Classical Qualitative RCA Toolkits

Introduction

Root cause analysis (RCA) is a disciplined approach for understanding why an undesired event occurred and for selecting controls that prevent recurrence. In Total Quality Management (TQM), RCA supports the movement from symptom correction to system improvement. A defective product, delayed service, customer complaint, safety incident, or process deviation is not itself a root cause. It is an observable outcome produced by interacting technical, human, organisational, and environmental conditions.

Classical qualitative RCA toolkits are especially valuable during the early stages of investigation, when evidence is incomplete, processes are complex, and the investigation team must make implicit knowledge visible. This chapter examines six widely used tools: brainstorming, the 5 Whys, cause-and-effect diagrams, affinity diagrams, Is–Is Not analysis, and cause-and-effect matrices. These tools are complementary rather than interchangeable. Each structures a different kind of reasoning, and each can produce misleading conclusions when used without evidence, clear problem definition, or appropriate facilitation.

A central principle is that a proposed cause is a hypothesis until it is supported by evidence. A useful RCA therefore combines qualitative reasoning with observations, records, interviews, process data, tests, and verification of corrective action.

Learning objectives

By the end of this chapter, learners should be able to:

1. Explain the purpose and logic of classical qualitative RCA tools.
2. Select an appropriate tool for a particular investigation stage.
3. Facilitate brainstorming and distinguish idea generation from cause verification.
4. Apply the 5 Whys without confusing sequential questioning with proof of causation.
5. Construct and interpret cause-and-effect and affinity diagrams.
6. Use Is–Is Not analysis to narrow the investigation boundary.
7. Build and use a cause-and-effect matrix to prioritise suspected inputs.
8. Evaluate the strengths, limitations, governance risks, and appropriate combinations of these tools.

1. RCA as disciplined qualitative inquiry

1.1 Events, symptoms, causes, and controls

An RCA begins with a clearly stated problem. A strong problem statement describes what happened, where and when it happened, its magnitude, and its impact. It avoids premature explanations. “The packing machine failed because the operator was careless” is a conclusion, not a neutral problem statement. “Three cartons were incorrectly labelled on Line 2 during the night shift on 14 May” is more suitable.

The investigation commonly distinguishes four levels:

• Symptom or effect: the undesirable result observed by a customer or process owner.
• Immediate cause: the condition directly associated with the event.
• Contributing cause: a condition that increased the likelihood or severity of the event.
• Root or systemic cause: an underlying condition whose correction materially reduces recurrence risk.

The term “root cause” should not imply that every event has one single cause. Many quality failures arise from causal configurations: several conditions must coexist before failure occurs. For example, a wrong component may reach assembly because of ambiguous labels, similar storage locations, inadequate point-of-use verification, and production pressure. Removing only one condition may reduce, but not eliminate, recurrence.

1.2 Qualitative does not mean unstructured

Qualitative tools organise observations, experience, and reasoning. They do not automatically establish causality. A mature investigation uses a chain of reasoning:

1. Define the effect precisely.
2. Generate possible explanations.
3. Organise and prioritise them.
4. Collect evidence.
5. Test whether each suspected cause could produce the effect under the observed conditions.
6. Implement controls and verify that the effect does not recur.

This sequence prevents two common errors: jumping to a familiar explanation and treating group agreement as evidence.

2. Brainstorming: expanding the field of possible causes

Brainstorming is a structured method for generating ideas rapidly from a group with relevant knowledge. In RCA it is generally used after the problem statement has been established and before causes are organised or tested.

2.1 Basic method

A facilitator should first state the problem, scope, available evidence, and rules. Participants then produce possible causes without debate during the generation phase. The facilitator records each idea in neutral language, seeks clarification, combines only genuinely identical ideas, and later leads a separate evaluation phase.

Useful rules include:

• defer judgement during idea generation;
• encourage specific observations rather than accusations;
• build on others’ ideas;
• include frontline, technical, customer, and managerial perspectives where relevant;
• distinguish facts from assumptions;
• record minority views rather than allowing hierarchy to determine the list.

Brainstorming may be conducted verbally, silently using written cards, or electronically. Silent or nominal brainstorming can reduce production blocking and conformity, particularly when senior staff dominate discussion.

2.2 Strengths and limitations

Brainstorming is fast, inclusive, and effective for surfacing tacit knowledge. It is particularly useful when the process crosses departmental boundaries. However, it is vulnerable to anchoring, groupthink, availability bias, and power differences. Participants may list causes that are familiar rather than probable, or they may protect their department by shifting blame elsewhere. Brainstorming also produces a list, not a validated causal model.

