A control chart method for ordinal data

Every day, professionals use control charts to increase the quality of goods and services. These charts help by signaling when special causes of process variation occur so that they are investigated. An influence identified as changing process variability, i.e., an assignable cause, is removed or made a permanent part of the process depending on which action improves quality. To monitor a process, quality characteristic measurements form control chart statistics that indicate when some parameter of the measurements' distribution changes. Sometimes a process is effectively monitored using ordinal data, which can be a cheaper alternative to measurement of a continuous variable. Items are graded and simply counted. We describe a robust control chart for ordinal data and demonstrate its advantages over a commonly used approach for categorical data.