Deming used to say, “There is no true value of anything measured or observed.” In fact, that is an integral part of his

System of Profound Knowledge, which he formed as the basis of competitive and effective management.

There is no true value, for example, for the number of homeless people in the United States. Many have claimed that

the census under-counted this group of people. others have claimed that this group was over-represented. Who is

right?

Lets see where the numbers come from. Perhaps that will help us decide who is right.

There was, in the census, a method devised to count homeless people.

It produced a number. A different method would produce a different

number. Which is “correct”? Neither in the sense of “Truly” correct. There are any number of methods that could have been used. One method was selected and that method produced a number.

It gets more complicated. The same method would produce a diffelent number on subsequent use. Measurement error is always present and can sometimes be crucial.

So the result of the method used by the Census Bureau produced a number. That number is not the ‘true’ number of homeless people in the United States. A different method would have resulted in a different number. So the result of the method is an estimate. The results of any measurement is virtually always an estimate.

It’s best to state predictions in terms of a range of possibilities rather than as a single estimate. In our example we can say that the number of homeless people in America is probably between a low boundary number and an upper boundary number. If these boundaries are statistically calculated they are called confidence intervals. A confidence interval is a range of values that describes the uncertainty surrounding an estimate. A wide

interval suggests a less precise estimate.

Some popular election polls provide an estimate and then a ‘margin of error’. It’s the same idea and it allows an interested party to see the spread (variation) in the data, he or she knows much more about

how to interpret the number.

People in every walk of life depend heavily on numbers, but numbers are almost always the result of some measurement. Measurements yield estimates, not “True Values”. What is the spread of usable data?