Sunday, February 10, 2013

Safety Culture - Lessons from the Social Science Literature

In 2011 the NRC contracted with the Pacific Northwest National Laboratory to conduct a review of social science literature related to safety culture (SC) and methods for evaluating interventions proposed to address issues identified during SC assessments.  The resultant report* describes how traits such as leadership, trust, respect, accountability, and continuous learning are discussed in the literature. 

The report is heavily academic but not impenetrable and a good reference work on organizational culture theory and research.  I stumbled on this report in ADAMS and don't know why it hasn't had wider distribution.  Perhaps it's seen as too complicated or, more importantly, doesn't exactly square with the NRC/NEI/industry Weltanschauung when the authors say things like:  

“There is no simple recipe for developing safety culture interventions or for assessing the likelihood that these interventions will have the desired effects.” (p. 2)

“The literature consistently emphasizes that effecting directed behavioral, cognitive, or cultural change in adults and within established organizations is challenging and difficult, requires persistence and energy, and is frequently unsuccessful.” (p. 7)

This report contains an extensive review of the literature and it is impossible to summarize in a blog post.  We'll provide an overview of the content, focusing on interesting quotes and highlights, then revisit Schein's model and close with our two cents worth.

Concept of safety culture

This section begins with the definition of SC and the nine associated traits in the NRC SC policy statement, and compares them with other organizations' (IAEA, NEI, DOE et al) efforts. 

The Schein model is proposed as a way to understand “why things are as they are” as a starting point upon which to build change strategies aimed at improving organizational performance.  An alternative approach is to define the characteristics of an ideal SC, then evaluate how much the target organization differs from the ideal, and use closing the gap as the objective for corrective strategies.  The NEI approach to SC assessment reflects the second conceptual model.  A third approach, said to bridge the difference between the first two, is proposed by holistic thinkers such as Reason who focus on overall organizational culture. 

This is not the usual “distinction without a difference” argument that academics often wage.  Schein's objective is to improve organizational performance; the idealists' objective is to make an organization correspond to the ideal model with an assumption that desired performance will follow. 

The authors eventually settle on the high reliability organization (HRO) literature as providing the best basis for linking individual and organizational assumptions with traits and mechanisms for affecting safety performance.  Why?  The authors say the HRO approach identifies some of the specific mechanisms that link elements of a culture to safety outcomes and identifies important relationships among the cultural elements. (p. 15)  A contrary explanation is that the authors wanted to finesse their observation that Schein (beloved by NRC) and NEI have different views of the the basis that should be used for designing SC improvement initiatives.

Building blocks of culture 

The authors review the “building blocks” of culture, highlighting areas that correspond to the NRC safety culture traits.  If an organization wants to change its culture, it needs to decide which building blocks to address and how to make and sustain changes.

Organizational characteristics that correspond to NRC SC traits include leadership, communication, work processes, and problem identification and resolution.  Leadership and communication are recognized as important in the literature and are discussed at length.  However, the literature review offered thin gruel in the areas of work processes, and problem identification and resolution; in other words, the connections between these traits and SC are not well-defined. (pp. 20-25)

There is an extensive discussion of other building blocks including perceptions, values, attitudes, norms**, beliefs, motivations, trust, accountability and respect.  Implications for SC assessment and interventions are described, where available.  Adaptive processes such as sense making and double-loop learning are also mentioned.

Change and change management

The authors review theories of individual and organizational change and change management.  They note that planned interventions need to consider other changes that may be occurring because of dynamic processes between the organization and its environment and within the organization itself.

Many different models for understanding and effecting organizational change are described.  As the authors summarize: “. . . change is variously seen as either pushed by problems or pulled by visions or goals; as purposive and volitional or inadvertent and emergent; as a one-time event or a continuous process. It is never seen as easy or simple.” (p. 43)

The authors favor Montaño and Kaspryzk’s Integrated Behavioral Model, shown in the figure below, as a template for designing and evaluating SC interventions.  It's may be hard to read here but suffice to say a lot of factors go into an individual's decision to perform a new behavior and most or all of these factors should be considered by architects of SC interventions.  Leadership can provide input to many of these factors (through communication, modeling desired behavior, including decision making) and thus facilitate (or impede) desired behavioral changes.

From Montaño and Kaspryzk
Resistance to change can be wide-spread.  Effective leadership is critical to overcoming resistance and implementing successful cultural changes.  “. . . leaders in formal organizations have the power and responsibility to set strategy and direction, align people and resources, motivate and inspire people, and ensure that problems are identified and solved in a timely manner.” (p. 54)

Lessons from initiatives to create other specific organizational cultures

The authors review the literature on learning organizations, total quality management and quality organizations, and sustainable organizations for lessons applicable to SC initiatives.  They observe that this literature “is quite consistent in emphasizing the importance of recognizing that organizations are multi-level, dynamic systems whose elements are related in complex and multi-faceted ways, and that culture mirrors this dynamic complexity, despite its role in socializing individuals, maintaining stability, and resisting change.” (p. 61)

“The studies conducted on learning, quality, and sustainable organizations and their corresponding cultures contain some badly needed information about the relationship among various traits, organizational characteristics, and behaviors that could help inform the assessment of safety cultures and the design and evaluation of interventions.” (p. 65)  Topics mentioned include management leadership and commitment, trust, respect, shared vision and goals, and a supportive learning environment.

