than companies 

with medium reputations did. In a study that focused on how motivations change over time, 

Haunschild et al. (2015) found that a recent failure (in their case, a pharmaceutical product 

recall) made the organization work to reduce future such events; however, as time passed, 

other performance metrics, such as profits, reclaimed the organization’s attention and lowered 

the focus on failure reduction.  

Some findings on how attention impacts motivation contradict one another. While more 

media coverage after an accident motivates greater investment in infrastructure for train 

companies, thus reducing accident risks (Desai, 2011); more media coverage of the near-

misses involving air traffic controllers instead reduced the effect of prior near-misses on 

learning (Desai, 2014). This divergence could be explained by near- misses looking alarming 

to outsiders since an actual failure (airline accident) would have catastrophic consequences, 

while organizational members, in contrast, see a near-miss as a success since an error was 

rectified and a failure was avoided (Kessels-Habraken, et al. 2010; Dillon & Tinsley, 2008) . 

 

ABILITY TO LEARN FROM FAILURE 

Page 20 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
21

Ability to learn from failure concerns the capacity to identify and report failure; 

understanding failure leads to finding and implementing solutions to prevent future failures. 

Individual and group level studies are concerned with training, emotional responses, shared 

goals and managerial style, as well as the interaction between ability and motivation. 

Organization-level studies rarely measure directly ability but rather conclude that the 

unobservable variances across units or organizations are due to differences in ability. In health 

care, where geography makes competition between hospitals less of an issue, checklists are 

used for the transfer of best practices across units, which raises the ability to learn in 

organizations.  

 

Individual- and group-level ability to learn from failure 

The question of how to improve individuals’ ability to learn from their errors and failure 

experiences has led to a series of studies that focus on comparing different training methods 

and the role of post-event reviews. Keith and Frese (2005) compared error management 

training, which explicitly addresses individuals who make errors during training and uses 

these errors as learning exercises and error avoidance training, which instead focuses on 

preventing participants from making errors. Error management training enhanced individuals’ 

capacity to cope with, and generate solutions to, new problems. Error management training 

also led to better emotion control and metacognitive activity, which in turn improved 

individuals’ ability to cope with new problems. Individual differences interact with training 

method so that individuals with better cognitive ability and higher openness to experience 

were found to perform better during error management training than if they had received no 

training or only training with a focus on error avoidance situations. Error management 

training not only provides trainees with opportunities to make errors but also to receive 

informative feedback; overall, these programs increase an individual’s ability to learn from 

Page 21 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
22

failure. Thus, the ability to learn from failure can be actively developed, and such training is 

more effective when it is paired with high motivation to learn from failure (Katz-Navon, et al. 

2005). 

A laboratory study on how to enhance individuals’ ability to draw lessons from previous 

experiences found that after-event reviews of both successful and failed events had a positive 

impact on individual learning (Ellis and Davidi, 2005). The after-event reviews followed a 

similar but more extensive process than that of error management training because it also 

included successful events. While the participants’ mental models of failure events were 

richer than those of success events, the performance improvements were greater when the 

after-event reviews focused on both failures and successes. In a follow-up study, any type of 

after-event review (success, failure or both types of events) was found to lead to better 

performance compared with no after-event review. Interestingly, for individuals who 

experienced successful events, the after-event reviews that focused on the failure factors that 

led to the greatest performance improvements had a greater impact than no reviews, reviews 

that focused on success factors, or on both success and failure factors (Ellis et al., 2006). 

Individual differences in how failure is processed has an impact on failure learning. 

Studying three types of coping orientations: loss, restoration, and oscillation, Shepherd and 

colleagues (2011) analyzed how these orientations affected (self-reported) learning. Loss 

orientation refers to the explicitly processing of a failure to break the emotion that is 

associated with the failure (a failed project). Restoration orientation refers to suppressing 

feelings of loss and instead proactively focusing on the tasks that arise as a consequence of 

the failure rather than preventing future failures. An oscillation orientation refers to moving 

back and forth between loss and restoration orientations. Individuals who have stronger loss 

and oscillation orientations reported a better ability to learn from previous project failure than 

those with a restoration orientation. Apparently, the necessary element is the capacity to 

Page 22 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
23

emotionally disconnect from the failure, which suggests that effective learning involves 

managing the emotions that are evoked by a failure.  

An individual’s ability and motivation to learn from failures are affected by emotional 

states. The ability to learn is enhanced when the environment is emotionally supportive: 

Individuals must feel comfortable applying their knowledge and acquiring new knowledge to 

learn from failures. An individual’s perception of psychological safety and the quality of his 

or her relationships with others in an organization are positively associated with failure 

learning (Carmeli and Gittell, 2009). High-quality relationships, which are manifested in 

shared goals, shared knowledge, and mutual respect, not only promote psychological safety 

but also enhance information processing and coordination capacity, which in turn have 

positive effects on the capacity to learn from failure. 

Individuals can also encourage failure learning at the group level with a positive 

leadership style, which includes clear direction and effective coaching (Cannon & 

Edmondson, 2001; Katz-Navon et al., 2009): leadership style can motivate learning; however, 

it also reflects the group’s ability to learn and a leader’s ability to enhance group learning. For 

example, when a CEO fosters psychological trust in a top management team (between the 

CEO and the team, and within the team), the team is more likely to engage in failure learning 

and to produce high quality strategic decisions (Carmeli, et al. 2012). 

Because individuals in organizations are embedded in groups, studies have stressed 

teamwork to be an important factor for failure learning (Baker, et al. 2006; Morey, et al. 

2002). Many studies of ability involve overlapping and interactive effects between motivation 

and ability, and some determinants are related to both mechanisms. Efficient teamwork relies 

on coordination and communication, which can be actively promoted (Baker et al. 2006) 

using formal training to improve team collaboration (Morey et al. 2002). Ability to foster 

cooperative goals also triggers failure learning (Tjosvold, et al. 2004). Most ability-type 

Page 23 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
24

factors can be taught, and they can, in turn, nurture group motivation to learn. Other group-

level factors that improve group learning ability are group member stability, understanding of 

team processes, and training to reduce attribution biases (Buljac-Samardzic, van Woerkom, 

& Paauwe, 2012; Morey, et al, 2002; Naquin & Tynan, 2003). 

A process study of of how ability is developed focuses on Israeli fighter crews (Ron, 

Lippschitz & Popper, 2006). To reduce errors in flight procedures, crews not only relied on 

their own learning based on flying more missions and thus learning through direct observation 

and experience; rather, because they knew that subjective perceptions are incomplete and 

sometimes faulty, given the intensity and massive information processing that is required 

when piloting a fighter jet, after each mission the entire cadre held a debriefing. In the 

debriefing crews watched footage from aircraft cameras while talking through their 

perceptions of what had occurred. The review process allowed them to compare their 

perceptions with the footage, clearly see how imprecise real-time perceptions are, which 

helped them to make corrections, and convert subjective experiences to objective ones. The 

review process and the debriefing are imposed by the organization, it enhanced the teams’ 

abilities and also increased information-based opportunities since the debriefing sessions 

provided the crews with additional information. 

 

Organization-level ability to learn from failure 

We found fewer studies on organizational-level ability than on organizational-level 

opportunity or motivation. A study analyzing failure reduction across clinics that use a new 

heart surgery method used an indirect measure of clinic ability (Pisano et al., 2001). Carefully 

establishing that given the same opportunity (cumulative number of procedures, which had a 

large impact on learning) and motivation, organizations differed in how well they converted 

experience into higher performance: the 16 clinics in the study demonstrated different 

Page 24 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
25

learning curve slopes. After ruling out other opportunity-based explanations, the conclusion 

was that differences across clinics must be due to (unobserved) ability. A paper argues that 

organizational form, whether an airline is a specialist or generalist, can influence its ability to 

learn from failure (Haunschild and Sullivan, 2002). Generalists have a more complex 

organizational structure with the potential to complicate information processing and 

coordination, which in turn hampers learning. In the airline industry specialists compared to 

generalists were better able to learn from failures with heterogeneous, that is, more complex 

causes. Similarly, smaller hospitals (more specialized), compared to large hospitals (more 

generalists), have been found to engage in more learning behaviors related to major adverse 

events and near-misses (Ginsburg et al, 2010). 

More detailed measures of ability demonstrate that ability affects the absolute number 

of errors as well as the reduction of errors: hospitals achieve faster learning and fewer errors 

by implementing protocols and checklists developed from best practices across the industry 

(WHO, 2017; Thornlow & Merwin, 2009; Thomassen et al., 2-13). Checklists were originally 

introduced in aviation, where they have been partly credited with the rapid decline in 

accidents (Clay-Williams & Colligan, 2015). Hospitals with patient-centered units were better 

at processing and reporting errors and near misses, which suggests that the organization of 

work matters for an organization’s ability to manage failures (Rathert & May, 2007).  

 

 

SYNTHESIS ACROSS LEVELS OF ANALYSIS 

A large body of literature finds that individuals, groups and organizations learn from prior 

failure experience. Information opportunities positively affect learning rate. The richer the 

information, the faster the reduction in errors and failure. Failure experiences generally 

contain more information than successful experiences (Kim & Miner, 2007). Most successes 

are “business as usual” contributing to learning by doing and other automatic responses, thus 

Page 25 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
26

enforcing existing routines. In contrast, failures trigger analyses and have greater potential to 

improve routines (Stan & Vermuelen, 2013). A lack of routines for error and failure 

management stops the information from passing through an organization (Gersick & 

Hackman, 1990; Kim & Miner, 2007). Larger-magnitude, more frequent and salient errors 

have greater information content and, hence, learning opportunities (Baum & Dahlin, 2007; 

Chuang & Baum, 2003; Madsen, 2009; Desai, 2011). Near misses also provide information 

about how to recover from a bad situation, which increases opportunities to learn (Kim & 

Miner, 2007). In general, studies find that “more is better” in regard to information-based 

learning opportunities; however, we would expect an overload when failures and errors are 

too frequent (Dahlin & Roulet, 2014). Information overload that leads to limited failure 

learning can hamper the ability to learn from failure. Accepting frequent errors due to 

information overload, that is, not learning from them, can also be a sign of low motivation to 

learn. These two factors—low ability and low motivation to learn—may be difficult to 

distinguish (Baum & Dahlin, 2007; Dahlin & Roulet, 2014).  

Time-based, or temporal, learning opportunities operate in the same direction and in a 

similar fashion as information-based learning opportunities: a lower work load leads to fewer 

errors and failures since work load reduces the time that is available for learning (Malone et 

al. 2007; Lawton, et al. 2012). Nevertheless, some groups are able to analyze and process 

information faster, taking advantage of temporal learning opportunities to enhance their future 

performance (Edmondson, et al. 2001). Whether temporal opportunities can trigger learning 

also depends on whether individuals in organizations have autonomy to process and reflect on 

the errors and failures that they encounter (Kerr, 2009; Stern, et al., 2008). To that end, 

organizational design (e.g., workload, task autonomy) has the potential to lead to latent failure 

that hampers learning opportunities. 

