How Can We Help?
< All Topics

What are the Basic Data Collection in SPC

The first chapter introduced the concepts of statistical process control. But before we can apply statistics, we must first collect data. This chapter explains how to identify characteristics and begin the task of collecting data. The first five sections discuss data collection and data analysis methods and theories. The last section discusses practical data collection methods.

Types of Data 

Data can be classified generally by how it is collected. Data that is measured is called variable data. Data that is counted or classified is called attribute data. Variables data has these characteristics: 

  • It is measurable, by units of length, diameter, weight, temperature, or Newton meters, for example. 
  • It is continuous (Figure 2.1.1). How we verify it depends entirely on the accuracy and resolution of our gauging. Something could weigh 2 kilograms (kg) on our scale, but weigh 1.98 kg on a more precise scale, and weigh slightly different on other scales. 
  • It can be compared numerically. We can find the mean, range, and standard deviation of the same unit of measure.

Variables data.

Attribute data has one or more of these characteristics: 

  • It is countable. Either it exists or it does not, such as with defects on a painted surface (Figure 2.1.2). 
  • It is classified or graded using a scale, such as small, medium, and large eggs. Sometimes an arbitrary scale is substituted for measuring when the exact measurement is not important. 
  • It can be pass/fail data, such as the picture tube works or does not. 

Attribute Data

Other types of data, such as serial numbers, build sequence numbers, and piece counts are used for production control. These types are sometimes used to support the quality control function and are very often collected and analyzed for the same purposes. Data collection techniques have their application in many aspects of manufacturing, whether for production control or quality control.

Attribute data

Attribute data is often collected during final inspection. An assembled machine may be composed of parts that can be individually measured, but once assembled, it either works or it doesn’t. The success of a manufacturing process can be expressed as the percentage of good parts produced. This type of data provides an overall measure of quality improvement when the inspection is used, but it does not provide a clue as to how to make the improvement. In general, attribute data is not a good substitute for variable data. Improving a process often depends on the ability to distinguish between minute differences in dimension, weight, or some other quality characteristic. The gauging must be able to detect these differences in order to establish the variability of a process. GO/NO-GO gauges based on the low and high limits of a part specification do not provide a means for process improvement (Figure 2.1.3).

GO/NO-GO gauges do not detect the variability in a process — the main indicator of process improvement

Selecting Characteristics 

To improve quality, we must first identify the characteristics of quality in a part. These characteristics may have to do with fit or finish or perhaps something very intangible, such as product desirability. Selecting characteristics involves a clarification of purpose, and aims at identifying these characteristics as either variables data or attribute data. Some of the questions to be answered include: 

  • Purpose — Is our purpose a general one, or are we addressing a specific quality problem? Can it be defined? 
  • Problem clarification — Where is the problem noticeable? Where does it first appear? Is it a compound problem? Can it be addressed within our factory? Is there currently a method to detect the characteristics of this problem? 
  • Selecting characteristics — Can we specify the characteristics? Are they measurable? At what point in the process can they be measured or verified? Can we take action on data collected on these characteristics? Are results verifiable? The importance of selecting characteristics becomes evident when we are faced with the task of collecting data. To be worthwhile, data collection must serve our objectives. The clarification of these objectives, and the selection of characteristics that provide evidence and serve as a basis for action, will allow data collection to provide results. If the problem is not clear, or the wrong characteristics are measured, time will be wasted, and we will still have a problem.

Means of Analysis

Selecting the Means of Analysis 

How data is analyzed depends on the type of data, whether variables or attributes and the purpose for collecting it. There are also several ways to analyze data for any one purpose. One or all methods may have to be used. Table 2.1.1 classifies several means of analysis described in this book.

Table of Contents