An I-MR Chart is a pair of control charts used to monitor individual observations and moving ranges when rational subgroups are not available.
Definition
An I-MR Chart combines an Individuals chart and a Moving Range chart. The Individuals chart tracks each observation over time; the Moving Range chart tracks short-term variation between consecutive observations.
It is used when data are collected one observation at a time or when rational subgrouping is not practical. It helps distinguish common-cause variation from special-cause signals.
History
I-MR charts are part of the Shewhart control chart family and became common in Statistical Process Control for low-volume, transactional, batch, or individual-measurement situations.
They remain useful because many real processes do not produce natural subgroups, yet still need time-ordered stability monitoring.
When to Use
Use an I-MR chart for cycle time, order processing time, daily output, single-part measurements, batch results, lab values, downtime duration, or service metrics where individual values arrive sequentially.
Use Xbar-R or other subgroup charts when rational subgroups are available and meaningful.
Step-by-Step
- Define the measure and preserve time order.
- Collect individual observations under consistent definitions.
- Calculate moving ranges between consecutive observations.
- Compute centerlines and control limits for both charts.
- Plot individuals and moving ranges over time.
- Check for points beyond limits, runs, shifts, trends, and unusual moving ranges.
- Investigate special causes using process context.
- Recalculate limits only after a real process change stabilizes.
Examples
- Service: Daily claims cycle time is monitored to detect unusual delays.
- Manufacturing: A low-volume dimension is measured one unit at a time.
- Maintenance: Repair duration is tracked for special-cause spikes.
- Healthcare: Turnaround time for lab results is monitored by completed case.
Common Pitfalls
- Using data out of time order.
- Recalculating limits too often.
- Treating specification limits as control limits.
- Ignoring autocorrelation where consecutive values are strongly related.
- Overreacting to common-cause variation.
- Failing to annotate process changes.