Page 26 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
27

Even when opportunity to learn from failure—whether information- or time-based—is 

high, learning may not occur. While opportunity studies are more concerned with what 

enables learning, motivation studies focus on why learning does not happen. Conditions under 

which individuals attribute the causes of errors and failures to other factors, groups attribute 

the causes of errors and failures to factors other than the group collective, and organizations 

attribute the causes of errors and failures to individuals rather than organizational factors, 

reduce the motivation to learn, which leads to low error and failure reporting and lower 

learning rates (Chuang et al., 2007; Ellis et al., 2006; KC, Staats & Gino, 2013). A cost-

benefit model that is proposed by Zhou and Olivera (2006) offers an overarching explanation 

to this problem: the unbiased reporting of errors is only expected when the perceived 

reporting cost is low to both an individual and his or her organization, and the perceived 

benefits for reporting is high to both as well as to any victim(s) who are associated with the 

errors. All other combinations of costs and benefits distort the motivation to report and what 

will be reported, which in turn affects the opportunity to learn from failure. In empirical 

studies, climate or psychological safety can be thought of as a reporting cost. Patient-centered 

climate (Rathert & May, 2007), low hierarchical distance, non-blaming cultures (Kathri et al. 

2009) and high psychological safety (Tjosvold et al. 2004) lead to better reporting and 

learning. The perception of the benefits to reporting are affected by the same factors; 

however, this side of the explanation for failure learning is less investigated. 

On the organizational level, conflicting goals, often safety and profitability, affect the 

motivation to learn from failure: A recent failure highlights safety goals for employees; 

however, as time passes, the focus reverts to financial performance metrics (Haunschild et al., 

2015). This highlights the risk of taking failure reduction for granted, assuming that learning 

is irreversible. Unless learning is embedded in physical artifacts (better brakes, a new IT 

Page 27 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
28

trading system) or it becomes a part of organizational routine, there is always the risk that 

error and failure rates will reverse. 

Ability studies show that training programs and after-event analyses can improve 

individual and group abilities to correctly analyze situations (Keith & Frese, 2005). It is clear 

that such training is quite common, with checklists being used in different industries, such as 

aviation and health care, to enhance organizational safety work (Clay-Williams & Colligan, 

2015). 

 

Integration across mechanisms 

Most studies focus on one or (at most) two mechanisms, either opportunity, motivation or 

ability, and we know that they all matter for failure learning. We know less about how the 

mechanisms affect one another. While some studies have considered moderating effects (KC, 

et al., 2013; Madsen & Desai, 2010; Shepherd at al., 2011), mediation is more seldom 

mentioned. If opportunity to learn sets the stage (providing information for failure analysis 

that can improve routines and failure responses), motivation causes actors to be willing to 

attend to such information, and ability is the conversion of opportunity into higher 

performance. What we do not clearly know is how, and whether, these three mechanisms 

jointly affect failure learning. 

Theoretical arguments about how to combine the different mechanisms argue for a 

three-way interaction (Blumberg & Pringle, 1982; Reinholt et al., 2011). However, it may just 

as well be a moderation-mediation process where the opportunity to learn interacts with the 

motivation to learn, and such interaction is mediated by the ability to learn, which results in 

failure learning. The empirical research focuses on one or two of the mechanisms, at most 

testing two-way interactions (for instance, how ability and motivation jointly determine 

learning). The notion that reality is complex is illuminated by different findings, some of 

Page 28 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
29

which find interactions between factors, others of which find that one factor affects learning 

but is mediated by a second factor. Motivation is thus found to affect ability, but ability also 

affects motivation (Lawton et al., 2012; Hofman & Stetzer, 1998) with both paths leading to 

learning. Further, motivation and opportunity jointly lead to learning (Haunschild et al. 2015), 

opportunity affects ability, which leads to learning (Stan & Vermuelen, 2013) and opportunity 

and motivation interact to produce learning (Baum & Dahlin, 2007). It is clear that the 

interplay between the mechanisms is more complex than we originally thought, and this 

promises many different ways to stimulate failure learning. 

 

MOVING BEYOND THE THREE FAILURE LEARNING MECHANISMS 

The papers in the review section covered both error and failure reduction. While there are 

many similarities between them, errors and failures are different types of events (Zhao & 

Olivera, 2006). Errors are mistakes, slips or violations of procedures, and they might or might 

not lead to an adverse outcome (Rasmusen, 1982; Reason, 1990). Failures are adverse 

outcomes
3
, such as accidents, unexpected patient deaths or bankruptcies (Tucker & 

Edmondson, 2003). While reducing errors should reduce failures, in this section we will 

discuss how these two types of events have a complex relationship; through a better 

understanding of this relationship, we can also better understand the differences in learning 

rates.   

 

Differentiating process and outcomes 

         When we study failures, that is, adverse outcomes such as product recalls, accidents, 

bankruptcies or unexpected hospital deaths, we assume that the cause of an adverse outcome 

is an erroneous process. Correspondingly, when we study traditional learning with successful 

outcomes, we assume that processes that lead up to the outcomes are correctly executed. 

                                                        
3
 We use the terms adverse outcomes, failures and undesirable outcomes interchangeably.  

Page 29 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
30

Questioning the strong link between the correctness of processes and outcomes, there is an 

increasing interest in situations in which erroneous processes can still result in good 

outcomes, such as near-misses (cf. Kim & Miner, 2007) and latent errors (Ramanujam, 2003; 

Reason, 1990). It is also possible, but rarely discussed, that a correct process can lead to an 

adverse outcome, such as a patient dying, even after well-executed surgery
4
. We propose that 

to better understand failure learning we must decouple processes and outcomes, or, more 

precisely, independently assess whether a process is correctly performed and whether an 

outcome is desirable. We also argue that different settings have different frequency 

distributions for the process-outcome pairs and that opportunity, ability and motivation to 

learn from failures (and successes) depend on this distribution, which explains why lessons in 

failure learning in one setting can be difficult to translate to another. 

If we regard processes as either correct or erroneous, and outcomes as successes or 

failures, we end up with four possible process-outcome combinations (see Table 2). 

———————————————————————- 

TABLE 2 TO BE INSERTED HERE 

———————————————————————- 

          Success and failure learning. A correct process with a favorable outcome represents 

traditional learning in which the outcome motivates actors who continuously improve and/or 

exploit the process to further enhance the outcome. This is what we expect from production 

learning curves (box 1 in Table 2). A faulty process with an undesirable outcome represents 

the “normal” failure learning case—an error is made, which yields a bad result (a train driver 

falls asleep, ignores a signal, and collides with another train; box 4). The main thrust in 

success learning is on how to improve existing processes to enhance the number of successful 

operations per time unit and thus lower the cost per unit produced (Argote, 2012; Yelle, 

                                                        
4
 Over time the number of different health interventions has greatly increased, many of which have been directed 

towards terminally ill patients (Maile, 2012). 

Page 30 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
31

1979). The main thrust in failure learning is the analysis of how to improve the processes, 

reduce errors, and thereby lower the risk of failure or reduce the number of failed operations 

per time unit (or as a share of all operations) (Reason, 1997).  

          The assumption is that box 4 is a good representation of failure situations: if there is a 

failure outcome, there must be a preceding error. As a consequence, failures should be 

reduced when errors are reduced, and this approach of error reduction lies behind much 

successful failure reduction (Reason, 1997; Van Dyck et al. 2005). Similarly, the assumption 

is that successful outcomes are due to an error-free process. However, there are two other 

possibilities in the matrix that complicate learning: the off-diagonal combinations where (1) 

an error has no effect on the outcome, that is, does not lead to a failure (Ramanujam, 2003), or 

(2) when there is a failure outcome without an error that has been committed (e.g., an “act of 

God”). The very existence of the off-diagonal combinations weakens the link between a 

process as being either correct and not, and its outcome as being either successful or not. We 

label the off-diagonal combinations spurious successes and spurious failures. An increase in 

spurious events leads to noisier learning processes because the cause-effect analysis becomes 

more complex, which in turn introduces difficulty in learning from both successes and 

failures. We expect that both spurious successes and failures complicate learning in general 

and failure learning in particular. Spurious successes and failures are likely to shift the 

attention of individuals and organizations away from the “true” causal-effects of failure, 

making failure learning challenging. Because successes tend to be vastly more common than 

failures,
5
 we failure learning should be more sensitive to the occurrence of spurious events.  

Spurious success. Not all errors lead to bad outcomes. We expect that both the 

motivation and ability to learn from an erroneous process is lower when an organization 

experiences a spurious success, that is, there is no negative outcome (Table 2, box 3). Further 

                                                        

 

Page 31 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
32

lowering learning ability, many process errors are unreported (and sometimes also 

unobserved) and unreported errors are known as latent errors (cf. Ramanujam, 2013; Reason, 

1997). Latent errors can lower failure reduction because a near-failure without an adverse 

outcome can strengthen an erroneous behavior.. The lower the likelihood that an error will 

lead to a failure, the more the error is accepted and the lower the motivation to correct the 

error or to learn from it is (Dillon & Tinsley, 2008; Banja, 2010). The motivation to reduce 

errors is also compromised because risk perception changes when actors experience errors 

without negative effects (a train driver falls asleep but wakes up before missing the signal, or, 

if the driver misses the signal, no other train is on the line, thus, there is no collision—the 

conclusion is that being tired on the job is not such a dangerous thing). Japanese nuclear firms 

did not learn from other firms’ errors without adverse outcomes when these errors were 

similar to non-adverse outcome errors the focal firm had experienced itself (Mitsuhashi, 

2012). This suggests that spurious success also lowers the motivation to learn from others’ 

errors: Knowing that other firms in the industry experience similar errors without adverse 

consequences signals that these errors pose no real risk and thus require little attention. When 

latent errors start being accepted by organizational members as not leading to adverse events, 

we obtain what is called normalization of deviance (Banja, 2010; Vaughan, 1996). 

Normalization of deviance involves accepting errors and rule breaking. Deliberate rule 

breaking is usually a dismissal of rules that are considered to be ill-conceived or overly 

complex. It is a gradual process (Vaughan, 1996) and in the majority of cases, normalized 

deviance has no negative outcome.  

              Normalized deviance is often exposed after a dramatic failure event leads to patient 

death (Maxfield, Grenny, Patterson, McMillan, & Switzler, 2005), nuclear meltdown 

(Dekker, 2011), spectacular trading losses (The Economist, 2014) or the crash of a space 

shuttle (Vaughan, 1996). Normalized deviance means that there is an implicit or explicit 

Page 32 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
33

agreement among organizational members to ignore certain safety procedures or regulations. 

One effect of normalized deviance is that it complicates cause-effect analyses when a failure 

strikes: it is easy for analysis to focus on the deviance behavior, which might not be the main 

cause of the failure. After all, some procedures or regulations are probably outdated or 

ineffectual and ignored for good reasons. In health care, normalized deviance often involves 

the violation of safety rules that impede work flow and signal a lack of trust in operators 

(Banja, 2010). Normalized deviance can also come from institutional logics that became 

dominant over time inside an industry despite their clashing with broader order values and 

beliefs that lie outside of the industry (Roulet, 2015; Shymko & Roulet, 2017).  

