Ppk Study FAQ

Mathews Malnar and Bailey, Inc.

Quality engineering, applied statistical consulting,
and training services for R&D, product, process,
and manufacturing engineering organizations.

Ppk Study FAQ

Q: What are process capability/performance statistics?
A: Process capability/performance statistics (e.g. cp, cpk, Pp, Ppk) are calculated from sample data and used to indicate how the process is performing relative to its specification limits. When certain conditions are satisfied, a pair of statistics like cp and cpk or Pp and Ppk can be used as surrogates for the process fraction defective.

Q: What's the difference between process capability and process performance statistics?
A: Process capability statistics, like cp and cpk, only take into account short term process variation. Consequently, they will give overly optimistic values when the process is actually drifting. Process performance statistics consider both short term and long term variation so they tend to be more representative of the true quality of the process. Consider cp and cpk to be the goal for Pp and Ppk, respectively, which would happen when the process is brought into statistical control.

Q: What's the relationship between cp and Pp versus cpk and Ppk?
A: cp and Pp compare the natural width of the process to the process tolerance. cpk and Ppk account for the additional effect of process centering errors.

Q: What conditions must be satisfied to validate process capability/performance statistics?
A: There are four conditions that must be satisfied to make process capability/performance statistics meaningful:

The sample must be representative of the process.
The distribution of the quality characteristic must be normal.
The process must be in statistical control.
The sample size must be sufficiently large.

Q: What happens when these conditions are not satisfied?
A: When one of more of the above conditions are not satisfied, the calculated process capability/performance statistics may not be representative of the true state of the process. For example, the actual defective rate of the process might be much lower or much higher than is indicated by a given pair of statistics.

Q: Since process capability/performance statistics are sensitive to so many conditions, aren't there alternative methods available for evaluating a process that are more robust?
A: There are more robust methods available (e.g. nonparametric tolerances), but they are so conservative that they aren't practical.

Q: How does Ppk Study help me validate these conditions?
A: For a given situation, you will have to provide Ppk Study with time-ordered process data, specification limits, and the performance goal. Then Ppk Study creates several graphs and related output to help you assess the latter three conditions. The first condition - that the sample is representative of the process - is determined by your sampling plan and cannot be validated using Ppk Study.

Q: Why does Ppk Study focus attention on Ppk and not cp, cpk, or Pp?
A: The four process capability/performance statistics obey the following inequalities: Ppk < Pp < cp and Ppk < cpk < cp. Ppk is always smaller than the other statistics because it accounts for process centering error and both short term and long term variation. This means that, of the four statistics, Ppk is the most conservative and the most indicative of the true performance of the process.

Q: Why don't more people realize that cp and cpk are insufficient indicators of process quality?
A: Ignorance.

Q: Who uses Ppk Study?
A: Ppk Study is used by operators to continuously monitor their processes, by quality inspectors and auditors to validate completed jobs, by purchasing agents to validate a potential supplier's claims about their process capability, and by receiving inspection operators to validate incoming material.

Q: Why did you write Ppk Study?
A: We had customers who identified several locations in their organizations where they needed a tool like Ppk Study, but they couldn't afford to install a more comprehensive software package in all of those locations. Furthermore, they couldn't expect the users at those sites to learn a more complicated interface or interpret a more complicated output.

Q: What are typical goals for Ppk?
A: Until a few years ago, most organizations used a goal of Ppk > 1.33, however, processes that must deliver Six Sigma performance require Pp > 2.0 and Ppk > 1.5.

Q: What support is provided for Ppk Study?
A: Ppk Study comes with a complete user's guide (download) and you may call or e-mail our office for help installing or using the program or interpreting its output.

Q: How does the trial version of Ppk Study work?
A: The trial version of Ppk Study is fully functional, however, the program will only start 20 times. You may analyze as many data sets as you like each time you start the program. If you decide to purchase Ppk Study, send us the ten digit registration number from the splash screen and we will send you back the unique activation code for your computer. Each computer will require its own unique activation code.

Q: Can I install Ppk Study on a network?
A: We should be able to help you install Ppk study on your network, but every network is unique and it may take us several attempts to find a configuration that will work on your system.. In this mode, you will only be able have Ppk Study open on as many computers as you have licenses.

Q: What is gnuplot?
A: Gnuplot is a shareware plotting program that we chose to implement the necessary graphics. The version of gnuplot packaged with Ppk Study is stripped down to minimize the size of the installation, but you can find the full version of gnuplot and more information at www.gnuplot.info.

Q: Why are there three curves shown in the Ppk plot?
A: The middle curve in the Ppk plot shows the running Ppk value versus observation number. The other two curves indicate the running 95% confidence limits for the true but unknown population Ppk value determined from the sample data. If the confidence interval falls completely below the target Ppk, the process is not capable. If the confidence interval falls completely above the target Ppk and if all other conditions are satisfied, the process is capable. If the confidence interval contains the target Ppk, then the data are insufficient to make a decision.

Q: Why do the confidence limits on the Ppk plot get narrower?
A: The confidence interval width decreases as the sample size increases.

Q: What's the benefit of knowing the Ppk confidence interval if the data are insufficient to determine if the process is or is not capable?
A: The rate of convergence of the Ppk confidence interval provides some clues about how much data will be required to finally determine if a process is capable or not.

Q: Can the Ppk Study graphical output be used in a Word document?
A: Yes. Right click the title bar of the graph, select Options> Copy to Clipboard, and then paste the graph into your Word document. The graphical output takes up about 1/2 of a page which leaves the other half of the page for administrative info, analysis, and interpretation. Using this format, each quality characteristic usually can be documented on a single page.

Q: Describe a typical circumstance under which someone buys Ppk Study.
A: In response to several instances of receiving defective material, a purchasing agent added the requirement that his suppliers provide cp and cpk statistics for each CTQ characteristic to demonstrate that their processes were capable. Unfortunately, even when cp and cpk statistics met minimum requirements, there were still instances of excessive defectives. Closer examination indicated that these processes were not normal, not in control, contained outliers, or a combination of these problems. The purchasing agent acquired a copy of Ppk Study to make his own evaluations of his suppliers' data. He eventually bought and distributed copies of Ppk Study to his suppliers so that they could create reports that meet his specific requirements.

Q: Are there other software packages that perform analyses similar to Ppk Study?
A: Sure, however, most of those packages also do many other things which increases their price and complexity. We specifically limited the scope of Ppk Study to keep the user interface and output as simple as possible. This keeps the price down so that companies can put a copy of Ppk Study everywhere that they need one.

Q: How long does it take to learn to use Ppk Study?
A: Ppk Study's user interface is almost intuitive and can be learned in a few minutes. If you run through all of the examples provided with the software and read the corresponding analyses, you'll quickly become an expert in the use of Ppk Study and the interpretation of its output. If you need to train many users quickly, do the training in a computer lab where each person can run the software themselves. Use the example data sets provided with the software to guide the training. Plan to spend a total of about one hour explaining the interface and interpreting the example problems.

Q: How do you typically enter data into Ppk Study?
A: Although you can always type your data directly into the Ppk Study data window, the data usually already exist in a worksheet or a .dat or .txt file. Simply copy the data from the original source file and paste them into Ppk Study. See Ppk Study's Help menu for more details on entering data.

Return to MM&B Software

Return to MM&B Home Page