260 CHAPTER 11 MANAGING RISKS
You can pick any number in column A. For example, take 0.5. The corresponding value
in column B is about 30%. So, 30% of trees exceed the average by half a metre or more.
There is a 30% likelihood that sales will exceed $10.5 million.
The distribution of values is symmetrical. So for the previous example, you can also say
that 70% of trees are shorter than 10.5 metres. You can turn it upside down also. Thirty per
cent are shorter than 9.5 metres; 70% are taller than 9.5 metres. Just in case you are not
too fond of subtracting from 100, column C does it for you. Put another way, column B +
column C = 100%.
Columns B and C look at one side of the picture. They reveal how much of a normal
distribution lies on either side of a specific point. Columns D and E relate to both sides of
the middle. Go back to the $1 million example. Against 1 in column A, column E indicates
that there is a 68% probability that sales will be within (i.e. plus or minus) $1 million of
expected value – between $9 million and $11 million. Column D says that there is a 32%
chance that they won’t be. Clearly, columns D and E must also add to 100%.
IN THE REAL WORLD
I do not know about you, but I am seriously weary of all these numbers. Hang in there a
little longer, there is only one more step and it is the most interesting one. This is how you
apply all this to real-world situations.
1 Make a realistic (be honest) central observation, estimate or forecast.
2 Make a realistic second observation, estimate or forecast.
3 Attach a probability to the value in step 2.
4 Find the standard deviation implied by step 3.
5 Use this to estimate other values.
Box A opposite takes you step by step through the process of finding standard deviation.
For example, if you estimate that a machine can produce 150 widgets, but there is a 10%
chance that it will stamp out only 126, you can establish that the standard deviation for
this situation is 20.
Armed with the standard deviation you can do one of the following.
1 Choose any other value, find the z score and read off the percentage likelihood of
that value (see Box B on page 262). For example, if you chose 126 you would find
that there is a 10% chance of making this many or fewer widgets.
2 Choose any percentage, read off the z score, and convert it into a useful value. (See
Box C on page 262.) For example, there is a 90% likelihood of producing 176 or
fewer widgets.