Despite a weak link between latent errors and failure outcomes (cf. Dekker, 2011), 

some settings still exhibit strong learning under such circumstances (airline safety, automobile 

safety), and it would enhance learning if we can determine factors that trigger learning from 

latent errors. First, the human error and safety literature is focused on errors regardless of 

outcomes, with a clear acceptance that adverse outcomes are quite infrequent; however, 

despite the low probability of an error that leads to a failure, there is nevertheless a link 

(Reason, 1997). When the focus is on errors rather than outcomes, both motivation and ability 

to learn should be enhanced. Creating agencies whose mission is error detection and 

reduction, such as the NTSB in transportation, leads to a low tolerance for errors. Some 

agencies have the power to close down or fine error-prone organizations with the aim of 

limiting such trade-offs in organizations, thus enhancing the organizations’ motivation to 

engage in failure reduction. Organizations’ and their members’ ability to reduce errors is also 

improved as regulators help with cause-effect analyses and recommend or regulate safer 

behaviors (FRA, 2016).  

        Spurious failures. The fourth process-outcome combination is the case where a faultless 

process produces an adverse outcome (Table 2, box 4). For instance, well-executed surgery 

Page 33 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
34

can still lead to a patient dying, or correct driver behavior can produce an accident. We call 

this a spurious failure, and this particular outcome is problematic for failure learning for a 

number of reasons. A failure outcome often triggers a search for causes even if there is none. 

Such a search risks misattributing the process as faulty, and there is a risk that the 

organization will replace a good routine with a worse one or make inefficient changes, and 

thus the ability to learn is compromised. The good-process-bad-outcome option is fairly rarely 

studied; however, it is likely to be frequent in complex settings where many processes co-

occur and involve many different actors. Because organizations are twice as likely as an 

impartial observer to assign blame to an operator after a failed event (Perrow, 1999), spurious 

failures are risky for individuals who might be unfairly blamed for adverse outcomes when 

the organization looks for failure causes. Thus, spurious failures can lead to lower trust and 

lower reporting of actual errors, which lowers the motivation to learn (Hofmann & Stetzer, 

1998).  

Health care is an obvious case where we expect frequent spurious failures because very 

sick patients eventually tend to die regardless of treatment. Staff is acutely aware that many 

failures (such as death) occur for reasons that are unrelated to any procedures that they 

perform, and this leads them to accept bad outcomes as an unavoidable part of everyday 

activities. While the acceptance of adverse outcomes is necessary, there is a risk of acceptance 

spilling over and allowing errors and latent errors to be forgiven: Because death is an 

expected and even unavoidable outcome for many patients, even when the cause of death is 

an error instead of an underlying disease and the outcome should be recorded as a failure, the 

high incidence of spurious failures might mean that errors are not detected. In addition, even 

if an error is detected, it might be ignored and therefore not corrected (Kohn, Corrigan & 

Donaldson, 2000). In other words, normalized deviance can be affected by spurious failures 

as well as by spurious successes. 

Page 34 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
35

In summary, the more frequent spurious learning and failure, or bad-good 

combinations, there are (boxes 2 and 3), the more difficult it is to perform a correct causal 

analysis. The combination of process error—good outcome is probably more common than 

that of the process error—bad outcome. An analysis of airline crews found that an error was 

made in the cockpit during a flight at least every four minutes but that very few incidents or 

accidents resulted (Bird, 1969; Reason, 1997). Similarly, nurses commit errors once every 

hour; however, this rarely leads to bad outcomes (Tucker & Spear, 2006). In contrast, we 

have almost no information about how often a correct process results in an adverse outcome. 

Learning motivation is low when the off-diagonal events are frequent, as is the opportunity to 

learn given the noisy information. 

When there are many processes that can simultaneously go wrong, attending to all 

potential errors becomes cumbersome and this also makes cause-effect analyses difficult. 

Simulations are well suited to investigating these trade-offs as well as experimenting with the 

complexity of tasks, the number of involved parties, risk levels and how these factors affect 

motivation and ability to learn (cf. Denrell, 2003).  

 

The impact of spurious successes and spurious failures on slow- and fast-learning 

settings 

          Noise and uncertainty in the process-outcome relationship help to explain learning-rate 

differences across settings. In settings with more off-diagonal outcomes (boxes 2 and 3), we 

expect lower learning rates since both ability and motivation to learn are lowered, and causal 

inferences are more difficult to draw. Spurious success (box 3), for example, is more likely to 

occur in situations in which tasks are easy and the rate of success is naturally high, while 

spurious failures (box 2) will, by contrast, be more common when tasks are complex, the 

chance of making an error is high and the risk of failure is also high. In the health care sector, 

Page 35 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
36

relatively routine procedures would be expected to lead to a higher rate of spurious success. 

By comparison, complex and high-risk surgical operations are more likely to lead to spurious 

failures. Those contexts offer fewer opportunities to learn because there is less information on 

which to draw for further success. The learning process is also hampered by the difficulty in 

assessing and taking stock of ability: the more spurious failures there are, the greater the 

doubt about current abilities will be. We also expect that motivation to learn will be lower in 

settings with high rates of spurious failure and success given the ambiguity with regard to 

cause and effect.  

 

FUTURE RESEARCH DIRECTIONS 

How can organizations best learn from errors and failures? We will suggest a number of 

approaches that are based on the ideas of maximizing the triad of learning mechanisms, 

opportunity, ability and motivation while taking into account the role of spurious failures and 

spurious successes. 

 

Opportunity to learn from failure 

Using multiple sources of event information and the role of regulations. What can 

organizations do when they experience few failures but still want to reduce future failure risk? 

There is an increasing emphasis in learning studies on vicarious learning in the form of 

organizations’ learning from similar others’ failures, successes and near-miss experiences 

(Baum & Dahlin, 2007; Haunschild & Sullivan, 2002; Kim & Miner, 2007; Madsen & Desai, 

2010), and thus one way to increase learning opportunities is to learn from others. Moreover, 

information about events is also provided by an array of industry stakeholders such as the 

press, unions, regulators, industry associations, insurance companies, equipment 

manufacturers, trade press and academics. All of these groups have an interest in failure 

Page 36 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
37

reduction, and some are mandated to collect data, investigate accidents and issue 

recommendations. They provide information, analysis and suggestions on how to reduce 

failures, and some of them are very active. However, management scholars have mostly 

ignored these stakeholders and the role that they play in identifying, analyzing and suggesting 

remedies for failures. As a consequence, there is an omitted variable bias in many studies, 

which overestimate the effect of one’s own or others’ ability to learn from failures (Dahlin & 

Roulet, 2014). In addition, it would be interesting to examine the relative impact of different 

learning sources on failure learning to better understand stakeholder roles. We call for future 

studies in management to include more industry stakeholders, or at least to control for their 

actions to better understand the sources of learning. While policy studies are engaging with 

this question to study the effect on an entire industry (Silbey, 2009; Dekker, 2011) they rarely 

analyze firm-level factors. 

Some key questions to ask when investigating the role of multiple parties are: Given 

different learning rates across industries: are there fewer sources of learning in slow-learning 

settings? Can a slower learning rate be explained by a lower opportunity to learn? 

Noise also lowers opportunity to learn: If slow-learning settings are more likely to 

combine a high failure volume and a high spurious failure volume, is it possible that lower 

failure learning rates are due to the difficulty to learn from spurious failures? Spurious failures 

create a great deal of noise, which makes cause-effect analyses complicated and makes more 

common failure cases (erroneous process—adverse outcome) difficult to analyze. Exploring 

these questions can further advance our understanding of the differences in failure learning 

across settings. 

Opportunity and the transfer of learning. One way to transfer best practices developed in a 

high-performing organization or industry to a lower-performing organization or industry is to 

use checklists (Degani & Wiener, 1993; Thomassen, Storesund, Søfteland, & Brattebø, 2014). 

Page 37 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
38

Checklists are required in aviation and their success has led to their adoption in health care. 

They are, for example, increasingly used in surgical procedures. Checklists have been 

effective in accelerating learning, with more rapid failure reductions in units that use surgery 

checklists than in units that do not use surgery checklists (Walker, Reshamwalla & Wilson, 

2012). However, in a meta-analysis of checklist studies, Thomassen et al. (2014) reported 

either improvement or no effects in the use of checklists. When some non-learning situations 

were more closely analyzed it was found that the checklists were not properly implemented—

some surgeons have resisted their use (Leape, 2014).  

Checklists are also used for data collection and analysis by US transportation safety 

agencies, such as the National Transportation Safety Board when they investigate accidents 

and has helped to reduce airline and train accidents (NTSB, 1998). Since the transportation 

sector is heavily regulated, all agencies are tasked with safety interventions and trained in 

using systematic tools while it is not clear if enforced checklists use by external agencies 

would be as effective in the health care setting. Maybe health care is too complex (different 

professional groups, hierarchical structures that complicate communication between groups, a 

large number of diagnostic and treatment options, complex information flow), thus requiring 

different tools to facilitate failure learning beyond learning on the procedural level? At the 

same time there are few other tools to facilitate information transfer that have been as 

extensively developed and whose impact is as well understood as the learning effects of 

checklists (Leape, 2014). We wonder what other methods and/or tools could be used to 

increase learning opportunities in general and which methods and/or tools might be effective 

for slow-learning environments in particular.  

 

Ability to learn from failure 

Page 38 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
39

         Ability is the learning mechanism that is most difficult to investigate, and we find a 

large gap in our understanding of how to improve ability. While there are case studies that 

conclude that failure learning is difficult, with many factors that are used to explain when 

learning will not occur (cf. Baumard & Starbuck, 2005), there are fewer studies that explain 

high learning rates (i.e., a better ability to learn from failure).  

Can ability to learn from failure be improved? After-event analysis and error management 

are ways to improve failure learning. Our review revealed that psychological safety and non-

hierarchical environments with good communication and coordination within and between 

teams also lead to more learning from errors and failures. This result then begs the question 

for organizations that lack good communication and have low psychological safety: how can 

they increase their failure learning ability? Can psychological safety be promoted by the same 

manager who pushes the norms of non-reporting? What types of action are necessary for 

organizations that wish to adopt the climate and norms of more successful failure learning 

organizations? Conversely, in organizations with good managerial support for error reporting 

and psychological safety, how can learning ability be further enhanced? 

Page 39 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
40

How much of learning is automatic? How does ability evolve? Some performance 

improvements in the learning-curve literature are virtually automatic; learning-by-doing is 

played out at the individual level (Thompson, 2012), and learning-by-doing implies that 

individuals’ abilities increase by themselves (but plateau after fairly few experience cycles, 

ibid.). How important is automatic learning in the context of failure learning? We expect 

some types of errors to diminish as operators gain experience. For instance, car drivers lower 

their accident risk within the first five years of driving, and we tend to ascribe this to success 

experiences. However, the failure learning studies have ignored the automatic learning effect. 

A general assumption is that performance improvements are due to active learning attempts. 

One of the most studied groups in the error research is nurses (e.g., Edmondson, 2004; Tucker 

& Spear, 2006); however, we see little discussion with regard to whether nurses’ error rates 

go down with tenure in the profession. Surgeons demonstrate failure reduction over time, thus 

it would be surprising if we would not see the same for nurses with respect to errors. 