Designing and evaluating targeted interventions 

This section emphasizes the potential value of the evaluation science*** approach (used primarily in health care) for the nuclear industry.  The authors go through the specific steps for implementing the evaluation science model, drilling down in spots to describe additional tools, such as logic modeling (to organize and visualize issues, interventions and expected outcomes), that can be used.  There is a lot of detail here including suggestions for how the NRC might use backward mapping and a review of licensee logic models to evaluate SC assessment and intervention efforts.  Before anyone runs off to implement this approach, there is a major caveat:

“The literature on the design, implementation, and evaluation of interventions to address identified shortcomings in an organization’s safety culture is sparse; there is more focus on creating a safety culture than on intervening to correct identified problems.” (p. 67)

Relation to Schein

Schein's model of culture (shown on p. 8) and prescriptions for interventions are the construct most widely known to the nuclear industry and its SC practitioners.  His work is mentioned throughout the PNNL report.  Schein assumes that cultural change is a top-down effort (so leadership plays a key role) focused on individuals.  Change is implemented using an unfreeze—replace/move—refreeze strategy.  Schein's model is recommended in the program theory-driven evaluation science approach.  The authors believe Schein's “description of organizational culture and change does one of the best jobs of conveying the “cultural” dimensions in a way that conveys its embeddedness and complexity.” (p. 108)  The authors note that Schein's cultural levels interact in complex ways, requiring a systems approach that relates the levels to each other, SC to the larger organizational culture, and culture to overall organizational functioning.

So if you're acquainted with Schein you've got solid underpinnings for reading this report even if you've never heard of any of the over 300 principal authors (plus public agencies and private entities) mentioned therein.  If you want an introduction to Schein, we have posted on his work here and here.


This is a comprehensive and generally readable reference work.  SC practitioners should read the executive summary and skim the rest to get a feel for the incredible number of theorists, researchers and institutions who are interested in organizational culture in general and/or SC in particular.  The report will tell you what a culture consists of and how you might go about changing it.

We have a few quibbles.  For example, there are many references to systems but very little to what we call systems thinking (an exception is Senge's mention of systems thinking on p. 58 and systems approach on p. 59).  There is no recognition of the importance of feedback loops.

The report refers multiple times to the dynamic interaction of the factors that comprise a SC but does not provide any model of those interactions.  There is limited connectivity between potentially successful interventions and desired changes in observable artifacts.  In other words, this literature review will not tell you how to improve your plant's decision making process or corrective action program, resolve goal conflicts or competing priorities, align management incentives with safety performance, or reduce your backlogs.

*  K.M. Branch and J.L. Olson, “Review of the Literature Pertinent to the Evaluation of Safety Culture Interventions” (Richland, WA: Pacific Northwest National Laboratory, Dec. 2011).  ADAMS ML13023A054

**  The authors note “The NRC safety culture traits could also be characterized as social norms.” (p. 28)

***  “. . . evaluation science focuses on helping stakeholders diagnose organization and social needs, design interventions, monitor intervention implementation, and design and implement an evaluation process to measure and assess the intended and unintended consequences that result as the intervention is implemented.” (p. 69)

1 comment:

  1. Great catch Lew,

    Given the date of publication of this information it seems strange that there is no reference to it on the NRC's NSC webpage - unless...

    I'm pretty sure that the prevailing "wisdom" in NRC/IAEA/NEI/INPO, regarding this artifact those refer to as Nuclear Safety Culture, have no answer for the assertion in these two sentences from the Executive Summary:

    "The framework presented in this report is based on a particular view of safety culture
    that is consistent with most of the relevant theoretical and empirical literature regarding
    organizational culture. Safety culture cannot be separated either logically or empirically
    from the dynamics of the broader organizational culture of which it is a part."

    The second sentence is a whopper - transliterating it says - "only in the nuclear enterprise is there a belief that 'nuclear safety culture' can be isolated for purposes of analysis and for the development of performance improvement initiatives."

    "Can't be separated logically or empirically" is tantamount to concluding that Positive Nuclear Safety Culture as referenced in the NRC's Safety Culture Policy is the figment of someone's imagination.

    That conclusion would seem to rule out the possibility that the many climate surveys being administered, those to test the waters of worker sentiments regarding how well they are being protected from work place hazards, bears no verifiable relationship to what everyone outside the nuclear industry terms "culture."

    I conclude from your generous review and the quick look I've taken of the report that Branch and Olson have made a good faith effort to "speak the truth to power." That "power," is not ready to hear such truth, should come as no surprise for those who insist on applying a Questioning Attitude to the dissonance between the "nuclear view" of culture and how it is described elsewhere.

    Its unfortunate that Dekker (who is not cited by the way) had not completed Drift into Failure before this research was done. Also Ed Schein continues to have his name taken in vane as the model at figure 2.1 bears little relationship to anything his full body of work would seem to recognize. The absent of reference to recent writing by him is also notable.



Thanks for your comment. We read them all. We'd like to display them under their respective posts on our main page but that's not how Blogger works.