SPC Charts Explained: Which Control Chart Should You Use and When?

Statistical Process Control (SPC) charts are the backbone of modern manufacturing quality management. First developed by Walter Shewhart at Bell Labs in the 1920s, control charts give process engineers and quality teams a window into whether a production process is operating in a state of statistical control or whether something has gone wrong. But with more than a dozen chart types available, knowing which control chart to use is a skill in itself.

This guide cuts through the confusion. Whether you're monitoring continuous measurements from a temperature sensor, counting defects per unit, or tracking the proportion of non-conforming parts, there is a specific chart designed for your data type. Using the wrong chart doesn't just produce misleading signals. It can mask real process problems or trigger false alarms that erode confidence in the entire quality system.

What Is a Control Chart?

A control chart plots a process metric over time and compares each point against statistically derived control limits — typically set at ±3 standard deviations from the process mean. Points inside the limits suggest common-cause variation (normal process noise). Points outside the limits, or non-random patterns within the limits, signal special-cause variation, meaning a data point is drifting and requires investigation.

Every control chart has three horizontal lines:

• Upper Control Limit (UCL): The upper bound of expected variation
• Center Line (CL): The process mean or average
• Lower Control Limit (LCL): The lower bound of expected variation

 

The Two Chart Families: Variables vs. Attributes

All control charts fall into one of two families based on the type of data being measured:

Variables charts measure continuous, numeric data — temperature, pressure, fill weight, tensile strength. These charts are generally more sensitive to process drift and are preferred when measurement is practical.

Attributes charts count discrete events — number of defects, proportion of defective units, number of non-conformances. These are used when measuring a characteristic is impractical or when data is inherently binary (pass/fail).

 

Variables Control Charts

Xbar-R Chart (Average and Range)

Best for: Subgroups of 2–8 measurements from a continuous process.

How it works: The Xbar chart monitors the subgroup mean; the R chart monitors subgroup range. Together they track both process centering and spread. The R chart is simpler to calculate than the standard deviation chart and works well for small subgroup sizes.

Example: Monitoring fill weight on a packaging line by sampling 5 units every 15 minutes.

Xbar-S Chart (Average and Standard Deviation)

Best for: Subgroups of 8 or more measurements.

How it works: Replaces the range with the sample standard deviation (S), which is statistically more efficient for larger subgroups. The Xbar-S chart provides a more accurate picture of process variability when subgroup size is large.

Example: Monitoring tensile strength in a materials lab with 15-sample subgroups.

Individuals and Moving Range (I-MR) Chart

Best for: Processes where only one measurement is collected at a time, or where subgrouping is impractical.

How it works: The Individuals chart plots each single observation; the Moving Range chart plots the absolute difference between consecutive measurements to estimate short-term variation.

Example: Monitoring daily reactor yield, batch viscosity, or chemical purity when only one batch is produced per cycle.

 

Attributes Control Charts

P Chart (Proportion Defective): Monitors the proportion (fraction) of defective units in a variable-size subgroup. Use when subgroup size changes from sample to sample — common in production lines with variable output.

NP Chart (Number Defective): Tracks the actual count of defective units rather than the proportion. Requires a constant subgroup size. Easier to interpret for operators since it shows raw defect counts.

C Chart (Count of Defects): Counts the total number of defects (non-conformances) per inspection unit, where the inspection area is constant. Suitable for products with multiple possible defects per unit — e.g., paint flaws on a panel, weld defects on a pipe section.

U Chart (Defects per Unit): Like the C chart but accommodates variable-size inspection areas. Calculates defects per unit, making it comparable across samples of different sizes. Common in web or continuous process industries.

 

Quick Reference: Choosing Your Chart

Data Type	Subgroup Size	Chart	Monitors
Continuous	2–8	Xbar-R	Mean & range
Continuous	8+	Xbar-S	Mean & std dev
Continuous	1 per period	I-MR	Individual values & MR
Attribute (proportion)	Variable	P	Fraction defective
Attribute (count)	Constant	NP	Number defective
Attribute (defects)	Constant area	C	Defect count
Attribute (defects)	Variable area	U	Defects per unit

 

Common Mistakes When Selecting Control Charts

• Using an Xbar-R chart with subgroup sizes greater than 8, which underestimates process variability
• Applying a P chart when subgroup sizes are constant — the NP chart is simpler and equally valid
• Using the I-MR chart when consecutive measurements are autocorrelated (e.g., in continuous flow processes), which inflates the moving range
• Charting both a variables and an attributes chart on the same characteristic — choose one
• Ignoring rational subgrouping: subgroups should represent conditions that are as homogeneous as possible within each subgroup

 

Getting chart selection right is foundational to any SPC program. Northwest Analytics' NWA Quality Analyst® includes built-in chart selection guidance and automatic control limit calculation, helping engineering teams implement the right chart from day one without needing a statistician on every shift. Download a free 30-day trial of NWA Quality Analyst below to learn more.