However, if the link between errors and failures is weak, the effect of professional tenure may 

not be a reduction in errors but a reduction in errors that result in failed outcomes, that is, in 

converting potential failures into near-miss events.  

Again, considering the difference between settings, is there less room for automatic 

learning in slow-learning settings, thus requiring more deliberate learning for failure 

reduction? Examining these questions is both theoretically and practically meaningful. These 

questions help to advance our understanding as to how different types of learning occur. The 

answers to these questions have implications for practices as they help to develop intervention 

in organizations to enhance failure learning. 

 

Motivation to learn from failure 

Page 40 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
41

Scholars assume that safety and risk avoidance are central to any organization, such as 

airlines, mining companies, hospitals, or banks, and they often ignore that most organizations 

to some degree accept failure. Safer practices compete with productivity-enhancing 

investments and since viable operations are necessary for the future of organizations, 

productivity is usually prioritized over the potential risk of a future failure, impacting the 

motivation to invest in failure reduction (Haunschild et al., 2015). This is illustrated by 

airlines that experience more accidents after filing for Chapter 11 protection (ABCnews, 

2005) and by a famous quote in the freight rail industry: “Uphill slow, downhill fast, freight 

comes first and safety last” (Ahear & Schick, 2014). Low-prime-lending practices are risky 

behaviors that lead to short-term gains but jeopardize organizations and the banking industry. 

In other words, failure reduction attempts will almost always compete with other activities, 

which is a situation that we must acknowledge when theorizing about and investigating 

failure learning. In addition, some failures are seen as being unavoidable by managers, and 

they are relegated to productivity equations’ error terms and more or less accepted as a 

necessary evil rather than something to improve upon (cf. Jovanovic & Nyarko, 1995). Trade-

offs therefore help to explain the low motivation to reduce failure risks (Haunschild et al., 

2015). In many settings, regulators move in to change the balance, for instance, imposing 

fines if safety targets are not met. However, it ultimately comes down to decisions that are 

made by individuals and organizations on the costs that they are ready to allocate to learning 

and further improving their learning rate. 

This leads us to ask if slow-learning settings face stronger trade-offs between 

productivity-enhancing activities and failure reduction.  

Motivation and the case of non-learning: is motivation THE key factor in failure learning? 

A handful of empirical studies describe settings without failure learning even when similar 

events provide learning opportunities (Baumol & Starbuck, 2005; Tucker & Edmondson 

Page 41 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
42

2003). Their findings led some scholars to question if it failure learning is generally to be 

expected at all (Baumold & Starbuck, 2005). Analyzing these studies, we find that they 

describe situations with low motivation to learn (Baumold & Starbuck, 2005; Eggers & Song, 

2015), limited opportunity to learn in combination with low motivation to solve underlying 

problems (Tucker & Edmondson, 2003) and potentially an inability to learn due to low 

motivation (Eggers & Song, 2015). A conclusion is that failure learning is difficult since is 

only likely to happen when all three mechanisms are sufficiently activated, and that 

motivation is a necessary condition for deliberate learning. 

In an in-depth case study of a large European telecom firm that suffered 14 strategic 

failures with no learning, Baumol & Starbuck (2005) describe lack of systematic reporting 

about failed projects, lack of interest in better understanding what went wrong and managers 

making external attributions to explain away bad outcomes caused by internal factors. The 

authors find opportunity but low-to-no motivation to learn, which in the end resulted in no 

learning. Low motivation involved managers expecting that admitting failure would have 

negative career ramifications and possibly also harm the organization (many projects were 

imposed by external stake-holders and the firm was publicly listed). Applying Zhao & 

Olivera’s (2011) cost-benefit reasoning, the cost to the individual managers was high, the 

benefit to the organization not clear and, hence, low motivation to report was to be expected.  

We find a similar argument around individuals using external attributions rather than 

changing their own behaviour among Chinese serial entrepreneurs (Eggers & Song, 2015). 

Entrepreneurs with failed start-ups were less likely to alter the way they structured their 

companies than were entrepreneurs with successful start-ups. Rather, they pursued the same 

firm strategies and structures when launching a new venture which, in turn, increased the risk 

for failure. They preferred a new industrial setting which made the authors conclude that this 

inability to learn is in line with self-serving attribution theory: Using external attributions to 

Page 42 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
43

avoid altering ones’ method of working is an individual-level defense mechanism 

demonstrating both lack of motivation and inability to learn. In contrast, entrepreneurs with 

successful ventures who stayed in the same industry kept enjoying more success.  

Tucker and Edmondson (2003) report high incidences of errors without organizational-level 

learning in a study of nine hospitals known for nursing excellence. Front line nurses that 

constantly saw patients were confronted with many problems, such as errors and mistakes 

made by others, when executing their tasks. Surprisingly, the nurses’ error learning was low 

despite their skill and motivation. The reason for low error learning is that the nurses solved 

the problems themselves on an ad hoc basis since they rarely had time to deal with the 

underlying causes. As a consequence, the same errors were repeatedly made and failure rates 

remained high. The insight from this study is that the nurses were highly motivated to execute 

their work, got a confidence boost in effectively managing problems created by others, and 

had little time to provide feedback to the organization. This is both a story of lack of 

motivation that emphasizes that the motivation in question is about addressing underlying 

problems caused by others, and that the support structure did not offer sufficient opportunity 

to get to the root cause of the problems, both in terms of managerial support but also the lack 

of time-based opportunity to analyze errors and come up with solutions to reduce their 

recurrence.  

 Across these three non-learning cases motivation plays different roles: ego protection, 

career protection, protecting the organization and focus on other parts of the job than error, or 

problem reduction. In what way besides motivation do these non-learning cases contrast with 

learning cases? Are there more competing motivations in non-learning cases?  

Is it an illustration of how motivation is a necessary factor for learning to occur and without it 

neither opportunity nor ability matter? Also, how common are non-learning outcomes and 

Page 43 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
44

when do they matter? In the telecom case, the firm still did well and clearly was not very 

concerned with the failed strategic projects. When do organizations ignore failures?  

 

The nature of the failure event 

We suggest three areas of research to expand how we study failure events to improve 

causality in studies: using counterfactuals by applying the process-outcome matrix in Table 2 

when selecting events; extending our failure measures to also consider per-event learning; and 

standardizing or at least improving the measures of the learning process.  

Selection of the dependent variable. Few learning studies allow for both successes and 

failures to be key events in the research design (Denrell, 2003). Most studies choose to focus 

on either successful or failed outcomes (sometimes controlling for the other), and the set of 

factors that cause such outcomes. However, to extend our understanding of the relationship 

between processes and outcomes, the inclusion of both types of events in the research design 

would allow for stronger causal linkages. Currently, this is done in studies in which an 

intervention is randomly assigned to different organizational units and the outcomes are 

monitored, such as the introduction of checklists in health care into one subset of hospitals, 

while the comparison group works as before, and the reduction of adverse events is monitored 

(cf. Walker, Reshamwalla & Wilson, 2012). In checklist studies an outcome is either a 

success or a failure, thus the rate includes both possibilities while the focus remains on the 

effect of a single mechanism, the use of a checklist or not. The initiation of such studies are 

usually third-party organizations, such as the WHO, in the health care checklist example 

(WHO, 2017).   

Including both successful and failed outcomes would also allow for a study of the role 

that is played by noise in the learning process (the off-diagonal outcomes in Table 2) to cover 

the full range of process-outcome combinations, thus allowing us to better understand 

disturbances in learning processes.  

Page 44 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
45

Extending failure measures. The failure learning literature covers a diverse set of events, 

from bankruptcies to patient deaths to large-scale accidents that involve hundreds of victims 

who have been injured or killed. Some events occur once per hour and some occur once per 

week, while others occur once per decade. While the magnitude of events matter for learning, 

higher-impact events motivate organizations to respond more forcefully. An important 

question to ask is whether the sensitivity to failure events varies across industries, and, if so, 

what determines the level of sensitivity. In settings with many adverse events that are caused 

less by errors and more by the nature of operations (very elderly patients dying in health care, 

for instance), we expect a greater insensitivity to the adverse events that are caused by errors. 

Three dimensions that should matter when considering events are (1) the frequency with 

which failures occur; (2) the frequency with which adverse events that are NOT caused by 

errors occur; (3) the magnitude of the failure, including the failure magnitude when compared 

to industry averages (killed, injured, failure costs). Few studies show us how much a single 

failure event affects learning (see Dahlin & Baum, 2012 for an exception), which might 

provide a clue to how frequency and magnitude matter. A per-event measure would also make 

comparisons across setting more applicable. Here, a meta-analysis could reveal the different 

effects of failure events. 

 We expect the industries that experience a combination of high failure volume and 

high spurious failure volume to have an elevated error acceptance. Elevated error acceptance 

reduces the motivation to learn, which would explain a lower learning rate in such settings.  

 

Are learning rates overestimated? A paper critiquing the methods used in econometric 

studies of failure learning, argues that statistical estimations used in failure learning studies 

are prone to yielding falsely positive results (Bennett & Snyder, 2017). The authors point out 

that the classical learning model where current performance (cost per unit) is a logarithmic 

Page 45 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
46

function of accumulated experience (number of units produced) has some econometric issues 

when translated to failure learning and risk overstating the case for failure learning. Similarly, 

they point out the risk for false positive coefficient results increase when including successful 

as well as unsuccessful learning opportunities in the same equation. They recommend using 

moving time windows of past failures, and separating the success and failure opportunities 

into different estimations. Most studies already use moving windows and also discount events 

further back in time (albeit they do this for theoretical and data rather than estimation reasons) 

(cf. Baum & Dahlin, 20017; Kim & Miner, 2007).  

Even if Bennett and Snyder overstate the risk of falsely positive results, their paper 

raises a more fundamental question about econometric failure learning models. Usually a 

learning curve is assumed (often without this functional form being established as the one 

best-fitting the data) which means that a key assumption is that success and failure learning 

follow the same learning pattern. Similarly, when including both successful and failed prior 

events these are seen as additive which is an untested assumption. For instance, in traditional 

success learning, failed outcomes can be seen as reducing the learning rate but it is usually 

relegated to the error term. To model the interaction between success and failure, we need to 

develop new models accounting for both failure and learning as potential complements rather 

than substitutes. This way we could for instance see spill-overs from number of successful 

repetitions in learning-by-doing (such as driving a car) – when skills increase, errors and 

mistakes decrease as a function of more successful events. Rather than substitutes, there is 

complementarity with success learning leading to fewer failures and also enhancing failure 

learning. Conversely, we can observe that failure has a complex effect on normal operations, 

slowing down success learning since overall output might be reduced after a serious accident. 

The recommendation from Bennett and Snyder (2017) is to, for econometric reasons, not 

include both successful and failed events in the same estimations, we would argue that they 

Page 46 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
47

should also be estimated separately for theoretical reasons – we simply don’t know how they 

relate. 

