What is AAA

Guide: Attribute Agreement Analysis (AAA)

Author's Avatar

Author: Daniel Croft

Daniel Croft is an experienced continuous improvement manager with a Lean Six Sigma Black Belt and a Bachelor's degree in Business Management. With more than ten years of experience applying his skills across various industries, Daniel specializes in optimizing processes and improving efficiency. His approach combines practical experience with a deep understanding of business fundamentals to drive meaningful change.

Guide: Attribute Agreement Analysis (AAA)

Understanding the reliability of your measurement system is crucial for effective process management, especially in the context of Lean Six Sigma. Attribute Agreement Analysis (AAA) serves as an invaluable tool for this purpose.

This guide aims to provide a detailed insight into AAA, covering its definition, importance, and methodology. We’ll explore how to assess the reliability of attribute (categorical) data by measuring the agreement between multiple appraisers’ ratings against a known standard. By the end, you’ll be equipped to conduct your own AAA, ensuring the robustness of your data and enhancing your continuous improvement efforts.

What is Attribute Agreement Analysis?

Definition and Purpose

Attribute Agreement Analysis  is a statistical technique used to evaluate the agreement among different appraisers’ judgments on categorical data. It assesses whether multiple individuals making judgments or assessments on the same item would reach a high level of agreement by evaluating the repeatability, reproducibility, and overall accuracy of the appraisers. Repeatability refers to the variation in assessments when repeated by the same appraiser, while reproducibility refers to variation when assessments are made by different appraisers on the same item.

Why is Attribute Agreement Analysis Important?

Benefits and Applications

AAA is pivotal for ensuring the quality and reliability of data in decision-making processes. It helps to:

Characterize the quality of data by identifying areas of non-agreement.Appraiser Agreement2

Calibrate appraisers, judges, or assessors for a higher level of agreement.

Appraiser Agreement3

Enhance the consistency and accuracy of judgments, thereby contributing to better decision-making and operational efficiency. This technique finds applications across various sectors, especially in quality control, manufacturing, and any domain where accurate categorization and assessment are crucial for operational success.

Types of Attribute Data: Nominal and Ordinal

In the context of Lean Six Sigma and continuous improvement, understanding different types of data is crucial for quality control and decision-making. Attribute data is a type of qualitative data that is categorized into labels or attributes, rather than numerical values. Attribute data is primarily divided into two types: Nominal and Ordinal.

Nominal Data

Nominal data consists of categories that have no inherent or meaningful order. These categories are mutually exclusive and exhaustive, meaning that every observation can only belong to one category, and all possibilities are accounted for.

Examples in Continuous Improvement:

  • Inspection results such as Pass/Fail
  • Types of defects like Scratch, Dent, or Crack
  • Employee departments like Manufacturing, Logistics, and Sales

Graphical Representation:

In most cases, a bar chart or pie chart is used to represent nominal data. Each category gets its own bar or pie slice, and the height or size of the slice represents the frequency or percentage of observations in that category.

Ordinal Data

Ordinal data is similar to nominal data, but the categories have a meaningful order. However, the intervals between these ordered categories are not uniform or measurable.

Examples in Continuous Improvement:

  • Customer satisfaction ratings from 1 to 5
  • Severity level of defects: Low, Medium, High
  • Skill level of workers: Novice, Intermediate, Expert

Graphical Representation:

Ordinal data can also be represented using bar charts, but it is important to maintain the order of the categories. Sometimes, a line chart may also be useful to highlight trends across ordered categories.

Components Involved in Attribute Agreement Analysis (AAA)

Attribute Agreement Analysis (AAA) is a statistical method used to assess the reliability and agreement among different appraisers. This is particularly useful in quality control processes where multiple individuals may be involved in assessing a product or service. Here are the key components:


The appraiser is the individual responsible for making the assessment or measurement. Their role is to apply the criteria or standard to the item being assessed.

Expert (Optional)

An expert is the Subject Matter Expert (SME) who sets the standard or criterion against which appraisers’ assessments are compared. While not always necessary, having an expert can lend more credibility and consistency to the analysis.


Repeatability refers to the variation that occurs when the same item is assessed multiple times by the same appraiser. High repeatability indicates that the appraiser is consistent in their assessments.


Reproducibility, on the other hand, measures the variation when the same item is assessed by different appraisers. High reproducibility indicates that there is agreement among different appraisers, which is ideal for quality control.

Graphical Representation:

A common way to visually represent AAA results is through the use of scatter plots or control charts, which can show the range of variation and help identify outliers or trends.

Step 1: Define the Objective

Defining the objective is the foundation upon which the entire AAA is built. Without a clear objective, the analysis can become aimless and may not provide meaningful insights.

Detailed Steps:

  1. Identify the Attribute: Determine what specific attribute you will be assessing. Is it the quality of a product, customer service ratings, or perhaps the accuracy of a machine?

  2. Set the Criteria: Clearly outline the criteria that will be used to assess the attribute. For instance, if you’re measuring product quality, the criteria might be Pass/Fail, or it could be various types of defects like scratches, dents, etc.