The facilitator should therefore ask for evidence associated with each important idea: “What observation, record, or test would support or refute this possibility?”

3. The 5 Whys: tracing a causal pathway

The 5 Whys is an iterative questioning technique in which investigators ask why an effect occurred and then ask why the stated cause occurred. The number five is a prompt, not a rule. Some problems require three questions; others require more than five or require several branches.

3.1 Application principles

A sound 5 Whys analysis:

• begins with a precisely described effect;
• uses verified facts at each step;
• asks “why did this condition exist?” rather than “who is to blame?”;
• allows multiple branches where causes are not sequential;
• stops when the analysis reaches a controllable systemic condition, not merely a convenient explanation;
• identifies evidence needed to test each link.

3.2 Illustrative example

Effect: A customer received a product with an incorrect label.

1. Why was the incorrect label applied? The operator selected a label from the wrong electronic job folder.
2. Why was the wrong folder selected? Two active jobs had nearly identical product names.
3. Why could similar jobs be confused? The naming convention did not require a unique product code at the beginning of the file name.
4. Why was the naming convention inadequate? The document-control procedure specified approval and revision status but not human-factors requirements for job selection.
5. Why were those requirements absent? The process-design review did not include an operational usability or error-proofing assessment.

This chain identifies a design and governance weakness rather than ending at “operator error.” It does not prove that the naming convention was the only cause. Training, workload, software design, and label verification may require parallel analysis.

3.3 Strengths and limitations

The method is simple, inexpensive, and accessible to non-specialists. It encourages investigators to move beyond immediate symptoms and examine management systems. Its limitations are equally important. Causal relationships are rarely a single linear chain. Different investigators can produce different chains from the same evidence. The method can also become accusatory, speculative, or artificially deep. A “why” answer must be tested; it is not validated merely because it sounds plausible.

For complex failures, the 5 Whys should be combined with a cause-and-effect diagram, timeline, fault tree, or data analysis. Where several branches exist, each branch should be analysed separately and then reconciled.

4. Cause-and-effect diagrams

The cause-and-effect diagram, often called the fishbone or Ishikawa diagram, displays possible causes of a specified effect. The effect is written at the head of the diagram, while major categories form the principal branches and detailed causes form sub-branches.

Common manufacturing categories include Man, Machine, Method, Material, Measurement, and Mother Nature (the six Ms). Service and knowledge-work investigations may use categories such as people, process, technology, information, environment, policy, and customer interface. Categories should be adapted rather than applied mechanically.

4.1 Construction

1. Write a measurable problem statement at the head.
2. Select categories appropriate to the process.
3. Ask participants to identify possible causes under each category.
4. Add sub-causes until the proposed explanations become sufficiently specific to investigate.
5. Mark causes supported by evidence, causes requiring evidence, and causes contradicted by evidence.
6. Select causes for verification and corrective action.

The diagram is a hypothesis map. It does not show that every listed cause occurred, nor does it establish relative importance. A long branch may reflect detailed discussion rather than greater causal significance.

4.2 Strengths and limitations

The diagram provides breadth and prevents premature concentration on one category. It exposes interactions between technical and organisational conditions and gives a team a shared visual model. Its limitations include possible clutter, arbitrary categories, duplication, and the tendency to list generic causes such as “poor training” without specifying the mechanism. It can also encourage a false assumption that all categories are independent.

A useful refinement is to convert vague entries into testable statements. Replace “poor maintenance” with “preventive-maintenance intervals were not defined for the sensor model installed on Line 3.”

5. Affinity diagrams: discovering natural themes

An affinity diagram groups numerous observations, statements, or suspected causes according to their natural relationships. Unlike a fishbone diagram, which begins with predetermined analytical categories, an affinity diagram is usually bottom-up. It is valuable when an investigation produces large volumes of interview comments, complaint descriptions, audit observations, or brainstorming cards.

5.1 Method

The team writes one idea per card, clarifies ambiguous statements, and places related cards together without initially imposing category names. After clusters emerge, the team assigns descriptive headers. Higher-level clusters may then be formed. The resulting structure can reveal themes such as information flow, equipment condition, workload, supplier variation, and governance.

The grouping process should preserve the original statements and record disagreements. Forced consensus can conceal important alternative interpretations.

5.2 Strengths and limitations

Affinity diagrams are effective for reducing unstructured qualitative material and giving equal visibility to diverse contributions. They are useful before constructing a fishbone diagram because the clusters can suggest context-specific categories. However, proximity does not prove causality. Two observations may belong together because they concern the same topic, while only one is causally relevant. Grouping is also sensitive to wording, facilitator influence, and the participants’ mental models.

