NOTE: The following will make more sense if you have taken training with TEAM 2000 on SPC and related topics.
If this question has been on your mind the following should help to a great extent. If not, feel free to send me an e-mail at: rai_chowdhary@yahoo.com.
The answer is "Depends..."
Typically when you begin working with SPC, you
will find that you are either measuring:
a - Something that can vary on a Continuous (or Discrete) basis. For
example: Time, Temperature, Magnetic Flux, Number of cars passing an
intersection per hour, etc. For such situations the chart suitable is
typically an IMR Chart, or an X Bar R Chart. The latter is used when
the data is in rational sub-groups; for example you are getting the same service
from 3 different vendors, and are interested in studying the Average Cycle Time
(X Bar), + the Variation in CT between the 3 vendors (R). The grouping can
be done several different ways, and each way of grouping will give you a
different message. Therefore sub-grouping can not be treated lightly;
considerable thought and planning should go into it - to get it right. Note
that the IMR charts do not require sub-grouping, and you use Individual values,
together with the differences in successive values (Moving Range).
b - Something that is either good or bad, acceptable or unacceptable (the answer can have only two choices, this or that). For example: Fraction (or %) of Cell Phone Calls that are dropped / week, Percentage of patients mis-diagnosed, % of Invoices that have errors / day, etc. For these conditions the best chart to use is a "p, or np Chart." Note that in this case, we treat a product / service as defective regardless of the number of errors. Thus a product is considered defective immaterial of whether it has just one defect, or six defects for example.
c - When measuring counts of errors, or defects, typical choice is a "c, or u Chart." For example: Number of errors per invoice, Number of defects per car, Errors in baggage handling / month, etc. Here the emphasis is the actual count of errors as opposed to whether the service / product was good or bad.
Caution:
Certain types of charts while providing a good overview of how a plant /
location / business may be doing in general, can mask a lot of
information. Thus while % of Cell Phones Defective can give a good sense
of what to expect overall, it fails to clarify several factors that may be of interest:
- How many defects / phone on average (remember - a cell phone can be labeled
defective regardless whether it has a "single defect", or 10
defects)
- Did the phone work for some time, or was it defective out of the box?
- Are the defects cosmetic, or render the unit unusable?
...
Last Updated 7/11/04
Copyright Rai Chowdhary
Rai Chowdhary, his associates or any other related entities and companies are
not responsible for any loss or gain of any kind, that you may incur from the
use of this information