  3. Justify the Need: Explain why this attribute is critical to measure. Does it relate to customer satisfaction, operational efficiency, or compliance with regulations?

  4. Define the Scope: Make sure to outline what is in and out of scope for the AAA. This helps to keep the analysis focused.

  5. Documentation: Document the objective, attribute, and criteria so that everyone involved understands what is being measured and why.

Example Objective:

“To evaluate the consistency and reliability of quality control inspectors in identifying and classifying defects in the manufactured widgets. The criteria for assessment will be categorizing defects as ‘Scratch’, ‘Dent’, or ‘Crack’.”

Step 2: Select the Appraisers

The appraisers are the individuals who will be conducting the assessments. Their reliability directly impacts the validity of the AAA.

Detailed Steps:

  1. Identify Potential Appraisers: List the individuals who are typically involved in these kinds of assessments in their day-to-day roles.

  2. Consider Experience: Consider the experience level and expertise of potential appraisers. A mix of novices and experts can provide a broader perspective.

  3. Commitment: Ensure that the selected appraisers are committed to participating fully in the AAA process, including possible retraining based on the results.

  4. Briefing: Brief the selected appraisers on the objective, criteria, and importance of the AAA.

  5. Documentation: Document who has been chosen and why they are considered appropriate for this analysis.

Example Appraisers:

Quality control inspectors who regularly check the widgets for defects.

Step 3: Identify an Expert (Optional)

An expert serves as the gold standard against which other appraisers are compared. Though optional, having an expert adds a level of rigor to the AAA.

Detailed Steps:

  1. Expert Identification: Choose a Subject Matter Expert (SME) who has significant experience and expertise in the attribute being measured.

  2. Confirm Availability: Make sure the expert is available for the duration of the AAA.

  3. Training: The expert should ideally also participate in training sessions, both to refresh their own understanding of the criteria and to provide guidance to the appraisers.

  4. Documentation: Document the expert’s credentials and role in the AAA.

Example Expert:

A senior quality control manager with over 10 years of experience in widget manufacturing.

Step 4: Prepare the Samples

Selecting a representative sample is crucial for the validity of the AAA. The samples you choose must adequately capture the variation in the attribute you are studying.

Detailed Steps:

  1. Define Population: Clearly state the population that the samples will be drawn from. For example, are you looking at all widgets produced in a month or just a specific type?

  2. Determine Sample Size: Choose a sample size that is statistically meaningful but also manageable for the appraisers. The larger the sample, the more reliable your results will be, but it should also be realistic in terms of time and resources.

  3. Select Samples: Actually draw the samples from your defined population. Make sure the selection is random and unbiased.

  4. Document: Record details of each sample and the methodology used for selection. This is important for traceability and for any future replication of the AAA.


For an AAA on widget defects, you might choose 50 widgets randomly from a month’s production to cover the variety of defects you are interested in.

Step 5: Randomize the Samples

Randomization eliminates order bias, ensuring that the results are not influenced by the sequence in which samples are assessed.

Detailed Steps:

  1. Shuffle Order: Use a random number generator or a similar method to randomize the order of the samples.

  2. Assign Codes: Optionally, you can assign codes to the samples to anonymize them, making it even less likely that appraisers will be biased.

  3. Document: Document the randomized order for auditability and future reference.


Use a software tool to randomize the order of the 50 widgets. Each widget could be labeled with a code like W1, W2, etc., and then shuffled.

Step 6: Train the Appraisers

Training ensures that all appraisers have a uniform understanding of the criteria and process, which is crucial for the reliability of the AAA.

Detailed Steps:

  1. Develop Training Material: Create or gather material that clearly explains the criteria for assessment.

  2. Conduct Training Session: Hold a session where appraisers can learn the criteria and ask questions.

  3. Test Understanding: Give appraisers a few test samples to ensure they have understood the criteria properly.

  4. Document: Record details of the training session and the attendees.


Hold a 1-hour training session with all quality inspectors, using presentation slides and example widgets to clarify the types of defects.

Step 7: Conduct the First Round of Assessments

The first round provides the initial data that will be used to evaluate repeatability and reproducibility.

Detailed Steps:

  1. Assign Samples: Give each appraiser their set of randomized samples.

  2. Record Assessments: As appraisers complete their assessments, record their judgments in a structured format like a spreadsheet.

  3. Ensure Independence: Make sure appraisers are not discussing their assessments with each other to maintain independence.

  4. Document: Keep detailed records of all assessments.


Each quality inspector assesses the 50 randomized widgets independently and records whether they find each widget to have a Scratch, Dent, or Crack.

Step 8: Conduct the Second Round (For Repeatability)

The second round is essential for assessing how consistent each appraiser is when making repeated assessments.

Detailed Steps:

  1. Time Interval: Wait for a specific time interval to reduce memory bias. This could range from a few hours to a few days, depending on the context.

  2. Repeat Assessment: Have appraisers reassess the same samples in a new randomized order.

  3. Record and Document: As in the first round, record all assessments in a structured format and keep detailed documentation.