6. Is–Is Not analysis: defining the boundary

Is–Is Not analysis compares what is occurring with what is not occurring. It is a disciplined form of contrast analysis that helps narrow the search space and identify distinguishing conditions.

A typical worksheet considers:

The “is not” information is not proof that excluded factors are irrelevant. It provides constraints and comparative evidence. The method is especially powerful when the effect is intermittent, localised, or associated with a specific product, shift, machine, supplier, or time period.

6.1 Strengths and limitations

Is–Is Not analysis reduces vague problem boundaries and helps investigators compare affected and unaffected cases. It supports efficient data collection and can expose patterns missed by open-ended discussion. Its limitations arise when “not” data are weak, samples are small, records are incomplete, or the comparison group is not genuinely comparable. Absence of an event is not necessarily evidence of absence of a cause.

7. Cause-and-effect matrices: linking inputs to outputs

A cause-and-effect matrix evaluates how process inputs may influence important outputs. It is useful when many possible inputs exist and the team needs a transparent prioritisation method.

The team lists critical outputs, assigns each output an importance weight, lists process inputs, and rates the strength of each input–output relationship. A common prioritisation score for input i is:

Here, P_i is the priority score for input i; w_j is the importance weight of output j; and r_ij is the relationship rating between input i and output j. The summation covers all selected outputs. Ratings may use a defined ordinal scale such as 0, 1, 3, and 9, where 0 means no apparent relationship and 9 means strong relationship. The numbers are prioritisation scores, not measurements of physical causality.

Worked example. Suppose customer safety has weight 5 and delivery accuracy has weight 3. Input A has relationship ratings 9 and 3 respectively. Its score is:

Input B, with ratings 3 and 9, scores:

Input A receives higher priority under this scoring convention. The result depends on the selected outputs, weights, rating definitions, and expert judgement; it should guide evidence collection rather than replace it.

7.1 Strengths and limitations

The matrix makes prioritisation assumptions visible and connects process inputs to customer or business requirements. It is useful for selecting measurement plans and concentrating scarce investigation resources. Its weaknesses include ordinal-rating subjectivity, false precision, double counting, and sensitivity to the chosen weights. Teams should document the rationale for ratings, conduct sensitivity checks when decisions are consequential, and validate high-priority inputs using data or controlled tests.

8. Integrating the toolkits into an RCA workflow

No single classical tool is sufficient for a complex investigation. A practical sequence is:

1. Define the effect: write a factual problem statement and establish scope.
2. Contrast the cases: use Is–Is Not analysis to identify affected and unaffected conditions.
3. Generate possibilities: use brainstorming and relevant records or interviews.
4. Organise information: use an affinity diagram for unstructured observations.
5. Model causes: use a cause-and-effect diagram to structure categories and interactions.
6. Trace important pathways: apply the 5 Whys to selected branches.
7. Prioritise verification: use a cause-and-effect matrix where multiple inputs and outputs compete for attention.
8. Verify: inspect records, observe work, sample products, test equipment, or compare conditions.
9. Control and learn: implement corrective actions, assess unintended effects, and monitor recurrence.

The sequence is not rigid. A serious safety event may require immediate containment and evidence preservation before group discussion. A recurring service complaint may begin with affinity analysis of customer narratives. The tool must fit the question.

9. Illustrative case: intermittent medication-packaging errors

Background

A fictional pharmaceutical packaging facility experienced seven customer complaints in one month involving cartons containing the correct medicine but an incorrect language label. No patient harm was reported, but the event created regulatory and recall risk.

Problem definition and Is–Is Not analysis

The quality team established that the errors occurred on Packaging Line 4, during late-shift changeovers, and affected two products with similar carton designs. Lines 1–3 had no recorded cases. The problem was intermittent: 7 errors in 18,600 cartons.

The contrast suggested that the investigation should examine line-specific configuration, changeover work, product similarity, and late-shift conditions rather than assuming a general labelling failure.

Brainstorming and affinity grouping

Participants generated 46 possible causes. An affinity diagram produced five clusters: label identification, electronic work orders, changeover practices, environmental and workload conditions, and verification controls. “Operator inattention” appeared several times, but the team required a mechanism and evidence before accepting it as a cause.

Cause-and-effect diagram and 5 Whys

A fishbone diagram showed that the label-selection branch connected human action with system design. The 5 Whys revealed that similar products were displayed in adjacent folders, the product code was not prominent, and the scanner confirmation step had been bypassed during a changeover because the scanner was positioned away from the operator’s normal work area.

