Identify the factors linked to process variation

Statistical analysis helps you separate evidence from opinion, so teams can focus their time on the factors most likely to matter.

Use a balanced ANOVA to examine how time, operator, and machine setting relate to coating thickness. You’ll fit a general linear model, check its residuals, and interpret the results to decide what to investigate next.

Move from observing variation to knowing where to focus your next investigation.

In 20 minutes, you'll:

• Define the response and factor variables for your analysis

• Fit a general linear model and determine which factors are statistically significant

• Check residual plots and turn the findings into a practical next step

MSC Trial - Statistical Analysis

USE CASE

Compare process performance and identify factors linked to variation.

DOWNLOAD DATASET

Download the Minitab worksheet dataset used in this lesson here.

DURATION

20 minutes

WHAT YOU'LL NEED

A logged-in Minitab Solution Center trial.

Identify significant process factors

Follow five guided steps in Minitab to fit, check, and interpret a model using manufacturing data.

Add Data

Open Minitab Statistical Software and open a data connection

First, download and save CoatingThickness.MWX to your organization’s cloud storage. 

From the Minitab Solution Center home page, open Minitab Statistical Software. From the left-hand navigation, select Open and select the CoatingThickness file from your saved location. 

Verify that the columns show Thickness, Time, Operator and Setting. Hit Load, and now you’re ready to start finding key factors. 

Navigate to the balanced ANOVA model

Navigate to the balanced ANOVA model

Go to the top ribbon, and navigate to Stat > ANOVA > Balanced Anova. We’re setting up to test which factors explain the output thickness you’re getting—which is crucial for quality improvement. 

Set up the ANOVA model

Fill in the model

In the Balanced Analysis of Variance dialog, tell Minitab what you want to explain and what might be causing the difference. Set it up as follows:

Responses
Click inside the Responses input box, select Thickness from the list of variables and hit Select.

This is the measurement you are trying to understand.

Model
Paste the following list into the Model input:

Time
Operator
Setting
Time*Operator
Time*Setting
Operator*Setting

 

This tells Minitab to test each factor by itself, plus whether factors work differently together.

Random factors 
Click inside the Random factors input box, select Operator from the list of variables and hit Select.

The operators are treated as a sample of possible operators, not the only people who could run the process.

Don't hit OK just yet. We'll keep this dialog open for the next step.

Check your residuals

Add in residual plots

Before hitting OK, click Graphs then select Four-in-one. 

Click OK to close the Graphs dialog, then OK again to close the Balanced ANOVA dialog.

Scroll through the top section of your screen and you'll see a histogram and a normal plot of your residuals, as well as graphs of residuals vs. fit and residuals vs. order.

  • Normal Probability Plot: The dots should mostly follow the line.
  • Histogram: The bars should be centered around zero.
  • Residuals vs. Fits: dots should not form a curve or funnel.
  • Residuals vs. Order: dots should not show a steady pattern over time.

 

Interpret your results

Interpret your results

ai-icon-1Minitab AI

You can use Minitab AI to summarize statistical output, highlight key findings, and explain results in clear, accessible language.

On the top right of your CoatingThickness canvas, click ai-icon-1to generate a Minitab AI summary.

 

Scrolling back up, the Factor Information table lays out what Minitab tested. In this case, you compared Thickness across 2 time periods, 3 operators, and 3 machine settings.

Then review the P column in the ANOVA table; values below 0.05 suggest a factor contributes meaningfully to variation in coating thickness.

In our dataset, Setting is the clearest driver of thickness, while Time does not appear to matter by itself. Significant interaction terms indicate that the effect of one factor depends on another. For example, the effect of Setting may differ by Operator.

Now you know which factors and interactions appear most important for explaining variation in coating thickness!

 

Recap what you learned

You’ve seen how balanced ANOVA can clarify which factors may be associated with process variation. You can use the same approach to explore other process problems, understand how factors work together, and narrow down where to look next. This helps your team spend less time following hunches and more time making focused, evidence-based improvements to quality and consistency.