Monday, February 13, 2012

Is Safety Culture An Inherently Stable System?

The short answer:  No.

“Stable” means that an organization’s safety culture effectiveness remains at about the same level* over time.  However, if a safety culture effectiveness meter existed and we attached it to an organization, we would see that, over time, the effectiveness level rises and falls, possibly even dropping to an unacceptable level.  Level changes occur because of shocks to the system and internal system dynamics.


Sudden changes or challenges to safety culture stability can originate from external (exogenous) or internal (endogenous) sources.

Exogenous shocks include significant changes in regulatory requirements, such as occurred after TMI or the Browns Ferry fire, or “it’s not supposed to happen” events that do, in fact, occur, such as a large earthquake in Virginia or a devastating tsunami in Japan that give operators pause, even before any regulatory response.

Organizations have to react to such external events and their reaction is aimed at increasing plant safety.  However, while the organization’s focus is on its response to the external event, it may take its eye off the ball with respect to its pre-existing and ongoing responsibilities.  It is conceivable that the reaction to significant external events may distract the organization and actually lower overall safety culture effectiveness.

Endogenous shocks include the near-misses that occur at an organization’s own plant.  While it is unfortunate that such events occur, it is probably good for safety culture, at least for awhile.  Who hasn’t paid greater attention to their driving after almost crashing into another vehicle?

The insertion of new management, e.g., after a plant has experienced a series of performance or regulatory problems, is another type of internal shock.  This can also raise the level of safety culture—IF the new management exercises competent leadership and makes progress on solving the real problems. 

Internal Dynamics    

Absent any other influence, safety culture will not remain at a given level because of an irreducible tendency to decay.  Decay occurs because of rising complacency, over-confidence, goal conflicts, shifting priorities and management incentives.  Cultural corrosion, in the form of normalization of deviance, is always pressing against the door, waiting for the slightest crack to appear.  We have previously discussed these challenges here.

An organization may assert that its safety culture is a stability-seeking system, one that detects problems, corrects them and returns to the desired level.  However, performance with respect to the goal may not be knowable with accuracy because of measurement issues.  There is no safety culture effectiveness meter, surveys only provide snapshots of instant safety climate and even a lengthy interview-based investigation may not lead to repeatable results, i.e, a different team of evaluators might (or might not) reach different conclusions.  That’s why creeping decay is difficult to perceive. 


Many different forces can affect an organization’s safety culture effectiveness, some pushing it higher while others lower it.  Measurement problems make it difficult to know what the level is and the trend, if any.  The takeaway is there is no reason to assume that safety culture is a stable system whose effectiveness can be maintained at or above an acceptable level.

*  “Level” is a term borrowed from system dynamics, and refers to the quantity of a variable in a model.  We recognize that safety culture is an organizational property, not something stored in a tank, but we are using “level” to communicate the notion that safety culture effectiveness is something that can improve (go up) or degrade (go down).

1 comment:

  1. Gentleman,

    This is a good discussion of systems dynamics; however, I suggest that there is one ambiguity that I'm not sure how you would clear up.

    Why "safety culture" effectiveness rather than "protection" or better yet "performance" (as the systems integral of "production" & "protection")?

    While there are lots of models of accident dynamics, I've yet to see one where "safety" is modeled as an outcome (I may be being a bit glib here saying never).

    Unlike protection I doubt that "safety" is a scalar variable even by analogy - for me its a go/no-go emotional state. As you point out snapshots of safety climate appear to be doable, but these rarely offer risk insights of momentum (functional direction and rate of decay).

    Risk insights about mission performance momentum are what risk managers need and if they don't develop them internally as a matter of course they generally prove resistant to being told about them by consultants or oversight bodies.

    As a case study, look at the Entergy Confirmatory Order that came out of some radcon problems at Fitzpatrick. Did the Entergy or NRC analysts of this handful of events correctly identify the ailing function (i.e. radcon); does the reported rate of functional failure appear to correspond to the scale of Entergy response?

    I would contend that NRC was looking for a developmental Safety Culture tightening-up guinea pig, and Entergy volunteered - on behalf of the entire US NP operations sector. This may prove a "useful" experiment for NRC; but were I a VPN at Excelon I'd not be too confident that Entergy has the "right stuff" leadership-wise to be recommending proposed changes to NEI-09-07.

    What became of INPO, it thought industry-wide performance improvement was their turf?


Thanks for your comment. We read them all. The moderator will publish comments that are related to our content.