# MGT510 - Total Quality Management - Lecture Handout 40

User Rating:  / 0
PoorBest

# Statistical Process Control….Contd.

## Example # 1 Shooting for Quality:

Mr. Khan observed that in basketball games, his son Ali’s free throw percentage averaged between 45 and 50 percent.
Ali’s process was simple: Go to the free throw line, bounce the ball four times, aim, and shoot. To confirm these observations, Ali shot five sets of 10 free throws with an average of 42 percent, showing little variation among the five sets.
Mr. Khan developed a Cause-and-Effect Diagram to identify the principal cause/s. After analyzing the diagram and observing his son’s process, he believed that the main cause was not standing in the same place on the free-throw line every time and having an inconsistent focal point. They developed a new process in which Ali stood at the centre of the line and focused on the middle of the front part of the rim. The new process resulted in a 36 percent improvement in practice. Toward the end of the 2004 season, he improved his average to 69 percent in the last three games.
During the 2005 season, Ali averaged 60 percent.
A control chart showed that the process was quite stable.
In the end of 2005, Ali attended a basketball camp where he was advised to change his shooting technique. This process reduced his shooting percentage during the 2006 season to 50 percent. However, his father helped him to reinstall his old process, and his percentage returned to its former level, also improving his confidence SP charting followed by drawing a fishbone helped to find the special cause of variation and corrected it improved the performance and hence making playing process under control limits.

Variation in the Outputs is a function of the variation in the Inputs
Y = f (X)
This idea provides some initial guidance for breaking down and understanding sources of variation. Changes in desired characteristics of the output are a direct result of changes or variation within inputs to that process. If there are inherent flaws or shortcomings within the process, they will result in variation in the outputs of the process. We examine all variation within the important input parameters in order to determine which factor plays the greatest role on variation within the output.
There is a distinct difference between the inherent failures of the system which are random. We will describe these as common cause based on the fact that we can predict the general area of the outcomes. However, these types of errors are very different than special cause problems which we can tie into specific events of conditions. These errors are special cause problems.

Variation in Product (Y) = Function of Process (X) or (Variation in Process)

Y = f (X)

How do we find Variation?

• Typically we collect data and use basic TQ tools to view it:
• Run chart
• Histogram
• Pareto……………….OR
• We can also summarize this data using more advanced statistical methods:
• Averages of smaller groups
• Ranges, or dispersions, within these groups…….and
• By combining summary statistics like averages and ranges with Run Charts to create a very powerful tool-----

## CONTROL CHARTS

Control charts examine information and data typically already collected as a metric or measurement of process output. We will examine averages and the differences in groups over a period of time in order to determine what is normal, what is expected, and what is predictable.

Here are some questions that might start the process: How do we currently analyze the problem area? Could we make more effective use of control charts to learn about the process by looking at the same information? Can we make the same pieces of data tell us more about the problem than they currently do? In short, are we getting our money’s worth out of our current analysis?

We are motivated to improve the outputs of the process. The big Y’s however, we know that variation in the outputs is a function of the variation in the inputs. As a result, we are draw to focus on these outputs. This often causes us to concentrate solely on the changes in the outputs without looking at the changes in the inputs and “adjusting” the process to manipulate the output instead of making real improvements. We concentrate primarily on the goals and not on how the system can truly perform. If we instead examine where we expect to perform and driven changes to support where we want the process to develop, the end result is improvement in the output.

When currently measure our process performance according to some standard. How will we quantify improvement? How will we know when we are done? Are we operating within our normally expected process limits? How do these relate to our specification limits? How do we measure the voice of the customer compared to the voice of the process? These are important questions to ask and answer.

## Special vs. Common Causes:

### There are distinct differences between actions designed to eliminate special cause and common cause variation:

• Special cause action eliminates a specific isolated event; does not involve a major process change
• Common cause action makes a change in the process that results in a measurable change in the normal process performance

Common cause variation is inherent to the system. It is the normal variation built into the process. There is a distinct difference between the inherent failures of the system and those caused by specific assignable events. We will describe these as common cause based on the fact that we can predict the general area of the outcomes. However, these types of errors are very different than special cause problems which we can tie into specific assignable events. These errors are special cause problems. Common cause variation is inherent to the system. It is the normal variation built into the process. There is a distinct difference between the inherent failures of the system and those caused by specific assignable events. We will describe these as common cause based on the fact that we can predict the general area of the outcomes. However, these types of errors are very different than special cause problems which we can tie into specific assignable events. These errors are special cause problems.