 

Measuring failure and its responses. The empirical studies on failure learning face two 

measurement issues. First, individuals and organizations often fail to identify or report all 

failure events. This makes examining failures’ effects on learning processes more of a 

challenge. Common approaches to collecting failure and error data include using public 

archival data (e.g., Baum & Dahlin, 2007; Desai, 2015; Lawton et al., 2012), experimental 

designs (Dillon & Tinsley, 2008; Ellis et al., 2006), or employee recall (Ginsburg, et al., 

2010). Archival data and employee recall increase under-reporting biases, and the question is 

how to account for this when interpreting studies. Audits are, for instance undertaken in a 

number of industries, with the Federal Railroad Administration performing spot-checks on 

regulation compliance (FRA, 2017). Legal cases where non-reporting leads to penalties 

usually state the expected underreporting. Union representatives also have addressed 

underreporting.   

Second, our review revealed that there is a lack of common measures with regard to 

failure learning behaviors, such as the schemas that are used when responding to failures. 

Without common measures, it is difficult to compare the results from studies with different 

research settings. With few exceptions (cf. Ginsburg, et al, 2010; Shepherd et al., 2011) most 

studies measure broad learning behaviors, such as whether employees can challenge work 

processes or if they have improved work procedures, rather than learning behaviors that are 

specifically related to failures, such as the identification of an adverse event, cause analysis 

and corrective action. Most importantly, to better understand errors and failure learning 

requires a research design with an explicit link between failure events and the different 

Page 47 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
48

elements of learning behavior, such as Israeli flight crews’ post-event analyses (Ellis et al., 

2006).   

 

CONCLUSION 

Reviewing the literature on error and failure learning, we clustered learning mechanisms into 

three categories of opportunity to learn (the factors of information and time), motivation to 

learn (willingness to act) and ability to learn (competence to act). Failure and error learning 

have been studied across academic areas and address multiple levels of analysis, from 

individual, group and organizational perspectives. While motivation and ability factors 

dominate studies at the individual and group levels of analysis, opportunity factors dominate 

studies at the organizational level. Studies in health management are also concerned with 

procedure-level studies, with a wealth of data showing time trends for procedures on the 

national level, where learning diverges greatly across treatment types from negative, over no 

improvement, to some positive learning (Downey et al., 2012).  

 In summary, the findings suggest that more information about errors and failures in 

the form of one’s own and others’ prior failures or near failures facilitate learning. They go on 

to report that leaders and organizations with a punitive attitude toward errors and failures 

obtain less information because individuals in such organizations consider the cost of 

reporting to be too high. Additionally, they state that the ability to process and learn from 

errors and failures is partly based on attribution and inherent attitudes; however, this ability 

can be boosted through active post-event reviews. We argue that the cross-industry 

differences in learning rates depend on a number of factors. Among those factors, a noisy 

learning environment in which organizations experience spurious successes and spurious 

failures exerts a strong influence on failure learning. Noisy information about cause-effect 

makes failure information more difficult to interpret, which lowers learning opportunities and 

Page 48 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
49

ability. Spurious successes and failures also create ambiguity, which lowers the motivation to 

learn because the awareness of weak error-failure connections risks leading to normalized 

deviance, and it can also lower individuals’ motivations to report errors and failures. 

 An organization that seeks to enhance error and failure learning should analyze the 

causes of its most common errors and failures. To maximize the opportunity to learn, the 

organization should not only study its own but also the events of similar organizations. 

Further, including near-misses would add information and suggestions for ways to avoid an 

adverse outcome after a process error has occurred. When attributing causes, management 

must ensure that operators are not unduly blamed and if the cause is found to be operator 

error, look for systematic such errors to find if there are structural or organizational factors 

leading to operator error.  

 We propose a number of unexplored research areas in error and failure learning, some 

of which are related to a lack of linkages across academic disciplines. In short: management 

scholars tend to ignore the role that is played by regulators in failure reduction; policy 

scholars tend to ignore organizational differences and the role that is played by management 

in failure reduction; safety scholars have devised methods for the transfer of learning and for 

the development of ability; however, they have not yet fully established the efficacy of these 

tools. To that end, scholars across academic disciplines have “failed” to learn from each 

other’s failure research. Through our review, we have found fruitful opportunities for future 

research to learn from the extant failure studies to enhance our understanding of failure 

learning in organizations.  

 

  

Page 49 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
50

References 

Ahear, A. & Schick T. 2014. Rail workers raise doubt about safety culture as oil trains roll on. 

EarthFix, July 2. Retrieved February 25, 2015 from http://www.opb.org/news/article/workers- 

question-safety-culture-in-railroads-hauli/ 

Argote, L. 2012. Organizational learning: Creating, retaining and transferring knowledge. 

New York, NY: Springer. 

Argote L, McEvily B, and Reagans R. 2003. Managing knowledge in organizations: An 

integrative framework and review of emerging themes, Management Science, vol. 49, no. 4, 

pp. 571–582. 

Argote, L., & Todorova, G. 2007. Organizational learning. International review of industrial 

and organizational psychology, 22, 193. 

Arrow, K. 1962. The Economic Implications of Learning by Doing. Review of Economic 

Studies, 29, 166-170.  

Baker D. P., Day R., Salas E. 2006. Teamwork as an Essential Component of High-Reliability 

Organizations. HSR: Health Services Research. 41:4, II 

Baum, J. A. C., & Dahlin, K. B. 2007. Aspiration performance and railroads' patterns of 

learning from train wrecks and crashes. Organization Science, 18(3): 368–385,543–544. 

Banja, J. 2010. The normalization of deviance in healthcare delivery. Business Horizons, 

53(2): 159. 

Baum, J. A. C. and Ingram P. 1998. ‘Survival-enhancing learning in the Manhattan hotel 

industry, 1898–1980’. Management Science, 44, pp. 996–1016. 

Baumard, P., & Starbuck, W. H. 2005. Learning from failures: Why it may not happen. Long 

Range Planning, 38(3), 281-298. 

Bennett, V. M., & Snyder, J. (2017). The Empirics of Learning from Failure. Strategy 

Science, 2(1), 1-12. 

Page 50 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
51

Bernstein, E. S. 2012. The transparency paradox: A role for privacy in organizational learning 

and operational control. Administrative Science Quarterly, 57(2), 181-216. 

Binder, L. 2013. Stunning news on preventable deaths in hospitals. Forbes. 

http://www.forbes.com/sites/leahbinder/2013/09/23/stunning-news-on-preventable-deaths-in-

hospitals/#19b3f2c4ad08, last retrieved July 15, 2016. 

Blinder, A. 2016. Donald Blankenship Sentenced to a Year in Prison in Mine Safety Case 

https://www.nytimes.com/2016/04/07/us/donald-blankenship-sentenced-to-a-year-in-prison-

in-mine-safety-case.html New York Times, last retrieved July 15, 2017 

Blumberg, M., & Pringle, C. D. 1982. The missing opportunity in organizational research: 

Some implications for a theory of work performance. Academy of Management Review, 7: 

560–569.  

Buljac-Samardzic, M., van Woerkom, M., & Paauwe, J. 2012. Team safety and innovation by 

learning from errors in long-term care settings. Health Care Management Review, 37(3): 

280–291. 

Cannon, M. D., & Edmondson, A. C. 2001. Confronting failure: Antecedents and 

consequences of shared beliefs about failure in organizational work groups. Journal of 

Organizational Behavior, 22(2): 161–177. 

Cannon, M. D., & Edmondson, A. C. 2005. Failing to learn and learning to fail (intelligently): 

How great organizations put failure to work to innovate and improve. Long Range Planning, 

38(3), 299-319 

Carmeli, A., & Gittell, J. H. 2009. High-quality relationships, psychological safety, and 

learning from failures in work organizations. Journal of Organizational Behavior, 30(6): 

709–729. 

Page 51 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
52

Carmeli, A., Tishler, A., & Edmondson, A. C. 2012. CEO relational leadership and strategic 

decision quality in top management teams: The role of team trust and learning from failure. 

Strategic Organization, 10(1), 31-54. 

Carroll, J. B. 1963. A model of school learning. Teachers college record. 

Chuang, Y., & Baum, J. A. C. 2003. It's all in the name: Failure-induced learning by multiunit 

chains. Administrative Science Quarterly, 48(1): 33–59. 

Chuang, Y., Ginsburg, L., & Berta, W. B. 2007. Learning from preventable adverse events in 

health care organizations: Development of a multilevel model of learning and propositions. 

Health Care Management Review, 32(4): 330–340. 

Clay-Williams, R., & Colligan, L. 2015. Back to basics: checklists in aviation and healthcare. 

BMJ Qual Saf, 24(7), 428-431. 

Cyert, R. M., J. G. March. 1963. A Behavioral Theory of the Firm. Prentice-Hall, Englewood 

Cliffs, NJ. 

Dahlin, K.B. and Roulet, T.J., 2014. Different Slopes for Different Folks? Train Accident 

Learning Curves. Academy of Management Proceedings (Vol. 2014, No. 1, p. 11938). 

Academy of Management. 

Degani, A. and Wiener E. L. 1993. Cockpit checklists: concepts, design and use. Human 

Factors 35(2), pp. 28-43. 

Dekker, S. 2011. The criminalization of human error in aviation and healthcare: A review. 

Safety science, 49(2), 121-127. 

Denrell, J. 2003. Vicarious learning, undersampling of failure, and the myths of management. 

Organization Science, 14(3), pp. 227–243. 

Desai, V. 2011. Mass Media and Massive Failures: Determining Organizational Efforts to 

Defend the Field's Legitimacy following Crises. Academy of Management Journal 54: 263-

278. 

Page 52 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
53

Desai, V. 2014. The impact of media information on issue salience following other 

organizations’ failures. Journal of Management, 40(3), 893-918. 

Desai, V. 2015. Learning through the Distribution of Failures within an Organization: 

Evidence from Heart Bypass Surgery Performance. Academy of Management Journal 58: 

1032-1050. 

Desai, V. 2016. Learning to learn from failures: the impact of operating experience on 

railroad accident responses. Industrial and Corporate Change, 25(2), 199-226. 

Dillon, R. L., & Tinsley, C. H. 2008. How near-misses influence decision making under risk: 

A missed opportunity for learning. Management Science, 54(8): 1425–1440. 

Downey, J. R., Hernandez‐Boussard, T., Banka, G., & Morton, J. M. (2012). Is patient safety 

improving? National trends in patient safety indicators: 1998–2007. Health services 

research, 47(1pt2), 414-430. 

van Dyck, Frese, M., & Baer, S. Sonnentag. 2005. Organizational error management culture 

and its impact on performance: A two-study replication. Journal of Applied Psychology, 

90(6): 1228–1240. 

Edmondson, A. C. 1996. Learning from mistakes is easier said than done: Group and 

organizational influences on the detection and correction of human error. The Journal of 

Applied Behavioral Science, 32(1), 5–28. 

Edmondson, A. C., Bohmer, R. M., & Pisano, G. P. 2001. Disrupted routines: Team learning 

and new technology implementation in hospitals. Administrative Science Quarterly, 46(4), 

685–716. 

