Are you overreacting to normal variation?

Have you ever seen a measurement result and immediately thought something was wrong?

Maybe the scale showed one kilogram more than expected, or production data indicated a slight increase in defect rates? But what if this was just normal variation – what happens if you start making adjustments?

When and how should you react to measurement results?

Do you rely on facts or intuition when interpreting measurement changes? Do you let random fluctuations dictate the adjustments you make in production?

All processes have natural fluctuations, and it is important to distinguish between normal variation and special variation before taking action.

Everyday example: Weight measurements

Let’s say you weigh yourself a few weeks apart, and this time the scale shows 1 kg more than last time. Have you actually gained weight, or is this just natural variation?

To find out, we need to ask a few important questions:

• What is the measurement uncertainty of the scale?
• Were the measurements taken at the same time of day? (e.g., before or after a meal?)
• Were you wearing the same clothes, or could this affect the weight?
• Is a 1 kg difference a real change, or is it within normal daily variation?

To answer these questions, we can conduct a Component of Variation (CoV).

Component of Variation (CoV) – What is normal weight variation?

To understand weight variation, I conducted a CoV analysis over three weeks. I measured three times a day, taking three readings each time. This resulted in a total of 189 measurements, as illustrated below:

How do we analyze these measurements? Control charts help us determine expected variation.

Example 1: Measurement uncertainty

To quantify how much variation is caused by measurement differences, we can use a Control Chart with subgroups (subgroup size = 3, since we have three parallel measurements). In this case, the control limits represent expected variation for three parallel measurements, i.e., measurement uncertainty.

Interpretation:

• Control limits (UCL and LCL) are calculated based on variation in the parallel measurements.
• The distance from the average to the control limits is ± 0.17 kg.
• This represents measurement uncertainty – meaning how much the weight can fluctuate even if actual body weight has not changed.

Example 2: Daily variation

To see how weight fluctuates throughout the day, we created another control chart using the average of the nine daily measurements:

Interpretation:

• Control limits are set based on variation in the daily measurements.
• The distance from the average to the control limits is ± 1.35 kg.
• This represents normal daily variation – meaning natural fluctuations within a day.

All daily averages are within the control limits. This means that variation within a single day can explain differences between days. If we only compare two measurements without considering this variation, we might misinterpret changes.

🔎 What does this mean?

• A 1 kg weight change does not necessarily mean you have gained weight.
• Variation within a day is normal, and it’s important to look at trends over time.
• Overreacting to small changes can lead to poor decisions – both in everyday life and in production.

This is a great example of how control charts can help us interpret variation and avoid incorrect conclusions. In production, misinterpreting variation can lead to unnecessary adjustments that make processes unstable.

Want to learn more about using control charts to understand variation?

🚀 Take the next step with our online SPC course!

• Learn to interpret SPC data and use control charts correctly
• Avoid incorrect adjustments and ensure stable processes
• Improve quality and reduce costs

Enroll in the SPC course