Advanced Process Capability Analysis with Minitab

By the end of this lesson, participants will be able to:

• Understand the difference between nominal-scaled quality data (e.g., "good" vs. "bad") and continuous data, and recognize its implications for statistical analysis.

• Correctly classify quality characteristics using the concept of scale levels and identify when binomial distribution is the appropriate model.

• Import and explore real-world production data in Minitab, including daily assembly volumes and defective unit counts.

• Calculate and interpret the daily defect rate as a key quality indicator.

• Recognize that binomially distributed data require specific stability and capability analysis methods, distinct from those used for normally distributed variables.

• Prepare the groundwork for applying p-charts and binomial capability analysis in Minitab by ensuring proper understanding of the data structure and distribution type.

This foundational unit enables participants to correctly assess whether their categorical quality data meet the prerequisites for process capability analysis—and prepares them to apply Minitab tools tailored to binomial data.

Validating Process Stability for Binomial Data Using Minitab

By the end of this section, participants will be able to systematically assess the process stability of binomially distributed quality data using Minitab’s p-chart and the cumulative % defective plot.

They will gain a deep understanding of how p-charts operate in contexts with variable subgroup sizes, and why p-charts—unlike np-charts—are the correct control chart type for evaluating proportions when daily production volumes fluctuate.

In Minitab, learners will:

• Read and interpret p-charts for defect rates, including mean lines and control limits

• Understand the statistical basis for variable upper and lower control limits depending on subgroup size

• Detect signs of process instability using Minitab’s four binomial control tests

• Identify and interpret red control chart signals and corresponding test numbers

• Navigate and analyze the “cumulative % defective” plot to monitor stabilization over time

• Calculate cumulative defect rates step-by-step based on aggregated subgroup data

• Recognize that a flattening trend in cumulative defect rate indicates a stable process mean

• Understand how random scatter within control limits validates statistical control

Participants will also learn why a process capability analysis is invalid unless the underlying process is stable, and how Minitab supports this distinction visually and analytically through multiple diagnostic outputs.

By practicing this structured stability validation in Minitab, learners will build both conceptual and operational confidence to judge whether a process is fit for capability analysis—and if not, how to diagnose and improve it.

Learning Outcome (Extended – Minitab Focused):

By the end of this section, participants will be able to critically verify whether a process with attribute data truly follows the binomial distribution, using graphical diagnostic tools available in Minitab. This is a crucial step to ensure that subsequent process capability analyses are statistically valid and reliable.

In practical terms, learners will:

• Understand why confirming process stability is not sufficient on its own

• Analyze the “Rate of Defectives” plot in Minitab to check for random scatter of defect rates across irregular subgroup sizes

• Recognize that if defect rates show no systematic trend and scatter randomly around the mean, this supports the assumption of binomial distribution

• Detect visual patterns or trends that may suggest non-binomial behavior, such as correlation between defect rate and subgroup size

• Learn that non-random trends invalidate binomial assumptions and may indicate hidden causes or influencing factors in the process

Participants will also understand when Minitab provides the “Binomial Plot” instead of the rate plot—namely, in cases where the subgroup size is constant.

In the binomial plot, they will:

• Compare observed defect counts with the expected defect counts under the binomial model

• Use the diagonal reference line to assess the goodness of fit between observed and theoretical values

• Understand that alignment with the red line confirms the applicability of binomial distribution, while significant deviations raise red flags

• Know that if the data clearly follow the binomial pattern, they can proceed with the capability analysis confidently

Through this section, participants gain the statistical reasoning and technical ability to validate model assumptions visually and analytically. They will understand that Minitab’s plots serve as a diagnostic checkpoint and that skipping this step could lead to misleading capability conclusions.

By mastering these Minitab tools, learners will reinforce their ability to ensure statistical validity in capability studies for attribute data—and make data-driven quality decisions with greater confidence.

Interpreting Summary Statistics and Capability Indicators in Minitab

Learning Outcome (Extended – Minitab Focused):

At the end of this section, participants will be able to confidently interpret the summary statistics and key performance indicators of a binomial process capability analysis in Minitab, linking statistical results to practical quality decisions.

They will learn how to:

• Use histograms of defect rates to visually assess the distribution of defects over time

• Identify the most frequent defect rate intervals and the overall spread of the data

• Understand that the visual shape of the histogram begins to resemble a normal distribution with large sample sizes, in line with the law of large numbers

• Relate the histogram to theoretical principles, such as the Laplace theorem and the binomial-to-normal convergence

In the summary output of Minitab, participants will:

• Identify and interpret the observed mean defect rate (e.g., 36.53%)

• Understand the use of confidence intervals (e.g., 36.31% to 36.75%) and what they imply about the expected process behavior in the general population

• Quantify the expected number of defective units in large-scale production by calculating PPM (parts per million)

• Learn how Minitab derives PPM values based on defect proportions and confidence intervals

• Recognize that a result far above the target (e.g., 365,000 bad parts per million) indicates a critically incapable process

Participants will also be able to interpret the Z benchmark, Minitab’s key sigma-level indicator, and understand how it applies to binomial distributions just as it does to normally distributed data. They will learn how the Z-value reflects the distance of the process performance from the desired quality threshold, and how this relates to Six Sigma standards.

By completing this section, learners will not only understand the numerical outputs provided by Minitab, but they will also be equipped to communicate their meaning effectively within a quality management context—translating statistical results into clear action for process improvement.

Process Capability Analysis for Count Data Using the Poisson Distribution in Minitab

By the end of this lesson, participants will be able to confidently conduct a process capability analysis for discrete defect data, specifically where the number of defects per unit or per subgroup is modeled using the Poisson distribution. This type of analysis is essential when the interest is not in classifying products as “good” or “bad,” but in counting how many defects (e.g., scratches) occur within a sample.

Participants will understand the transition from binomial to Poisson modeling, and why the binomial distribution no longer applies when the variable of interest is the number of defects per unit or subgroup—rather than the number of defective items.

In Minitab, learners will:

• Import a real-world data set that contains subgroup-level scratch counts from automatic surface inspection

• Use the Worksheet Information panel to explore the dataset structure and interpret the meaning of each column

• Distinguish between subgroup size and total defect count across samples

• Use Descriptive Statistics in Minitab to identify key location and spread parameters, such as the mean, minimum, and maximum scratch counts

• Understand that the scratch counts per subgroup are summed values across 300 skateboards, not individual product values

• Interpret whether the observed data might follow a Poisson distribution, based on domain knowledge and data type

• Learn that, like with binomial data, Minitab allows users to combine distribution check, stability analysis, and capability analysis in a single, streamlined procedure

They will also grasp the conceptual differences between binomial and Poisson quality data:

• Binomial → Good/bad decision per unit

• Poisson → Number of events (defects) per sample or unit

By understanding this distinction and applying the correct tools in Minitab, participants will:

• Be equipped to move beyond binary defect classification and enter the realm of discrete defect frequency analysis

• Lay the foundation for applying U-charts, Poisson plots, and Poisson capability analysis in subsequent sections

• Be able to evaluate and communicate whether their assembly process is stable and capable based on the average number of defects, not just the presence of defective products

Ultimately, this lesson empowers participants to correctly apply Poisson-based tools in Minitab, interpret key process indicators for count data, and prepare for deeper statistical analysis related to defect minimization and process optimization.