Edmondson, A. C., & Lei, Z. 2014. Psychological safety: The history, renaissance, and future 

of an interpersonal construct. Annual Review of Organizational Psychology & 

Organizational Behavior, 1(1), 23–43. 

Page 53 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
54

Eggers, J. P., & Song, L. 2015. Dealing with Failure: Serial Entrepreneurs and the Costs of 

Changing Industries Between Ventures. Academy of Management Journal, 58(6), 1785–

1803. 

Ellis, S., Mendel, R., & Nir, M. 2006. Learning from successful and failed experience: the 

moderating role of kind of after-event review. Journal of Applied Psychology, 91(3), 669. 

Ellis, S., & Davidi, I. 2005. After-event reviews: Drawing lessons from successful and failed 

experience. Journal of Applied Psychology, 90(5): 857–871. 

Forgas, J. P. 1995. Mood and judgment: the affect infusion model (AIM). Psychological 

bulletin, 117(1), 39. 

FRA. 2016. FRA Issues New Rule Requiring Passenger Railroads to Proactively Identify and 

Reduce Safety Risks. Federal Railroad Agency, Press Release FRA 21-16, July 2016 

https://www.fra.dot.gov/eLib/details/L18291#p1_z25_gD_lPR_y2016 

Frese, M., & Keith, N. 2015. Action errors, error management, and learning in organizations. 

Annual review of psychology, 66, 661-687. 

Gersick, C. J., & Hackman, J. R. 1990. Habitual routines in task-performing groups. 

Organizational behavior and human decision processes, 47(1), 65-97. 

Ginsburg, L. R., Chuang, Y., Norton, P. G., Berta, W., Tregunno, D., Ng, P., & Richardson, J. 

2009. Development of a measure of patient safety event learning responses. Health Services 

Research, 44(6): 2123–2147. 

Ginsburg, L., Chuang, Y.T., Berta, W.B., Norton, P.G., Ng, P., Tregunno, D., and 

Richardson, J., 2010. The relationship between organizational leadership for safety and 

learning from patient safety events. Health Services Research, 45: 607–632. 

Gully, S. M., Payne, S. C., K Lee Kiechel Koles, & Whiteman, J. K. 2002. The impact of 

error training and individual differences on training outcomes: An attribute-treatment 

interaction perspective. Journal of Applied Psychology, 87(1): 143–155. 

Page 54 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
55

Haunschild, P. R., & Rhee, M. 2004. The role of volition in organizational learning: The case 

of automotive product recalls. Management Science, 50(11), 1545–1560. 

Haunschild, P.R., Polidoro Jr. F. and D. Chandler. 2015. Organizational Oscillation Between 

Learning and Forgetting: The Dual Role of Serious Errors. Organization Science, 26(6). 

Haunschild, P.R. and B. Sullivan. 2002. Learning from complexity: Effects of 

accident/incident heterogeneity on airline learning. Administrative Science Quarterly 49: 

607–643. 

Heimbeck, D., Frese, M., Sonnentang, S., & Keith, N. 2003. Integrating errors into the 

training process: The function of error management instructions and the role of goal 

orientation. Personnel Psychology, 56(2): 333–361. 

Hodgkinson, G. P., & Wright, G. 2002. Confronting strategic inertia in a top management 

team: Learning from failure. Organization Studies, 23(6), 949-977. 

Hoff, T., Jameson, L., Hannan, E., & Flink, E. 2004. A review of the literature examining 

linkages between organizational factors, medical errors, and patient safety. Medical Care 

Research and Review, 61(1), 3-37. 

Hofmann, D.A. & Frese, M. 2011. Errors, error taxonomies, error prevention, and error 

management: Laying the groundwork for discussing errors in organizations. In D. Hofmann & 

M. Frese (eds), Errors in organizations, New York: Routledge, pp: 1–44. 

Hofmann, D. A., & Mark, B. 2006. An investigation of the relationship between safety 

climate and medication errors as well as other nurse and patient outcomes. Personnel 

Psychology, 59(4): 847–869. 

Hofmann, D. A., & Stetzer, A. 1996. A cross‐level investigation of factors influencing 

unsafe behaviors and accidents. Personnel Psychology, 49(2), 307-339. 

Huber, G. P. 1991. Organizational learning: The contributing processes and the literatures. 

Organization science, 2(1), 88-115. 

Page 55 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
56

Iedema, R. A. M., Jorm, C., Long, D., Braithwaite, J., Travaglia, J., & Westbrook, M. 2006. 

Turning the medical gaze in upon itself: root cause analysis and the investigation of clinical 

error. Social science & medicine, 62(7), 1605-1615. 

Ilgen, D., & Davis, C. 2000. Bearing bad news: Reactions to negative performance feedback. 

Applied Psychology, 49(3), 550-565. 

Ito, T. A., Larsen, J. T., Smith, N. K., & Cacioppo, J. T. 1998. Negative information weighs 

more heavily on the brain: the negativity bias in evaluative categorizations. Journal of 

personality and social psychology, 75(4), 887. 

Jones, E. E.; Harris, V. A. (1967). The attribution of attitudes. Journal of Experimental 

Social Psychology. 3 (1): 1–24.   

Jovanovic B, and Nyarko Y. 1995. A Bayesian learning model fitted to a variety of empirical 

learning curves. Brookings Papers on Economic Activity Microeconomics, 1995: 247–305. 

Kanfer, R., & Ackerman, P. L. 1989. Motivation and cognitive abilities: An 

integrative/aptitude-treatment interaction approach to skill acquisition. Journal of applied 

psychology, 74(4), 657. 

Katz-Navon, T., Naveh, E., & Stem, Z. 2009. Active learning: When is more better? The case 

of resident physicians' medical errors. Journal of Applied Psychology, 94(5): 1200–1209. 

KC, D., Staats, B. R., & Gino, F. 2013. Learning from my success and from others' failure: 

Evidence from minimally invasive cardiac surgery. Management Science, 59(11), 2435-

2449. 

Keith, N., & Frese, M. 2005. Self-regulation in error management training: emotion control 

and metacognition as mediators of performance effects. Journal of Applied Psychology, 

90(4), 677. 

Kerr, A. 2009. A problem shared…? Teamwork, autonomy and error in assisted conception. 

Social Science & Medicine, 69(12), 1741-1749. 

Page 56 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
57

Kessels-Habraken, M., Van der Schaaf, T., De Jonge, J., & Rutte, C. 2010. Defining near 

misses: Towards a sharpened definition based on empirical data about error handling 

processes. Social science & medicine, 70(9), 1301-1308. 

Khatri, N., Brown, G.D. and Hicks, L.L., 2009. From a blame culture to a just culture in 

health care. Health care management review, 34(4), pp.312-322. 

Kim, J. & Miner, A. S. 2007. Vicarious learning from the failures and near-failures of others: 

Evidence from the U.S. commercial banking industry. Academy of Management Journal, 

50(3): 687–714. 

Kohn, L.T., Corrigan, J. and M.S. Donaldson. 2000. To err is human: Building a safer 

health system. Washington, D.C.: National Academy Press. 

Kralewski, J. E., Dowd, B. E., Heaton, A., & Kaissi, A. (2005). The influence of the structure 

and culture of medical group practices on prescription drug errors. Medical care, 43(8), 817-

825. 

Lapré, M.A., Mukherjee, A. S. and L. N. Van Wassenhove. 2000. Behind the Learning Curve: 

Linking Learning Activities to Waste Reduction. Management Science  Vol. 46, No. 5, pp. 

597–611 

Lawton, R., Carruthers, S., Gardner, P., Wright, J. & McEachan R. (2012) 

Identifying the Latent Failures Underpinning Medication Administration Errors: An 

Exploratory Study. HSR: Health Services Research. 47:4 

Leape, L. L. 1994. Error in medicine. Journal of the American Medical Association. 

273(23): 1851-1857. 

Leape, L.L. 2014. The checklist conundrum. New England Journal of Medicine 

2014;370:1063 4.  doi:10.1056/NEJMe13158 

Page 57 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
58

Lei, Z., Naveh, E., & Novikov, Z. 2016. Errors in Organizations An Integrative Review via 

Level of Analysis, Temporal Dynamism, and Priority Lenses. Journal of Management, 

0149206316633745. 

Loewenstein, G. F., Weber, E. U., Hsee, C. K., & Welch, N. 2001. Risk as feelings. 

Psychological bulletin, 127(2), 267. 

Madsen, P. M., & Desai, V. 2010. Failing to learn? The effects of failure and success on 

organizational learning in the global orbital launch vehicle industry. Academy of 

Management Journal, 53(3): 451–476. 

Madsen, P., Dillon, R. L., & Tinsley, C. H. 2016. Airline Safety Improvement Through 

Experience with Near‐Misses: A Cautionary Tale. Risk analysis, 36(5), 1054-1066. 

Madsen, P. M. 2009. These lives will not be lost in vain: Organizational learning from 

disaster in US coal mining. Organization Science, 20(5), 861-875. 

Malone, D. C., Abarca, J., Skrepnek, G. H., Murphy, J. E., Armstrong, E. P., Grizzle, A. J., ... 

& Woosley, R. L. (2007). Pharmacist workload and pharmacy characteristics associated with 

the dispensing of potentially clinically important drug-drug interactions. Medical care, 45(5), 

456-462. 

Maxfield, D., Grenny, J., McMillan, R., Patterson, K., & Switzler, A. 2005. Silence kills. 

Vital Smarts, Crucial Conversations. 

Mitsuhashi, H., 2012. Almost identical experience biases in vicarious learning. Industrial and 

Corporate Change, 21(4), pp.837-869. 

Morey, J. C., Simon, R., Jay, G. D., Wears, R. L., & et al. 2002. Error reduction and 

performance improvement in the emergency department through formal teamwork training: 

Evaluation results of the MedTeams project. Health Services Research, 37(6): 1553–1581. 

Page 58 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
59

Muehlfeld, K., Sahib, P. R., & Van Witteloostuijn, A. 2012. A contextual theory of 

organizational learning from failures and successes: A study of acquisition completion in the 

global newspaper industry, 1981–2008. Strategic Management Journal, 33(8): 938–964. 

Naquin, C. E., & Tynan, R. O. 2003. The team halo effect: Why teams are not blamed for 

their failures. Journal of Applied Psychology, 88(2): 332–340. 

Neal, A., & Griffin, M. A. 2006. A study of the lagged relationships among safety climate, 

safety motivation, safety behavior, and accidents at the individual and group levels. Journal 

of applied psychology, 91(4), 946. 

Nisbett, R. E., & Ross, L. 1980. Human inference: Strategies and shortcomings of social 

judgment. Prentice-Hall. 

Paget, M. A. 1988. The unity of mistakes. Philadelphia: Temple University Press. 

Perrow, C. 1999. Normal accidents: Living with high-risk technologies. Princeton, N.J.: 

Princeton University Press. 

Pisano, G. P., Bohmer, R. M., & Edmondson, A. C. 2001. Organizational differences in rates 

of learning: Evidence from the adoption of minimally invasive cardiac surgery. Management 

Science, 47(6), 752-768. 