Cause-and-effect matrix

The team rated label identification, scanner availability, changeover checklist quality, and staffing against safety, regulatory compliance, and throughput outputs. Scanner availability and label identification received the highest scores because they strongly affected the two highest-weight outputs.

Verification and solution

Record review showed that all seven errors occurred when the scanner was unavailable or bypassed. Direct observation found that the scanner cable was routinely disconnected during cleaning and not reconnected before the next job. A controlled observation confirmed that the electronic system permitted completion without a successful scan.

Corrective actions included relocating and protecting the scanner, making scan confirmation a system-required step, revising the changeover checklist, and redesigning folder names to begin with a unique product code. Training was used as a supporting control, not as the sole solution.

Outcome and learning points

During the following three months, no comparable error occurred across 31,000 cartons, and scan-compliance records showed 100% completion. The case demonstrates the value of combining tools: Is–Is Not narrowed the field, brainstorming expanded it, affinity grouping organised it, the fishbone diagram exposed interacting categories, the 5 Whys traced a pathway, and the matrix supported prioritisation. Verification, rather than the diagrams themselves, established the basis for action.

10. Critical limitations, ethics, and governance

RCA can cause harm when it becomes a search for personal blame. Blaming an operator may obscure poorly designed interfaces, unrealistic schedules, inadequate maintenance, or contradictory policies. Investigators should use just-culture principles: distinguish human error, at-risk behaviour, and reckless behaviour, and match the response to the nature of the conduct and the system context.

Confidential interviews, personnel records, and customer information must be handled according to organisational policy and applicable privacy and employment law. Evidence should be retained, traceable, and protected from alteration. Investigators should disclose uncertainty and avoid claiming that a cause is proven when it is only plausible.

Technology can support electronic brainstorming, text clustering, process mining, and analysis of complaint narratives. Artificial intelligence may help identify themes or suggest questions, but generated suggestions are not evidence. Sensitive data require appropriate access controls, and human investigators must validate classifications, protect confidentiality, and examine algorithmic bias. Automation should improve inquiry, not replace professional accountability.

11. Best-practice checklist

• Use a neutral, measurable problem statement.
• Separate idea generation, organisation, prioritisation, and verification.
• Record evidence and uncertainty for every important suspected cause.
• Adapt categories to the process rather than using a fixed fishbone template.
• Treat 5 Whys as branching inquiry when the system is complex.
• Use Is–Is Not comparisons to guide sampling and observation.
• Document matrix weights and rating definitions.
• Prefer system and process controls over reliance on memory or vigilance.
• Confirm that corrective action changes the causal condition.
• Monitor recurrence and unintended consequences.

Chapter Summary

Classical qualitative RCA toolkits provide complementary ways to investigate quality problems. Brainstorming expands the range of possible explanations; the 5 Whys traces causal pathways; cause-and-effect diagrams organise possible causes by process categories; affinity diagrams reveal themes in unstructured qualitative information; Is–Is Not analysis narrows the investigative boundary through comparison; and cause-and-effect matrices prioritise inputs against important outputs. None of these tools proves causality by itself. Their outputs are hypotheses that require evidence, testing, and follow-up.

At postgraduate level, effective RCA requires more than knowing how to draw each tool. It requires judgement about tool selection, awareness of bias and power, explicit treatment of uncertainty, integration of technical and organisational factors, and verification that corrective action reduces recurrence. The strongest investigations use several tools in sequence while maintaining a clear distinction between symptoms, contributing conditions, systemic causes, and effective controls.

Key Terms

Root cause analysis: A structured investigation of why an undesired event occurred and how recurrence can be prevented.

Effect: The observable undesirable outcome being investigated.

Immediate cause: The condition directly associated with the occurrence.

Contributing cause: A condition that increases the likelihood or impact of an event.

Systemic cause: An underlying organisational or process condition that permits recurrence.

Brainstorming: A facilitated method for generating possible explanations.

5 Whys: Iterative questioning used to trace a causal pathway.

Cause-and-effect diagram: A visual structure linking a defined effect with categories and possible causes.

Fishbone diagram: Another name for a cause-and-effect diagram.

Affinity diagram: A method for grouping related qualitative observations or ideas.

Is–Is Not analysis: A contrast method that defines where, when, what, and how extensively a problem occurs.

Cause-and-effect matrix: A prioritisation table linking process inputs to important outputs.

Problem statement: A factual description of an undesired condition and its scope.

Causal hypothesis: A proposed explanation requiring evidence or testing.

Verification: The process of checking whether a suspected cause is supported by evidence.

Containment: Immediate action that limits the impact of an existing problem.

