TLDR
Use the X-bar and R chart when your subgroup size is 2 to 9 samples. Use the X-bar and S chart when your subgroup size is 10 or more. Both chart pairs monitor the process average and within-subgroup variation, but the R chart uses the range while the S chart uses the standard deviation, which is more statistically efficient for larger subgroups.
This article explains the difference between these two control chart types, when to use each, the advantages and disadvantages of both, and the sample size rules that drive the decision.
Two Chart Pairs, One Decision
If you are setting up a Statistical Process Control (SPC) program for variable data, you will almost certainly use either an X-bar and R chart or an X-bar and S chart. Both are variable control chart pairs designed to monitor the process average (X-bar) alongside within-subgroup variation. The difference is in how they estimate that variation.
Choosing the wrong chart will not produce catastrophically wrong results, but it will reduce the sensitivity and efficiency of your monitoring. Understanding when each chart is appropriate ensures you get the most reliable signal from your data.
What the X-Bar Chart Does
The X-bar chart is the same in both pairs. It plots the subgroup average (X-bar) over time against control limits calculated from the overall grand average (X-double-bar) and the estimated process variation. The X-bar chart detects shifts and trends in the process mean.
The difference between the two chart pairs is entirely in the companion chart that monitors within-subgroup variation: the R chart or the S chart.
The R Chart: Monitoring Variation with Range
The R chart tracks the range of each subgroup. The range is simply the maximum value minus the minimum value within the subgroup. It is the easiest measure of spread to calculate and the most intuitive to understand.
Control limit formulas for the R chart:
Center Line = R-bar (average of all subgroup ranges)
UCL = D4 x R-bar
LCL = D3 x R-bar
Where D3 and D4 are constants from standard SPC tables that depend on the subgroup size (n).
Strength: Simple to calculate by hand. Easy for operators to understand. Works well for small subgroups where the range captures most of the information about within-subgroup variation.
Weakness: The range only uses two data points (the highest and lowest) regardless of how many samples are in the subgroup. As the subgroup size increases, more and more data points in the middle are ignored. This makes the range a less efficient estimator of the true standard deviation for larger subgroups.
The S Chart: Monitoring Variation with Standard Deviation
The S chart tracks the sample standard deviation of each subgroup. Unlike the range, the standard deviation uses every data point in the subgroup, making it a more complete and statistically efficient measure of variation.
Control limit formulas for the S chart:
Center Line = S-bar (average of all subgroup standard deviations)
UCL = B4 x S-bar
LCL = B3 x S-bar
Where B3 and B4 are constants from standard SPC tables that depend on the subgroup size.
Strength: Uses all data points in the subgroup, making it a more efficient estimator. For subgroups of 10 or more, the standard deviation is significantly better at detecting changes in variation than the range.
Weakness: Slightly more complex to calculate by hand. Less intuitive for operators who are not comfortable with the standard deviation concept. Requires a calculator or software.
The Sample Size Rule
The decision between R and S charts comes down to subgroup size. Here is the rule.
Subgroup size of 2 to 9: Use X-bar and R. For these small subgroups, the range is nearly as efficient as the standard deviation. The simplicity of the R chart outweighs the marginal improvement in efficiency that the S chart would provide.
Subgroup size of 10 or more: Use X-bar and S. For larger subgroups, the range becomes increasingly inefficient. With a subgroup of 15 or 20, the range uses only the two extreme values and ignores 13 or 18 data points. The standard deviation is the better choice.
The threshold of 10 is the widely accepted convention in SPC practice. Some references place it at 8 or 9, but the principle is the same: as subgroup size grows, the efficiency advantage of the standard deviation becomes significant enough to warrant the additional calculation complexity.
Statistical Efficiency: Why It Matters
Statistical efficiency in this context refers to how well the statistic (range or standard deviation) estimates the true population parameter. A more efficient estimator gives you a more accurate picture of the process with the same amount of data.
For a subgroup of 5, the range captures about 95% of the information about the within-subgroup variation that the standard deviation would capture. That is why the R chart works perfectly well at this subgroup size.
For a subgroup of 10, the relative efficiency of the range drops to about 85%. By the time you reach a subgroup of 25, the range captures only about 70% of the information. You are throwing away data. The S chart avoids this loss.
Impact on the X-Bar Chart Control Limits
The choice of companion chart (R or S) also affects how the X-bar chart control limits are calculated.
When using the R chart, the X-bar control limits are: UCL = X-double-bar + A2 x R-bar, and LCL = X-double-bar – A2 x R-bar.
When using the S chart, the X-bar control limits are: UCL = X-double-bar + A3 x S-bar, and LCL = X-double-bar – A3 x S-bar.
The constants A2 and A3 come from different SPC tables and account for the different estimation methods. The resulting control limits will be slightly different between the two approaches, reflecting the different efficiency of the range versus the standard deviation.
Practical Considerations
Most automotive SPC applications use X-bar and R. The most common subgroup size in automotive manufacturing is 5 parts, which is well within the range where the R chart is appropriate. If your sampling plan calls for subgroups of 3 to 5, the X-bar and R chart is the standard and correct choice.
Larger subgroups are less common but do occur. Processes with very high production rates (such as stamping or injection molding) sometimes use larger subgroups. Laboratory testing may also involve larger sample sizes. In these cases, switch to X-bar and S.
SPC software handles both automatically. Most modern SPC software lets you select the chart type and calculates the correct constants and control limits accordingly. The manual calculation difference is only relevant if you are building charts by hand or in a spreadsheet.
Do not mix them. If you start an X-bar and R chart, continue with the R chart. Do not switch to an S chart mid-stream without recalculating all control limits from scratch. The two approaches use different constants and different estimation methods.
Quick Reference Comparison
X-bar and R Chart: Subgroup size 2 to 9. Variation measured by range (max minus min). Uses constants D3, D4, A2. Simple calculation. Most common in automotive SPC.
X-bar and S Chart: Subgroup size 10 or more. Variation measured by sample standard deviation. Uses constants B3, B4, A3. More statistically efficient for large subgroups. Less common in practice but required when subgroup sizes are large.
Frequently Asked Questions
The R chart uses the range (max minus min) to measure within-subgroup variation, while the S chart uses the sample standard deviation. The range only considers two data points; the standard deviation uses all of them. This makes the S chart more efficient for larger subgroups.
The standard threshold is a subgroup size of 10. For subgroups of 2 to 9, use the R chart. For subgroups of 10 or more, use the S chart. Some references use 8 or 9 as the cutoff, but 10 is the most widely cited convention.
Technically, yes. The S chart calculations work for any subgroup size of 2 or more. However, for small subgroups the efficiency gain over the R chart is negligible, and the R chart is simpler. There is no practical reason to use an S chart for subgroups of 5.
The X-bar chart concept is the same, but the control limit formulas use different constants. When paired with an R chart, the X-bar limits use A2 and R-bar. When paired with an S chart, they use A3 and S-bar. The resulting limits will be slightly different.
A subgroup size of 5 is the most common in automotive manufacturing. This is well within the range where the X-bar and R chart is the appropriate and standard choice.
The Individuals and Moving Range (I-MR) chart is used when the subgroup size is 1, meaning you are measuring one part at a time. It is not a variation of X-bar and R or X-bar and S. It is a separate chart type for different sampling conditions, typically low-volume or batch production.
Stay Current on Automotive Quality
PPAP updates, GD&T guidance, and practical quality resources delivered to your inbox. No spam, just what quality professionals need to know.
Subscribe
Leave a Comment