Call me a geek, but I love statistical process control charts


I’ve had a lot of jobs and done a lot of training over the years. The biggest “aha” moment of my professional career, hands down, was related to statistical process control (SPC) charts. In 2009 I had the privilege of learning about quality improvement measurement from Sandy Murray, who is a faculty member of the Institute of Healthcare Improvement. This experience fundamentally changed the way I see numbers, reports, statistics, and measurement — in every aspect of my life.

The value of SPC is hard to describe. But Mark Graban’s latest blog post “Can SPC Show if American Airlines Pilots are Staging a “Sick Out”? does a great job of showing how SPC charts can help us find the truth in the numbers we care about.

SPC charts are not the answer for everything. Using them takes time, data, and expertise in interpretation and investigation. But, if you have the right measures and the right data, they can provide great opportunities for learning and insight.

, ,

4 Responses to “Call me a geek, but I love statistical process control charts”

  1. john peter
    October 9, 2012 at 7:17 am #

    hey john peter here.statistical process control is very use full and nice software who have manufacturing electronics products this nice information thank you for giving this information..

  2. Karen Barber
    October 3, 2012 at 2:25 pm #

    Hi Gary,

    yes, and I love it when SPC software (like Chartrunner) identifies and highlights these patterns for the users as well!

    Your user-friendly data geek friend,

  3. Gary Teare
    Gary Teare
    October 1, 2012 at 10:54 pm #

    Hi Karen,

    Gary here – responding on Kyla’s behalf because she’s on a well deserved vacation!
    Thanks for your message – indeed statistical process control is a really valuable tool for looking at data and determining whether or not the variation seen is what would be “expected” given the system/process producing the measure.

    Interestingly – SPC charts can detect unexpected kinds of variation in several ways – the most obvious being if the data point is outside of the control limits. However there are other patterns of non-random variation that can be detected using SPC too – including an unusual number (i.e. 8) of consecutive data points on one side of the mid-line of the distribution of points; or an unexpected number of consecutive data points going in one direction or the other (i.e., 6 in a row going consistently up… or down) and a third rule that is a bit more obscure that indicate when a series of data points are bouncing back and forth across the mid-line too regularly.
    OK.. now I’ve demonstrated that I’m a real data geek and better stop before I put people to sleep. Thanks for your comment!

  4. Karen Barber
    September 27, 2012 at 11:32 am #

    I totally agree, Kyla! I see SPC analysis as the best way to avoid misinterpretation of data and tampering with the system. Much easier than remembering rules and interpreting run charts, in my opinion. The program sets up control limits and does that for you – totally user friendly when set up appropriately behind the scenes.. Should be the default method of continuous data display whenever we can! Even more important, as you point out, it changes the way we see things in light of understanding normal variation in a system.

Leave a Reply to john peter Click here to cancel reply.

Share your opinions and questions about this post. We welcome different points of view; but ask that you be respectful and constructive.  Your comments will be moderated before they appear on the site. Your Email address will be kept private.