Tim Wangler

Dr. Foster
10/21/02
Attribute Control Charts

INTRODUCTION: Attribute control charts can be used to monitor the stability of

systems where any count or percentage is accumulated. Attribute charts are the result of

an assessment using go/no go gauges, or pass fail criteria.

Attribute charts consist of:

 p chart (Graph of fraction non-conforming)

 np chart (Graph the number of non-conformances in a sample of n pieces)

 u chart (Graph the defect count / n)

 c chart (Graph the defects in a sample of n pieces)

p and np charts are used to graph non-conformities. c and u charts are used to graph

non-conforming units. The distinction is paramount to understand, and is described in

the next section.

An attribute control chart is reactive to a process, graphing the non-conformities,

or non-conforming units. An important distinction must be made regarding non-

conformities and non-conforming units. “Non-conformity may be a blemish or the

presence of some non-preferred feature” (Freeman, Mintzas, 2). A non-conforming unit

on the other hand is defined as, “A non-conforming unit, however, may fail to meet the

assessment of some criteria because of one or more non-conformities” (Freeman,

Mintzas, 2).

The relationship between quality and attribute charts is one where the attribute

chart is one more tool in the manager’s tool box in fighting the productivity killers like

scrap and rework. Reducing scrap and rework can increase productivity, decrease cost,
and decrease production capacity. “By interpreting control chart patterns, the

manufacturing manager can help reduce non-conformities and rejections that occur on-

line and thus assist with the prevention of quality problems” (Freeman, Mintzas, 2).

One important tip when computing attribute charts is to make sure the,

accept/reject boundaries are clearly defined. This will eliminate any ambiguity when

performing the necessary analysis on the particular item being sampled. Also, the

accept/reject characteristics should be agreed with by the customer.

HOW TO USE ATTRIBUTE CHARTS: Attribute charts start with a sample of items.

These items could be letters about to be mailed, bottles of beer, widgets, wajamas,

whatever your firm needs to look at. Then it is required to look at the number of non-

conformities or, non-conforming units, depending on what type of chart you are

calculating. In the case of sampling bottles of beer if there was a scuff on the label that

would be non-conformity and the chart desired would be either a c or u chart. However, if

the problem lies with the taste of the beer then the entire unit is defective and the desired

chart would be a p chart or np chart.

In the case of p and np charts the next step is to summate the column of data that

contain the proportion of defective items called p and then compute pbar. Pbar is equal to

the average number defective. The upper and lower control limits are computed using

pbar. The basic calculation for p charts is p = p +/- 3*sqrt(( p )(1-p)/n)). Some important

variables that need to be calculated for p charts are:

1. p is the proportion defective.

2. pbar is the average proportion defective.

3. n is the sample size.

 The main difference between p and np charts is that with np charts the sample size

must be the same number for every sample. The p chart allows for different sample size,

because of this nbar or the average sample size is also required to be calculated.

np charts are calculated much the same way as p charts, with one exception, the

standard error must be calculated. The standard error is calculated with the formula, Snp

= sqrt( npbar)(1-npbar/n)). The control limits are calculated with the formula, n(pbar) +/-

3Snp.

Some common variables with np charts are:

1. n is the sample size.

2. pbar is the average proportion defective.

3. Snp is the standard error.

c and u charts look at non-conformities, and specifically the c chart graphs the

number of defects, and the u chart graphs the average number of defects. The formula for

the control limits for the c and u chart are, CLc = cbar +/- 3*sqrt(cbar), CLu = ubar

3*sqrt(ubar/n).

The required variables needed to calculate c and u charts are:

1. n is the average sample size.

2. cbar is the process average number of non-conformities.

3. ubar is the process average number of non-conformities per unit.

EXAMPLE: An example of a p chart is given.

The Boise Cascade corporation is implementing new process for writing purchase

orders (PO’s). The implementation of the new system has not taken well. Personnel
authorized to write PO’s are using the new system wrong, and consequently writing

defective PO’s. A graph and the proper calculations are in figure 1.

The conclusion can be reached that the process is not stable . This can be seen by

looking at the two points that fall below the lower control limit, signifying the process is

statistically not stable. Therefore, action must be taken. That action can be additional

training using the system, or gathering user input to determine what the problem is. The

point is attribute control charts tell you when to take action on a situation and when to

leave the system alone. Although this tutorial only went through the p chart, the

calculation of np, c, and u charts are computed much the same way as the p chart.

As the use of computers grows increasingly common in the manufacturing

environment control charts will be pushed downward and not only used by engineers but

by factory floor workers. Control charts provide instantaneous feedback about the

stability of a system which can be useful to the floor worker. “Attribute data can be

collected from any type of process. Several types of non-conformities can be grouped

together on one chart, and the charts can be easily interpreted” (Zaciewski, 2).

WHERE TO GET MORE INFORMATION: The Journal of Quality Management has a

consistent amount of writings concerning control charts. Also, The TQM Magazine has

contains many articles about both attribute control charts and variable control charts.

Both publications provide articles that are accessible over the internet.
 Bibliography

1. Foster, Thomas S. Managing Quality an Integrative Approach, Prentice Hall

2001, New Jersey. Pages 365-370.

2. Freeman, Mintzas, “Simulating c and u Control Schemes”, The TQM Magazine;

Bedford, 1999 Vol. 11. Pages 1-5.

3. Levinson, “Multiple Attribute Control Charts”, Quality; Wheaton, Dec. 1994

Vol. 33, Issue 12. Pages 1-2.

4. Zaciewski, “Attribute Charts Are Alive and Kicking”, Quality; Troy, Mar. 1995

Vol. 34, Issue 3. Pages 1-3.