Probst, T.M., 2015. Organizational safety climate and supervisor safety enforcement: 

Multilevel explorations of the causes of accident underreporting. Journal of applied 

psychology, 100(6), p.1899. 

Ramanujam, R., & Goodman, P. S. 2003. Latent errors and adverse organizational 

consequences: A conceptualization. Journal of Organizational Behavior, 24(7): 815–836. 

Ramanujam, R. 2003. The effects of discontinuous change on latent errors in organizations: 

The moderating role of risk. Academy of Management Journal, 46(5), 608-617. 

Page 59 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
60

Rasmussen, J., Nixon, P. and F. Warner. 1990. Human Error and the Problem of Causality in 

Analysis of Accidents, Philosophical Transactions of the Royal Society, London, Volume B 

327, 449–462. 

Rathert, C., & May, D. R. 2007. Health care work environments, employee satisfaction, and 

patient safety: Care provider perspectives. Health care management review, 32(1), 2-11. 

Reason, J. T. 1990. Human error. New York; Cambridge University Press.  

Reason, J. 1997. Managing the risks of organizational accidents. Burlington, VT: Ashgate 

Publishing Co. 

Reinholt, M. I. A., Pedersen, T., & Foss, N. J. 2011. Why a central network position isn't 

enough: The role of motivation and ability for knowledge sharing in employee 

networks. Academy of Management Journal, 54(6), 1277–1297. 

Rerup, C. 2006. Success, Failure And The Gray Zone: How Organizations Learn Or Don't 

From Ambiguous Experience. In Academy of Management Proceedings (Vol. 2006, No. 1, 

pp. BB1-BB6). Academy of Management. 

Rhee, M. 2009. Does reputation contribute to reducing organizational errors? A learning 

approach. Journal of Management Studies, 46(4), 676-703. 

Ron, N., Lipshitz, R., & Popper, M. 2006. How organizations learn: Post-flight reviews in an 

F-16 fighter squadron. Organization Studies, 27(8), 1069-1089. 

Roulet, T. J. 2015. “What good is Wall Street?” Institutional contradiction and the diffusion 

of the stigma over the finance industry. Journal of Business Ethics, 130(2), 389-402. 

Salancik, G. R., & Meindl, J. R. 1984. Corporate attributions as strategic illusions of 

management control. Administrative science quarterly, 238-254. 

Shepherd, D. A., Patzelt, H., & Wolfe, M. 2011. Moving forward from project failure: 

Negative emotions, affective commitment, and learning from the experience. Academy of 

Management Journal, 54(6): 1229–1259. 

Page 60 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
61

Shymko, Y., & Roulet, T. J. (2017). When does Medici hurt DaVinci? Mitigating the 

signaling effect of extraneous stakeholder relationships in the field of cultural 

production. Academy of Management Journal. August 2017. 

Silbey, S. S. 2009. Taming Prometheus: Talk about safety and culture. Annual Review of 

Sociology, 35: 341–369. 

Sinclair, G., Klepper, S.& Cohen, W. 2000. What's experience got to do with it? Sources of 

Cost Reduction in a Large Specialty Chemicals Producer. Management Science, 46(1): 28-

45  

Sitkin, S.B. 1992. Learning through failure: The strategy of small losses. Research in 

Organizational Behavior, 14: 231–266. 

Stan, M., & Vermeulen, F. 2013. Selection at the gate: Difficult cases, spillovers, and 

organizational learning. Organization Science, 24(3), 796-812. 

Stern, Z., Katz-Navon, T., & Naveh, E. 2008. The influence of situational learning 

orientation, autonomy, and voice on error making: The case of resident physicians. 

Management Science, 54(9), 1553-1564. 

Thomassen Ø, Storesund A, Søfteland E, and Brattebøl, G. 2014. The effects of safety 

checklists in medicine: a systematic review. Acta Anaesthesiologica Scandinavica 

2014;58:5–18. doi:10.1111/aas.12207 

Thompson, P. (2001). How much did the Liberty shipbuilders learn? New evidence for an old 

case study. Journal of Political Economy, 109(1), 103-137. 
 

Thompson, P. 2012. The Relationship between Unit Cost and Cumulative Quantity and the 

Evidence for Organizational Learning-by-Doing, Journal of Economic Perspectives, 

American Economic Association, vol. 26(3), pages 203–24, Summer 

Tjosvold, D., Yu, Z. Y., & Hui, C. 2004. Team learning from mistakes: the contribution of 

cooperative goals and problem‐solving. Journal of Management Studies, 41(7), 1223-1245 

Page 61 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
62

Thornlow, D. K., & Merwin, E. 2009. Managing to improve quality: the relationship between 

accreditation standards, safety practices, and patient outcomes. Health care management 

review, 34(3), 262-272. 

Tucker, A. L., & Edmondson, A. C. 2003. Why hospitals don't learn from failures: 

organizational and psychological dynamics that inhibit system change. California 

management review, 45(2), 55–72. 

Tucker, A. L., & Spear, S. J. 2006. Operational failures and interruptions in hospital nursing. 

Health services research, 41(3p1), 643-662. 

Vaughan, D. 1996. The Challenger Launch Decision: Risky Technology, Culture, and 

Deviance at NASA.” 

Vogus, T. J., & Sutcliffe, K. M. 2007. The impact of safety organizing, trusted leadership, and 

care pathways on reported medication errors in hospital nursing units. Medical Care, 45(10): 

997–1002. 

Walker, I. A., Reshamwalla, S., Wilson, I. H. 2012. Surgical safety checklists: do they 

improve outcomes?. British Journal of Anaesthesy 2012; 109 (1): 47-54. doi: 

10.1093/bja/aes175 

WHO, 2017. The World Health Organization website, 

http://www.who.int/patientsafety/safesurgery/ss_checklist/en/ last accessed June 15, 2017. 

Wright, T. P. 1936. Factors affecting the cost of airplanes. Journal of Aeronautical Sciences, 

3(4): 122–128. 

Yelle, L. E. 1979. The learning curve: Historical review and comprehensive survey. Decision 

Sciences, 10(2), 302-328. 

Zhao, B., & Olivera, F. 2006. Error reporting in organizations. Academy of Management 

Review, 31(4), 1012-1030. 

Page 62 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
63

Zhao, B. 2011. Learning from errors: The role of context, emotion, and personality. Journal 

of Organizational Behavior, 32(3): 435. 

Zheng, Y. Miner, A. & George, G. 2013. Does the learning value of individual failure 

experience depend on group-level success? Insights from a University Technology Transfer 

Office. Industrial & Corporate Change. doi:10.1093/icc/dtt003 

Zohar, D. 2002. The effects of leadership dimensions, safety climate, and assigned priorities 

on minor injuries in work groups. Journal of Organizational Behavior, 23(1), 75-92. 

Zohar, D., & Luria, G. 2005. A multilevel model of safety climate: cross-level relationships 

between organization and group-level climates. Journal of applied psychology, 90(4), 616. 

 

 

 

Page 63 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
64

Table 1. Factors investigated in failure learning studies classified by level of analysis and mechanisms. 

 Opportunity Motivation Ability 

Individual 

level 

Perceived autonomy (Stern at el., 2008) 

Situational learning (Stern at el., 2008) 

Experience (KC, Staats, & Gino 2013) 

 

Motivation to learn (Zhao, 2011) and safety 

motivation (Buljac-Samardzic, et al. 2012; 

Neal & Griffin, 2006; Probst, 2015)  

Attribution (Ilgen and Davis, 2000; Nisbett and 

Ross, 1980; Ellis, Mendel & Nir, 2006; KC, Staats, 

& Gino 2013; Naquin & Tynan, 2003) 

Emotion (Edmondson, 1996; Paget, 1988; Keith & 

Frese, 2005; Neal & Griffin, 2006; Shepherd et al., 

2011; Zhao, 2011) 

Psychological safety (Carmeli & Gittell, 2009) 

Perception of outcomes (Dillon & Tinsley, 2008) 

Coping orientation (Shepherd et al. 2011) 

Training (Gully, et al., 2002; Keith & Frese, 2005) 

After-event reviews (Ellis & Davidi, 2005; Ellis, Mendel, & 

Nir, 2006) 

Coping orientation (Shepherd, et al., 2011) 

Relationships with others (Carmeli & Gittell, 2009) 

 

 

Group level Member rotation (Argote & Todorova, 2007) 

Group diversity and intergroup-linkages (Chuang et 

al. 2007; Tucker & Edmondson, 2003) 

Team stability (Edmondson, 1996) 

Routines to gather information (Lawton, et al. 2012; 

Edmondson, et al 2001; Tucker & Spear, 2006) 

Autonomy (Kerr, 2009) and load (Malone et al. 

2007) 

Group norms and team orientation (Katz-Navon, et 

al., 2009; Vogus & Sutcliffe, 2007; Zohar, 2002; 

Tjosvold et al. 2004) 

Psychological safety (Edmondson, et al. 2001) 

Tacit belief about failure (Cannon & Edmondson, 

2001; Edmondson, et al., 2001) 

Leadership style (Cannon & Edmondson, 2001; 

Katz-Navon et al., 2009; Carmeli et al., 2012) 

Safety climate (Zohar & Luria, 2005) 

Status dynxamics (Edmondson, et al. 2001) 

Routines (Gersick & Hackman, 1990) 

Debriefing and reviewing abilities (Ron, et al. 2006) 

Resistance to change (Hodgkinson & Wright, 2002) 

Workload (Lawton et al., 2012) and autonomy 

(Kerr, 2009) 

Failure climate (Buljac-Samardzic et al., 2012) 

Leadership style (Buljac-Samardzic et al., 2012; Zhao, 2011; 

Cannon & Edmondson, 2001) 

Training (Morey, et al. 2002) 

Membership stability (Buljac-Samardzic et al., 2012) 

Understanding of team process (Baker et al., 2006) 

Workload (Lawton et al., 2012) and work process 

(Edmondson, 1996) 

Coordination and communication (Baker et al. 2006) to 

develop cooperative goals (Tjosvold et al. 2004) 

Page 64 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
65

Organizational 

level  

(includes 

inter-

organizational 

factors) 

Own and others’ failure experience and processes to 

collect information on those failures (Tucker & 

Spear, 2006; Baum & Dahlin, 2007; Chuang & 

Baum, 2003; Haunschild & Rhee, 2004; Haunschild 

& Sullivan, 2002; Kim & Miner, 2007; Madsen & 

Desai, 2010; Madsen, Dillon & Tinsley, 2016) 

Own success experience (Madsen & Desai, 2010; 

Pisano et al., 2001) 

Recency of event (Haunschild, Polidoro & Chandler, 

2015) 

Magnitude of event (Desai, 2011; Madsen, 2009) 

Complexity of problems (Stan & Vermuelen, 2013) 

Organizational size (Desai, 2009; Slonim, 2007) 

Near failures / near misses (Kim & Miner, 2007; 

Kessels-Habraken, et al. 2010) 

Geographic proximity (Kim & Miner, 2007) 

Performance aspirations (Baum & Dahlin, 2007) 

Media attention (Desai, 2011 and Desai 2014; 

opposite results) 

Reputation (Haunschild & Rhee, 2004; Rhee, 2009) 

Safety climate (Hofman & Stetzer, 1998) 

Other’s similar errors (Mitsuhashi, 2011) 

 

Organizational form (Hanschild & Sullivan, 2002) 

Failure climate (Khatri et al., 2009) 

Error management culture (Dyck et al., 2005) 

Leadership style (Desai, 2015) 

Culture, work load (Kralewski, et al. 2005; Malone et al. 