Corrective action: Action intended to eliminate a cause or reduce recurrence risk.

Just culture: An approach that promotes learning while distinguishing error from reckless conduct.

Review Questions

Short Answer Questions

1. Why should a root cause be treated as a hypothesis until verified?
2. What is the principal purpose of Is–Is Not analysis?
3. Explain why the number five in the 5 Whys is not a strict requirement.
4. State one strength and one limitation of affinity diagrams.
5. Why should cause-and-effect matrix scores not be interpreted as direct measurements of causality?

Long Answer Questions

1. Compare brainstorming, cause-and-effect diagrams, and affinity diagrams in terms of purpose, process, strengths, and limitations.
2. Explain how a 5 Whys investigation can become misleading when it assumes a single linear cause.
3. Describe a complete RCA workflow integrating at least four classical qualitative tools.
4. Discuss how Is–Is Not analysis can improve the efficiency and validity of evidence collection.
5. Evaluate the ethical risks of using RCA as a mechanism for assigning individual blame.

Analytical / Application Questions

1. A service organisation receives complaints about delayed refunds only on Mondays and only for one payment method. Design an Is–Is Not analysis and identify the next two RCA tools you would use.
2. Construct a cause-and-effect matrix for a process with two outputs—customer safety and delivery accuracy—and three possible inputs. State your weights, rating scale, assumptions, and interpretation.
3. A team concludes after five questions that “insufficient training” caused a recurring assembly defect. Critique this conclusion and propose a more rigorous investigation using a fishbone diagram, evidence checks, and verification steps.

Multiple Choice Questions with Answers

Question 1

Which statement best describes the role of brainstorming in RCA?

A. It proves which cause produced the failure.

B. It generates a broad set of possible explanations.

C. It assigns numerical probability to every cause.

D. It replaces evidence collection.

Correct Answer: B. It generates a broad set of possible explanations.

Question 2

In the 5 Whys method, the number five is best understood as:

A. A mandatory number of questions for every investigation.

B. A statistical confidence requirement.

C. A practical prompt that may be adjusted to the problem.

D. The maximum number of causes that may be recorded.

Correct Answer: C. A practical prompt that may be adjusted to the problem.

Question 3

What is the main purpose of a cause-and-effect diagram?

A. To display possible causes of a defined effect in an organised structure.

B. To calculate process capability.

C. To prove that all listed causes occurred.

D. To rank employees by responsibility.

Correct Answer: A. To display possible causes of a defined effect in an organised structure.

Question 4

Which tool is primarily bottom-up and groups observations according to natural relationships?

A. 5 Whys

B. Affinity diagram

C. Cause-and-effect matrix

D. Control chart

Correct Answer: B. Affinity diagram

Question 5

What does Is–Is Not analysis contribute most directly to an RCA?

A. It defines contrasts that narrow the investigation boundary.

B. It eliminates the need for sampling.

C. It establishes legal responsibility.

D. It automatically identifies the root cause.

Correct Answer: A. It defines contrasts that narrow the investigation boundary.

Question 6

In a cause-and-effect matrix, what does a high priority score generally indicate?

A. That causality has been mathematically proven.

B. That the input deserves earlier investigation under the selected weights and ratings.

C. That the input is independent of all outputs.

D. That the input must be removed immediately.

Correct Answer: B. That the input deserves earlier investigation under the selected weights and ratings.

Question 7

Using Pᵢ = ∑(wⱼ × rᵢⱼ), an input has output weights 4 and 2 and relationship ratings 3 and 9. What is its score?

A. 18

B. 24

C. 30

D. 36

Correct Answer: C. 30

Question 8

Which is the strongest problem statement for an RCA?

A. The operator was careless.

B. The process is poorly managed.

C. Three cartons were incorrectly labelled on Line 2 during the night shift on 14 May.

D. Training caused the failure.

Correct Answer: C. Three cartons were incorrectly labelled on Line 2 during the night shift on 14 May.

Question 9

Why should matrix ratings and weights be documented?

A. They make subjective prioritisation assumptions visible and reviewable.

B. They guarantee that the highest score is the root cause.

C. They remove the need for process knowledge.

D. They convert ordinal ratings into physical measurements.

Correct Answer: A. They make subjective prioritisation assumptions visible and reviewable.

Question 10

Which corrective-action principle is most consistent with the chapter?

A. Always discipline the person closest to the failure.

B. Rely primarily on memory and vigilance.

C. Prefer controls that change the causal condition and verify their effectiveness.

D. Close the investigation once a fishbone diagram is completed.

Correct Answer: C. Prefer controls that change the causal condition and verify their effectiveness.