A Runs Test Uses Sequence Changes, Not the Magnitude of Each Reading

The statistical question

For a median-based runs analysis, readings are classified above or below a stated reference, and a run is a consecutive stretch with the same label. The number of runs describes sequence structure rather than the numerical size of deviations. The handling of ties and the reference distribution for the actual label counts must be identified. A result does not name the physical cause of a pattern.

Primary reference: NIST Engineering Statistics Handbook.

An illustrative review example

Consider a fictional ordered label sequence plus,plus,minus,minus,plus,minus. It contains four runs: a two-plus stretch, a two-minus stretch, one plus and one minus. Sorting by sign would leave only two runs and destroy the original temporal evidence. Six labels are used here only to demonstrate counting; this illustration does not justify a large-sample normal approximation or a significance conclusion.

Prepare the evidence for discussion

Retain the time or item sequence, labeling reference, ties and omitted readings. Ask the analyst to select the reference calculation appropriate to the positive and negative counts rather than copying a generic cutoff. Relate an observed pattern to recorded changeovers, replenishment or collection timing only through additional evidence. For a filling-trial file, the ordering used in analysis should be reproducible from the preserved raw record.

Runs preserve order rather than magnitude

This blank worksheet is for your own project. It contains no H M machine trial result.

Customer Questions

Primary References

These references support the technical principles discussed in this guide. The worked examples and review questions are educational.

• Primary reference: NIST Engineering Statistics Handbook