A day after the first round, the same quality inspectors reassess the same 50 widgets, which have been re-randomized, and record their new assessments.

Step 9: Analyze Repeatability

Analyzing repeatability helps to gauge how consistent each appraiser is when making assessments. The aim is to understand if the same appraiser is likely to give the same rating for the same sample across different instances.

Detailed Steps:

  1. Choose Statistical Methods: Decide on the statistical methods you will use, such as percent agreement, Kappa statistics, or control charts.

  2. Compute Metrics: Use the recorded data to compute the chosen metrics for each appraiser.

  3. Visualize Data: A control chart or similar graphical tool can help visualize the repeatability across different appraisers.

  4. Document: Record all calculations, assumptions, and findings for future reference and auditing.


Calculate the Kappa statistic for each appraiser to quantify the level of agreement between their first and second rounds of assessments.

Step 10: Analyze Reproducibility

Reproducibility analysis helps in understanding how different appraisers agree with each other and possibly with an expert, giving insights into the overall reliability of the measurement system.

Detailed Steps:

  1. Compute Cross-Appraiser Metrics: Use the recorded data to calculate metrics like percent agreement or Kappa statistics between different appraisers.

  2. Expert Comparison: If an expert is involved, compare each appraiser’s assessments with the expert’s assessments.

  3. Visual Representation: A scatter plot or similar graphical method can help visualize the degree of agreement between different appraisers.

  4. Document: As always, document all calculations and findings.


Calculate the percent agreement between each pair of appraisers and between the appraisers and the expert, if available.

Step 11: Interpret the Results

Interpreting the results helps in understanding what the statistical metrics imply about the reliability and validity of the measurement system.

Detailed Steps:

  1. Evaluate Repeatability: Assess whether individual appraisers are consistent in their assessments.

  2. Assess Reproducibility: Determine if there is a general agreement between different appraisers.

  3. Overall Reliability: Combine these analyses to get an overall picture of the measurement system’s reliability.

  4. Document: Prepare a detailed report or presentation that interprets these findings.


If Kappa values are below 0.6, consider the level of agreement as poor and the measurement system as potentially unreliable.

Step 12: Take Corrective Actions

Based on the interpretation, corrective actions may be needed to improve the measurement system.

Detailed Steps:

  1. Identify Weak Areas: Pinpoint where the issues lie, whether they’re in repeatability, reproducibility, or both.

  2. Plan for Improvement: Develop a plan that could include retraining, clarifying guidelines, or even revisiting the criteria.

  3. Implement Changes: Execute the improvement plan.

  4. Document: Keep a record of the corrective actions taken for audit trails and future reference.


If the issue lies in repeatability, consider retraining appraisers using more detailed guidelines.


A: Attribute Agreement Analysis (AAA) is a statistical method used to assess the level of agreement among multiple observers or raters when evaluating categorical attributes. It is commonly applied in quality control, market research, and various scientific fields.

A: To define the attribute, start by clearly describing the nature of the attribute and its possible levels or characteristics. For example, if the attribute is color, specify the color categories such as red, blue, green, etc.

A: The rating scale should have discrete categories that represent different levels or characteristics of the attribute. Ensure that the scale is comprehensive and covers all possible levels. For example, a color attribute might have categories like red, blue, green, etc.

A: Choose observers who have the necessary expertise or knowledge to evaluate the attribute accurately. Aim for a diverse group of observers to ensure a representative assessment.

A: Provide each observer with a set of samples or items representing the attribute. Randomize the presentation of items to avoid bias. Ask observers to independently rate or classify each item according to the established rating scale.

A: Agreement metrics are quantitative measures used to assess the level of agreement among observers. Common metrics include percent agreement, Cohen’s Kappa, and Fleiss’ Kappa.

A: Interpret the agreement metrics based on established guidelines or benchmarks specific to your field. Higher values indicate higher agreement. Refer to industry standards or previous research to determine if the agreement level is acceptable or needs improvement.

A: Investigate the reasons behind the discrepancies. Check for sources of bias, inconsistencies in the rating scale, or other factors that may have influenced the results. This investigation will help identify areas for improvement.

A: Yes, if the level of agreement is not satisfactory or if you need to evaluate the attribute again under different conditions, it is recommended to repeat the entire AAA process. This allows for tracking improvements and enhancing the reliability of attribute evaluations.


Picture of Daniel Croft

Daniel Croft

Daniel Croft is a seasoned continuous improvement manager with a Black Belt in Lean Six Sigma. With over 10 years of real-world application experience across diverse sectors, Daniel has a passion for optimizing processes and fostering a culture of efficiency. He's not just a practitioner but also an avid learner, constantly seeking to expand his knowledge. Outside of his professional life, Daniel has a keen Investing, statistics and knowledge-sharing, which led him to create the website www.learnleansigma.com, a platform dedicated to Lean Six Sigma and process improvement insights.

All Posts

Free Lean Six Sigma Templates

Improve your Lean Six Sigma projects with our free templates. They're designed to make implementation and management easier, helping you achieve better results.


Was this helpful?

Thanks for your feedback!