Statistical Process Control (SPC) FAQS

QI Macros has a control chart wizard which will select the right type of chart for you. The right chart is based on the type of data, attribute (counted) or variable (measured), and the sample size. For attribute (i.e., integer or counted defects) data use np, p, c and u charts

If you are counting defective items (e.g. incorrect orders)

• use a np chart if the sample size is constant
• use a p chart if the sample size varies

If you are counting the number of defects (e.g. number of errors on an order)

• use a c chart if the sample size is large or not available
• use a u chart if the sample size varies

If you are using a ratio of defects/period (e.g., patient falls/1000 patient days)

• use the XmR chart 
  

For variable or measured data (i.e., decimal) such as time, cost, length, weight, etc.

• use XmR, XbarR, and XbarS Charts
• If the sample size is one use an XmR chart If the sample size is 2-10 use an XbarR chart
• If the sample size is 11-25 use an XbarS chart 
  

These rules are summarized and shown pictorially in this control chart decision tree.

If you still need more help understanding which chart to use you might consider:

• SPC Simplified 1 hour Training Video
  (item #267 - choose between Healthcare and Manufacturing and DVD).

Standard deviation only works for populations of "normal" data. Control charts use different formulas based on sample size and the type of distribution.

For a more complete explanation see our formulas page.

Control limits are calculated statistically from your raw data. They can't be specified. Don't confuse them with Specification limits which are provided by your customer.

See these instructions to add a target or goal line to your control chart.

You're probably using standard deviation (stdev) to calculate Cp and Cpk, not Sigma estimator (Rbar/d2). Stdev is used to calculate Pp and Ppk. Compare the QI Macros Pp and Ppk to your Cp and Cpk; they'll probably match. For more information, check our formulas. You may have sorted your data which prevents the proper calculation of Cp and Cpk using Rbar.

We offer telephone coaching for $95 for 30 minutes. Call 1-888-468-1537 to schedule a time.

There are two possible problems:

1. You sorted your data before you ran a histogram. Cp and Cpk calculations include Rbar (the Average of the Ranges). If your sample is sorted, Rbar will be distorted and your Cp and Cpk will be incorrect. Solution run your histogram on unsorted data.
2. You may have an error in your data (e.g. .001 is .01).

We offer telephone coaching for $95 for 30 minutes. Call 1-888-468-1537 to schedule a time.

If Cp and Cpk differ from Pp and Ppk by 0.5 or more, then one of three things may be the problem:

1. You sorted your data before you ran a histogram. Cp and Cpk calculations include Rbar (the Average of the Ranges). If your sample is sorted, Rbar will be distorted and your Cp and Cpk will be incorrect. Solution run your histogram on unsorted data.
2. Your process is unstable (e.g., trending upward or downward). Run a control chart on your data and eliminate any special causes of variation. Then return to capability analysis.
3. If you have several rows and columns of data, you may need to "flip" your data using copy then paste special transpose.

We recently created a whitepaper on histograms and capability analysis for $20. We also offer telephone coaching for $95 for 30 minutes. Call 1-888-468-1537 to schedule a time.

Many people input dummy data into the Gage RR to try it out and then wonder why the results look weird. The answer is Garbage In Garbage Out. For free articles on how to use and interpret the Gage R&R see our QI Macros Tips Articles.