Talk:Nelson rules

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Statistics (Rated Start-class, Low-importance)
WikiProject icon

This article is within the scope of the WikiProject Statistics, a collaborative effort to improve the coverage of statistics on Wikipedia. If you would like to participate, please visit the project page or join the discussion.

Start-Class article Start  This article has been rated as Start-Class on the quality scale.
 Low  This article has been rated as Low-importance on the importance scale.

I couldn't find the reference stated in the following sentence on the publisher's web site: "They were first published in the October 1984 issue of the Journal of Quality Technology in an article by Lloyd S Nelson."

Specifically, I looked on and didn't find any related articles. Can someone find the correct citation?

Earlier references[edit]

I don't think it is correct to say that these were "first published by Nelson in 1984", as they date back much earlier. See:

  • AT&T (1956). Statistical quality control handbook, Select code 700-444. Indianapolis: AT&T Technologies.
  • Grant, E. L., & Leavenworth, R. S. (1980). Statistical quality control (5th ed.). New York: McGraw-Hill

Also, there is at least one earlier reference from Nelson (sorry - looks like this is the same reference):

  • Nelson, L. (1984). The Shewhart control chart - tests for special causes. Journal of Quality Technology, 15, 237-239 —Preceding unsigned comment added by (talk) 14:59, 24 January 2011 (UTC)

Small Confusion around Rule 3[edit]

Very informative and useful page, but I would like to clear up a couple of things which I find unclear with respect to Rule 3:

Would the following set of data be considered to include 5 points in a row continually increasing or 6 points in a row increasing?

10 1 2 3 4 5 6 0

Also, I guess that if a value is repeated, then it is NOT considered as increasing, could someone confirm this?

Similar questions exist for Rule 4, but I would hope that the same answers apply to both.

SparkyMarky 06:58, 19 December 2006 (UTC)

My understanding (in an uninformed way) is that "10 1 2 3 4 5 6 0" is six points in a continually increasing sequence. It is also five points and 4 points and 3 points in a sequence etc. It is six points (with 5 increasings). I have never seen a definition of what to do when values are equal. However in a real system, there will be noise on the measurements and an equality could probably be treated as either increasing or not at your choice. -- SGBailey 08:27, 22 December 2006 (UTC)

I think that it's five point in a row increasing. A point is only an increase if the one before it is lower in value. 2 is the first value in this sequence to be greater than the value before it so the increase starts there.
Similarly, we need to look at the point before the point we believe starts the zigzag on rule 4. The run of points needs to start with a peak or trough made by the suspected first point and the point before it.--GMcGlinn (talk) 16:22, 12 August 2008 (UTC)

Rule 5 and 6 ambiguous?[edit]

Rules 5 and 6 aren't entirely clear. For example, consider this sequence of data where +1 represents a point more than 2 sigma above the mean.

0 +1 +1 0 +1

It is clear that applying rule 5 at point 3, 2 of the last 3 points are beyond the 2 sigma limit and so the rule is violated. However, if you look at point 4, it is also true that 2 out of 3 points are beyond the 2 sigma limit, so do you count it also? And if you don't, and you restart counting after the rule is first violated, do you then skip point 5 where 2 of the last 3 points are beyond the two sigma limit? In other words, do you only apply rule 5 when the current point being considered is beyond the 2 sigma limit? (talk) 20:12, 4 March 2008 (UTC)


I've drawn a diagrammatic chart for each of the nelson rules. There is also a poster of all 8 on one canvas on wikicommoms. The diagrams are very similar in size and style of those in the Western Electric rules article but with less colour. Are there any error or can these diagrams be improved upon? --GMcGlinn (talk) 16:30, 12 August 2008 (UTC)