Attribute Charts
There are four main types of attribute chart which
differ in the type of data which they are used to record. Before
listing the charts it will be useful to provide a couple of
definitions:
Defects are errors in the
output of a process, for example:
- bubbles in paintwork
- typing errors in a document
- flaws in a length of material
Defectives are items which,
because of the presence of one or more defects are considered faulty
and unfit for use. Examples include:
- lightbulbs which don't work
- bad copies from a photocopier
- bearings outside the specified tolerance limits
For a given sample, the number of defective items cannot
be larger than the sample size. However, the number of defects found
within the sample is potentially infinite.
The four main attribute charts are:
- Number Defective (np-Chart): this chart monitors
the number of defective items in each sample, where the sample size
is constant. This chart is sometimes called an f-Chart
('f' for faulty).
- Proportion Defective (p-Chart): this chart is
similar to the np-Chart but instead of recording the number of
defective items, the p-Chart records the proportion of
defective items in a sample (the number of defective items divided
by the sample size). This means it can be used with samples of varying
size. Theoretically, the control limits should be recalculated
each time the sample size changes. However, in practice this is
tiresome and is not really necessary, so long as the sample size
does not vary from the average by more than ±20%.
- Number of Defects (c-Chart): this chart is used to
record the number of defects found in each sample, where the sample size
is constant.
- Number of Defects per Unit (u-Chart): this is similar
to the c-Chart, but is used for samples of varying size. In order to
be able to compare samples of different sizes, this chart records the
average number of defects per unit in the sample (calculated by dividing
the number of defects by the sample size). As with the p-Chart, the
same control limits may be used so long as the sample size does not
vary from the average by more than ±20%.
In addition the XmR chart can
be used as a generic 'catch-all' chart for any attribute data. It is
not as sensitive as the more specific charts, but means you don't have
to try and decide which is the appropriate chart for a given
situation. If in doubt, the XmR chart will give more reliable results
than the wrong type of attribute chart.
In this course we will only be looking in detail at the
u-Chart.