|
INVESTIGATING
INVESTIGATIONS
to advance the
State-of-the-Art of
investigations, through
investigation process
research.
Research Resources:
Search site for::
Launched Aug 26 1996.
| |
Contents
REVIEW OF A ROOT CAUSE ANALYSIS DOCUMENT
- Summary and conclusions:
SUBJECT OF DOCUMENT: ROOT CAUSE ANALYSIS
Purpose of document:
Root cause defined:
Effectiveness elements
Purpose:
References.
Definitions:
- Causal factor
- Contributing factor
- Contributing causes:
Discussion section
Appendices
Conclusion
Root Cause Analysis is growing in popularity in the investigation field. It is based on the conventional wisdom in the field. The review suggests that Root Cause Analysis could be characterized as another monument to deficiencies in traditional investigation concepts and methodologies, and their inevitable result -- investigation process failures as demonstrated by the RCA worksheet lists.
The discovery problem described by Boorstin is alive and well today.
This is one of many documents promoting use of Root Cause Analysis in investigations. RCA had its origins in the nuclear industrial field, reportedly originating at the Savannah River facility when it was run by duPont. This specific document is offered to assist organizations that feel their programs and method could be improved by using the document. It addresses human and equipment performance. Pros and cons include;
- + recognizes that some organizations want to improve investigations
- + recognizes that process should address both human and equipment performance.
- - implies that all problems are people or object performance, and ignores conceptual, cultural, organizational and perceptual influences on performance.
The document devines root cause as "the fundamental cause(s) that, if corrected, will prevent recurrence of an event or adverse condition."
- - fundamental, event are undefined.
- - repackages obsolete "silver bullet" or simplified corrective action illusion so prevalent in safety and quality assurance community
- - embraces technically indefensible single "cause" determination which dominates present state of the art of investigation
- RCA Phases: Defines process in terms of 5 phases.
- - basis for phase determination or definition not indicated
- + recognizes need for management emphasis on identification of problems that can affect human and equipment performance, including assignment of evaluation and implementation responsibilities.
- + recognizes need to train managers and employees how to recognize and report events, and encourage supportive working environment for open process
- - focuses on RCA effectiveness, not effectiveness in terms of performance improvements
- - claims to present example methodology to implement RCA???
Next section describes Example Methodology.
1. Identify specific root causes of performance problems
All references are to the organization's internal documents.
Only more critical ones are discussed here.
a condition that shapes the outcome of a situation - asserts there are 17 potential causal factors associated with RCA. It then alludes to event causal factors (undefined) , and lists 12 for human problems, and 5 for equipment.
- definition uses "condition" and "situation" which reflect static rather than dynamic perception of process being investigated. Actions by people or objects or nature create "conditions" which are transient. Thus investigation and causes focus on outcomes, rather than interactions that produced the outcomes or conditions.
- definition indicates all ecf can all be abstracted to 17 generalized ones?
To list is to limit - especially opportunities for discovery of new interactions. Every one requires experientially based individualized judgment call to categorize observations by investigator. Challenges replicability! Solution: get committee consensus - later.
"a condition that may have affected an event."
- same problem - outcome orientation vs. dynamic process interaction orientation. Does this rule out actions or interactions as contributing factors?
"causes that, if corrected, would not by themselves have prevent the event, but are important enough to be recognized as needing corrective action to improve the quality of the process or product."
This definition illustrates many of the fundamental conceptual shortcomings of RCA:
- cause, without adjectives, is not defined. In working with the accident phenomenon, it can be demonstrated that for the process to achieve the observed outcome, ALL related interactions between people and objects had to occur, in the observed sequence, times and relationships.. If ANY interactions had been different, the outcome would have been different. Thus to select for analysis less than all the interactions required to produce the outcome without criteria that can produce replicable results, becomes very subjective. Subjective, inadequately supported conclusions are thus vulnerable to second-guessing and controversy. This shortcoming becomes most evident when differences arise between the investigator and the recommendation implementer over the investigator's recommendations.
- event is not defined. Usage in root cause analysis is ambiguous. At times it means the entire occurrence or phenomenon, as in this definition. In later Events and Causal Factors Charting , usage refers to actions, errors, malfunctions, failures, etc. with consequences. Exasperated by an ambiguous definition of recurring event. This reflects the broader issue of differing and unresolved perceptions of the nature and scope of phenomena that should be investigated. (What are incidents, accidents, near misses, etc.?)
- important enough is another instance of the RCA reliance on subjective judgments, without providing criteria or decision guidance that will produce replicable results and support quality control of the investigation or findings. Important is what a supervisor or manager says it is, which must be based on experientially developed decision rules (recognized as needing corrective action) - precluding an objective new problem discovery-facilitating process. Nowhere is the issue of unknown unknown (unk-unk) identification and resolution addressed. This reflects a broad research issue of case selection criteria for any investigation, and how case selection decisions are or should be made. It also indicates deeper methodological issues.
- quality of the process or product suggests the narrow perspective of the potential uses for investigation work products, such as improved process efficiency, reduced cost, improved corporate memory, discovery of patentable process changes, real time performance monitoring, training duration, etc. It may also reflect the influence of quality control thinking on the RCA processes, suggested again later by the analytical techniques presented.
In a discussion section, criteria for determining whether the investigator should feel that the root causes of a performance problem have been identified consist of asking whether the corrective actions prevent recurrence and if proposed corrective action will correct the root causes. Again, subjective decisions totally dependent on experience of manager or committee consensus. No objective quality control or validation is possible.
In the presentation of the RCA process, these issues are exacerbated by the techniques presented. For example, events and causal factors charting is obsolete, having been overtaken by more rigorous data formulation, organizing, display and testing procedures. On the other hand, a plus is a presentation of the root cause analysis methods to guide investigators. Unfortunately the triggers must be either people or object problem causes. This puts the cart before the horses, because the problem is supposed to be characterized subjectively - before the analysis to define the problem is selected. Hmmmm. Further the discussion suggests that cause-effect analyses can not be applied to human performance questions. (However, it has been.) To its credit, it does not base the investigation process wholly on logic tree analysis methods, but rather puts logic trees into a more limited perspective.
The appendices contain a list of generalize Event Causal Factor Categories and Causal Factor Worksheets. The lists are comprehensive, and provide a valuable form of corporate memory. They can be used as thought-starters to provide insights that are applicable in repetitive incidents. However, discovery becomes more a matter of chance than methodological design, because the list is based on knowledge of yesterday's experiences. "Experience recycles yesterday's problems."
The fact that such lists exist suggest an unfortunate conclusion. The suggest that prior investigations failed to resolve the problems, which in turn inescapably suggests that prior investigations were failures, and that new ways of investigating are needed to achieve better performance.
The appendices contain examples of the analysis methods, which are seriously flawed, in large measure because of the conceptual deficiencies and ambiguities mentioned above as well as others, such as the incorporation of interaction timing, logic flow testing and validation, etc.
Root Cause Analysis is growing in popularity in the investigation field. It is based on the conventional wisdom in the field. The review suggests that Root Cause Analysis could be characterized as another monument to deficiencies in traditional investigation concepts and methodologies, and their inevitable result -- investigation process failures as demonstrated by the RCA worksheet lists.
End note
It should be noted that the comments - favorable and unfavorable -about this document are also applicable to the most recent copy (1989) of the Department of Energy publication on Root Cause Analysis of Performance Indicators (WP-21 from System Safety Development Center, Idaho Falls, ID)
Ludwig Benner PE
created 5/94 reviewed 8/95
|