Meaning
Statistical process control relies on the assumption that the data collected within a single subgroup represents a stable and homogeneous process state. An artificial increase in this within-subgroup variation, termed subgroup variance inflation, occurs when the sampling method groups heterogeneous items together or when the process experiences rapid shifts during the sampling interval. This inflation masks the true capability of the process and distorts the control chart limits.
It ceases to be an issue when the subgroup size is set to one, as in individual measurement charts.
Sampling Error
Incorrect subgrouping strategies often mix parts from multiple parallel production streams or different cavities of a single mold. This mixture increases the calculated standard deviation of the subgroup, which broadens the control limits on the range chart. Consequently, the control chart becomes less sensitive to actual process shifts.
Control Sensitivity
When the calculated within-subgroup variation is inflated, the control limits on the average chart become too wide. This loss of sensitivity prevents the detection of significant changes in the process mean. Operators are then unable to intervene before the process produces defective parts.
Diagnostic Strategy
Analyzing the individual measurements within the subgroup can reveal the presence of systematic patterns or distinct distributions. Engineers can then adjust the sampling plan to ensure that each subgroup represents a single, short-duration operating state. This adjustment restores the diagnostic power of the control charts.
