|
Harper Collins
Westerhill Road
Glasgow G64 2QT
316 pages
£8.99 paperback
ISBN 0-00-714501-2
Understanding the principles, methods and applications of statistics in business
can be a daunting task.
While computers take much of the tedium out of complex calculations, public
awareness of the subject of statistics has increased. It is now expected that
when the result of an opinion poll is announced the sample size should be given.
As a result, statistical terms, such a parameter and correlation have found
their way into our everyday English, often used incorrectly.
For credit professionals, the increasing use of statistics is evident. For
instance if, say from 15,000 invoices you just picked the first convenient 1,000
and tested them for proper authorisation, how reliable would your results be,
and what conclusion could you draw if you found none that were improperly
authorised?
If you found nothing wrong, on a non-statistical basis you would perhaps
conclude that 'all invoices were likely to be properly authorised'. However,
would not your conclusion rather depend on how typical of the rest of the
invoices you felt your sample to be?
For instance, did convenience mean that only ones for the first month of year
were picked, or the first few letters of the alphabet, or the first few cost
codes? Or was it 'convenient' at the time to pick items in a way that avoided
any significant bias?
Even if no errors were found in the sample, what is the chance of a sample of
this size including any errors if, say, the real error rate was 1 in 10,000? Or,
1 in 100?
What if only 10 had been sampled compared to what you thought was the likely
error rate in the whole population of invoices? In this case, perhaps the first
question to ask is 'what sort of error rate should I worry about in the first
place?'
Despite the fact that statistics can be used to address a range of situations
and questions, many people are using it with a somewhat limited theoretical or
mathematical background. It is for those that the Collins Dictionary of
Statistics will be particularly useful.
While its format is like that of a traditional dictionary, it differs in that
many of the entries are encyclopedic in treatment and include many worked
examples. These illustrate how the various statistical measures are calculated
and the tests applied.
Most of these examples use artificial data, as real data, the author informs us,
is usually more difficult to work with and can easily obscure the point being
made.
Diagrams, graphs and a range of appendices, such as lists of symbols and
formulae, statistical tables and advice on the use of statistical computer
software, all go together to make this dictionary a worthwhile addition to the
business environment.
Source:
Credit Control Journal
|