2007) 

Post-experience reviews (Ron, Lipshitz & Popper, 2006) and 

external pressures (Haunschild & Rhee, 2004). 

Patient-centered hospitals, climate (Rahert & May, 2007; 

Lawton et al., 2012) 

Standardized procedures/protocols (Thornlow & Merlin, 

2009) 

Organizational size (Ginsburg, et al., 2010) 

 

 

Page 65 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



 
66

Table 2. 2x2 table of four process and outcome combinations. 

 Outcome of event 

Success Failure 

 

 

Process/behavior/routine  

 

Correct 

 

1. Traditional learning 

 

 

2. Spurious failure 

 

Faulty 

 

3. Spurious success 

 

 

4. Traditional failure 

learning 

 

 

 

 

 

Page 66 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



Authors JOURNAL YEAR LEVEL OF  

ANALYSIS 

O/M/A
1
 DATA/SETTING METHOD N RESULT 

Edmondson Journal of 

Applied 

Behavioral 

Science 

1996 Group O, A Patient care groups Survey 8 units L=O, A 

Hofman & Stetzer Academy of 

Management 

Journal 

1998 Group M Utility company Field 

experiment 

Safety 

communication and 

attributions in 159 

groups 

L: A�M 

Edmondson, Bohmer 

& Pisano 

Administrative 

Science 

Quarterly 

2001 Group M Team in hospitals Qualitative 16 hospitals L=M 

Cannon & 

Edmondson 

Journal of 

Organizational 

Behavior 

2001 Group M, A Organization teams Field study 51 groups L=M+A 

Haunschild & 

Sullivan 

Administrative 

Science 

Quarterly 

2002 Organization O, A airlines Archival data 310 airlines L=O*A,  

Specialist airlines learn more  

from heterogeneous accidents 

Tucker & Edmondson California 

Management 

Review 

2002 Group O, M Hospitals Qualitative 9 hospitals No learning 

Zohar Journal of 

Organizational 

Behavior 

2002 Group M Work groups Survey 49 groups L=M 

Gully, Payne, Kiechel 

Koles & Whiteman 

Journal of 

Applied 

Psychology 

2002 Individual A Undergraduate 

students 

Experiment 188 L=A 

Morey, et al. Health 

Services 

Review 

2002 Group A Emergency 

departments in 

hospitals 

Field 

experiment 

9 emergency 

departments 

L=A 

Hodgkinson & Wright Organization 

Studies 

2002 Team M Top management 

teams 

Qualitative 1 team L=M 

                                                        
1
 Please note that this is our interpretation of how to classify a study’s mechanisms. 

Page 67 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



Naquin & Tynan Journal of 

Applied 

Psychology 

2003 Individual M Graduate students Experiments 131 L=M 

Chuang & Baum Administrative 

Science 

Quarterly 

2003 Organization O Naming strategy Archival data 32 nursing home 

chain 

L=O 

Tjosvold, Yu & Huy Journal of 

Management 

Studies 

2004 Group O and M Organization teams Survey 107 teams L=M,O 

Keith & Frese Journal of 

Applied 

Psychology  

2005 Individual A Undergraduate 

students 

Experiment 55 L=A 

Ellis & Davidi Journal of 

Applied 

Psychology 

2005 Individual A Organizations Experiment 98 L=A 

Van Dyck, Frese, 

Baer & Sonnentag 

Journal of 

Applied 

Psychology 

2005 Organization A Organizations Survey 112 companies L=A 

Zohar & Luria Journal of 

Applied 

Psychology 

2005 Multilevel M Production workers 

in plants 

Survey 3952 individuals in 

401 group 

L=M 

Ron, Lippshitz & 

Popper 

Organization 

Studies 

2006 Group, 

organization 

M, A Fighter crews Observations of 

after-event 

reviews 

13 pilots and 

navigators 

L=A, M 

Neal & Griffin Journal of 

Applied 

Psychology 

2006 Individual M Hospital Longitudinal 

survey 

135 staff L=M 

Ellis, Mendel & Nir Journal of 

Applied 

Psychology  

2006 Individual M, A Undergraduate 

students 

Experiment 119 L=M, A 

Tucker & Spear Health 

Services 

Review 

2006 Multilevel O Nurses in Hospitals Qualitative 6 hospitals L=O 

Baum & Dahlin Organization 

Science 

2007 Organization O, M train accidents Archival data 189 railroad years L=M*O 

Underperforming train companies  

learn more from others’ failure 

experiences 

Page 68 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



Rathert & May Health Care 

Management 

Review 

2007 Individual A hospitals Survey 304 nurses, 3 

hospitals, same chain 

A gives M 

Kim & Miner Academy of 

Management 

Journal 

2007 Organization O Bank near-failures 

and failures 

Archival data 2696 banks L=O with near-failures being the 

important events 

Vogus & Sutcliffe Medical Care 2007 Group M Hospitals Survey 78 units in 10 

hospitals 

L=M 

Malone, Abarca, 

Skrepnek, Murphy, 

Armstrong, Grizzle, 

Rehfeld & Woosley  

Medical Care 2007 Multilevel O Pharmacy and 

pharmacist 

Survey + 

archival 

672 pharmacies L=O 

Stern, Katz-Navon & 

Naveh 

Management 

Science 

2008 Individual O Residents in 25 

departments of two 

teaching hospitals 

Survey 123 resdients L=O 

Dillon & Tinsley Management 

Science 

2008 Individual M Students and NASA Experiment 292 L=M 

Thornlow & Merlin Health Care 

Management 

Review  

2009 Organization A hospital routines Archival data 1430981 patient 

records, 115 

hospitals, 20 states 

L=A as a function of procedure 

Rhee Journal of 

Management 

Studies 

2009 Organization M Car recalls Archival data 54 U.S. automakers L=M, inverted U 

Madsen Organization 

Science 

2009 Organization O Mining accidents Archival data Recency of events L=O 

Kerr Social Science 

and Medicine 

2009 Group O, M Hospitals Qualitative 2 clinics L=O+M 

Carmeli & Gittell Journal of 

Organizational 

Behavior 

2009 Individual M, A Organizations and 

graduate students 

Survey 228 L=M+A 

Katz-Navon, Naveh & 

Stern 

Journal of 

Applied 

Psychology 

2009 Group M Hospitals Survey 123 residents, 25 

medical wards 

L=M 

Page 69 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



Madsen & Desai Academy of 

Management 

Journal 

2010 Organization O Orbital launch Archival data 26 launch vehicle 

organizations 

L=O 

Ginsburg, Chuang, 

Berta, Norton, Ng, 
Tregunno & 

Richardson  

Health 

Services 

Research 

2010 Organization A Patient safety Survey 69 hospitals L=A 

Desai Academy of 

Management 

Journal 

2011 Press release 

Organization 

O, M Railroad accidents Archival data 234 press releases, 

391 firm-years 

L=M, O 

Mitsuhashi Industrial and 

Corporate 

Change 

2011 Organization M and O Japanese nuclear 

plants 

Archival data  56 reactors L=M, O has no effect 

Banja Business 

Horizons 

2011 Event A, M Health care Case histories 

and theory 

Description of 

systematic rule 

breaking 

L=M, A 

Carmeli, Tishler & 

Edmondson 

Strategic 

Organization 

2011 Group M Top management 

teams 

Survey 77 TMTs L=M*A 

Zhao Journal of 

Organizational 

Behavior 

2011 Individual M Undergraduate 

students 

Survey with 

experiment 

127 L=M 

Shepherd, Patzelt & 

Wolfe 

Academy of 

Management 

Journal 

2011 Individual M, A Research institutes Survey 585 L=M+A+M*A 

Downey, Hernandez-

Boussard, Banka & 

Morton  

Health 

Services 

Research 

2012 Procedure Trends, no 

mechanism 

Medical failure 

trends using US 

national data 

Archival data 7.6 million adverse 

events for 69 million 

hospitalizations 

14 procedures: 6 with decrease 1 with 

no change and 7 with increases in 

adverse events 

Muehlfeld, Sahib & 

Van Witteloostujin 

Strategic 

Management 

Journal 

2012 Transaction O Newspaper merger 

and acquisitions 

Archival data 4973 m&a attempts 

1981-2008 by 1964 

firms 

L=O, U-shaped relationship 

Buljac-Samardzic, 

Van Woerkom & 

Paauwe 

Health Care 

Management 

Review 

2012 Group A Hospitals Survey 152 teams L=A 

Page 70 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



Lawton, Carruthers, 

Gardner, Wright & 

Mceachan 

Health 

Services 

Research 

2012 Group O, A Hospitals Qualitative 3 wards L=O*A 

Stan & Vermeulen Management 

Science 

2013 Organization O Fertility clinics Archival data 116 UK fertility 

clinics 1991-2006 

O--> A --> L 

Zheng, Miner & 

George 

Industrial and 

Corporate 

Change 

2013 Patent (IVs on 

individual and 

group levels) 

M, O University 

technology transfer 

office 

Archival and 

interviews 

778 patents of which 

170 were licensed 

L=M*O 

Kc, Staat & Gino Management 

Science 

2013 Individual O, M Hospitals Archival data 71 cardiac surgeons L=O+M+M*O 

Desai Journal of 

Management 

2014 Organization M Air traffic 

controllers, near 

misses 

Archival data US class I railroad 

firms 

L=M 

Haunschild, Polodori 

& Chandler 

Organization 

Science 

2015 Organization O Pharmaceutical 

firms, drug recalls 

Archival data 146 firms L=O, M 

Motivation varies over  

Time. 

Eggers & Song Academy of 

Management 

Journal 

2015 Individual M, A Serial entrepreneurs 

in China 

Survey 253 entrepreneurs L=M, A tested, no learning 

Probst Journal of 

Applied 

Psychology 

2015 Individual M 33 organizations Survey 1238 L=M 

Desai Academy of 

Management 

Journal 

2015 Organization O Heart bypass 

surgeries 

Archival data 116 hospitals L=O 

Madsen, Dillon & 

Tinsley 

Risk Analysis 2016 Organization O Airlines Archival data 64 commercial 

airlines 

L=O 

Page 71 of 72 Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60



Khanna, Guler & 

Kerkar 

Academy of 

Management 

Journal 

2016 Organization O Pharmaceutical 

industry 

Archival data 97 pharmaceutical 

firms 

L=O 

Desai Industrial and 

Corporate 

Change 

2016 Organization O Railroad accident Archival data US railroad firms L=O 

Bennett & Snyder Strategy 

Science 

2017 Event O Liver transplant data 

but mainly methods 

paper 

Simulated and 

archival data 

111619 US liver 

transplants 

No learning + risk of type I errors in 

learning studies 

 

Page 72 of 72Academy of Management Annals

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60