In Statistical Process Control (SPC), data collection is not just about measuring parts, it is about collecting data in a way that correctly shows process variation.
One of the most important decisions during SPC implementation is defining the subgroup size.

The concept of a subgroup directly affects how accurately we can separate common cause variation from special cause variation, which is the core objective of SPC.

What Is a Subgroup in SPC? #

A subgroup in SPC is a set of multiple observations (measurements) collected close together in time, under similar process conditions.

In simple words:

A subgroup represents a snapshot of the process at a specific moment.

We group these observations together to estimate short-term (within-subgroup) variation, and then use it to understand overall process behavior.

Why Is Subgroup Important in SPC? #

Subgrouping is critical because it helps us:

• Estimate process variation correctly
• Identify special causes quickly
• Build accurate control charts (X̄-R, X̄-S, etc.)
• Avoid misleading conclusions from random data

If we do subgrouping incorrectly, a stable process may appear unstable, or we may fail to detect real problems.

Definition of Subgroup Size #

Subgroup size is the number of observations collected in one subgroup.

Common subgroup sizes in manufacturing:

• 2, 3, 4, or 5 → very common
• 10 or more → used when process variation is very small or data is easily available

The choice of subgroup size depends on:

• Process speed
• Cost of inspection
• Nature of variation
• Type of control chart used

Example 1: Subgroup in Shaft Diameter Measurement #

Consider an automotive parts manufacturing company producing shafts.

• The inspector measures the diameter of shafts
• Every 2 hours, 5 shafts are selected from the production line
• All 5 measurements are taken almost at the same time

Interpretation:

• These 5 measurements are considered as one subgroup
• Subgroup size = 5
• This subgroup represents the process condition at that moment

This data is typically used to create an X̄-R control chart, where:

• X̄ tracks the subgroup average
• R tracks the within-subgroup variation

Example 2: Subgroup in Oil Can Weight Measurement #

Now consider an oil canning (packaging) process:

• The inspector measures the weight of oil cans.
• Then samples 10 cans every 30 minutes.
• The machine fills all 10 cans using the same settings.

Interpretation:

• These 10 readings are one subgroup
• Subgroup size = 10
• Used to estimate short-term filling variation

This is useful when:

• Filling machines are fast
• Weight variation is very small
• High sensitivity to variation is required

Rational Subgrouping (Missing but Critical Concept) #

A rational subgroup is formed so that:

• Variation within the subgroup represents only common causes
• Variation between subgroups captures special causes

Good Rational Subgrouping:

• Parts made one after another
• Same machine, operator, material, and settings

Poor Subgrouping:

• Mixing parts from different shifts
• Mixing different machines in one subgroup
• Large time gaps within a subgroup

Incorrect subgrouping hides real process problems.

Relationship Between Subgroup and Control Charts #

If subgroup size is 1, variation cannot be estimated within the subgroup, so I-MR charts are used instead.

Key Points to Remember for subgroup #

• A subgroup is multiple measurements taken close in time
• Subgroup size directly affects SPC accuracy
• Always aim for rational subgrouping
• Wrong subgrouping = wrong decisions
• Choose subgroup size based on process behavior, not convenience

Conclusion #

Understanding what a subgroup is in SPC is fundamental for every quality engineer.
Subgrouping is not just a statistical requirement, it is a process understanding exercise.

When you select the subgroup size and sampling strategy correctly, SPC becomes a powerful tool to control variation, prevent defects, and improve process stability.