Root Cause Analysis: Why? Why? Why? William A. Lindley April 6, 2001
Why Do Root Cause Analysis? “Just fix it, there is too much to do.” “We don’t have time to think, we need results now.” Reality - fix symptoms without regard to actual causes Root Cause Analysis - structured and thorough review of problem designed to identify and verify what is causing the symptoms
Definitions Cause (causal factor): a condition or event that results in an effect Direct Cause: cause that directly resulted in the occurrence Contributing Cause: a cause that contributed to the occurrence, but by itself would not have caused the occurrence Root Cause: cause that, if corrected, would prevent recurrence of this and similar occurrences
How Is Root Cause Analysis Done? Teams identify all possible causes The actual root causes are identified and verified Corrective action(s) are identified to reduce or eliminate the problem
RCA Process Need for creative thought to identify all possible causes Collect data about the problem Analyze data Verify causes Relationship between cause and effect
Root Cause Tools Cause and Effect Diagram Scatter Diagram - prove cause-effect relationship Control Chart - process stable? Five Whys Tree Diagram Change Analysis Barrier Analysis Event and Causal Factor Analysis Management Oversight & Risk Tree Analysis (MORT)
Cause Effect Diagram Visual display of possible causes Cause categories include materials, machines, methods, and people Reveals gaps in existing knowledge Helps team reach common understanding of why loss exists
Cause Effect Diagram Problem PeopleProcedures EquipmentMaterials
Cause Effect Diagram Danger: The Cause Effect Diagram is a list of potential root causes. This includes both probable causes, real causes and guesses.
After The Cause Effect Diagram Identify likely candidates for root cause(s) by one of the following actions: Look for causes that appear repeatedly within or across major cause or process categories Look for changes or other sources of variation in the process or environment Use consensus decision-making to select Collect data to confirm a potential root cause as real
Scatter Diagram Test for possible cause and effect relationships Some variation should be expected Relationships being tested must be logical Visual depiction of relationship
Patterns of Correlation Quality Improvement Tools Juran Institute, 1989
Correlation Coefficients Quality Improvement Tools Juran Institute, 1989
Data shows strong positive correlation. Scatter Diagram
Statistical Process Control Process Variation - Common Cause & Special Cause Is the process stable? Points outside LCL/UCL warrant investigation Alert for problems
Five Whys Describe the problem in specific terms For each likely cause ask, “Why did this happen?” Continue for a minimum of five times Show logical relationship of each response to the one that preceded it Stop when the team has enough information to identify the root cause
Tree Diagram State the problem Causes are listed as branches to the right of the problem Continue to clarify causes, drawing additional branches to the right Repeat until each branch reaches its logical end
Tree Diagram Example Training Class Cancelled Not enough students signed up Too much work No reward Schedule not communicated Trainer not prepared New trainer assigned late No time to learn Turnover Materials not completed Late changes Changes up to class date Flexibility Current Training Dept - other projects Floating due date This project- low priority More info needed
Cautionary Note “It’s impossible to solve significant problems using the same level of knowledge that created them!” Albert Einstein
Cautionary Note - Part 2 Cause and effect analysis can’t get past existing knowledge - must have either observed (or considered) that the cause produced the effect in the past
Why not just ask “Why”? Need to systematically organize and analyze data First understand “What happened” then “Why” Typically multiple root causes Blame is an obstacle Guidance needed to investigate human performance problems Need to ask right questions to completely understand why Some RCA techniques may provide easy answers that are either incomplete or wrong (but easy to find)
Event and Causal Factor Analysis Used for multi-faceted problems or long, complex causal factor chains Cause effect diagram that describes time sequence Anything that shapes the outcome recorded Identifies what questions to ask to follow path to root cause
Event and Causal Factor Analysis Event Potential Event Event Condition Sequence of happenings Conditions that may exist, but not identified Found or existing state that influences outcome Condition
Change Analysis Used when problem is obscure Generally used for single occurrence Focuses on things that have changed Compares trouble-free process with occurrence to identify differences Differences evaluated for contribution to occurrence
Change Analysis Steps Occurrence with undesirable consequence Compare Comparable activity without undesired result Identify differences Analyze differences for effect on undesired consequences Integrate information relevant to the causes of undesired consequence
Change Analysis Steps Answer the following: What? When? Where? How? Who?
Barrier Analysis Systematic process to identify barriers or controls that could have prevented the occurrence >Physical >Administrative >Procedural Determine why these barriers or controls failed What is needed to prevent reoccurrence
Barrier Analysis Sequence of events: System Tagout Tag Hung Electricians Given Assignment Electricians Follow Procedure Reactor Trip Barriers Analysis Start Tagout Process Step 1 Tagout Process Step 2 Communications Process Interface ProcedureOccurrence Barrier Holds Barrier Holds Barrier Holds Barrier Fails Barrier Fails Barrier Fails
Management Oversight and Risk Tree (MORT) Used to prevent oversight in the identification of causal factors Specific factors listed Management factors that permit these factors to exist listed Questions for each factor on the tree are included