We will do Example 1.25, first including Puerto Rico in the unemployment rate data, and then excluding it.
Download and open the JMP file Table1.1.
Proceed as in the Histogram tutorial to get:
Select Normal Quantile Plot from the red triangle pull down menu next to Percent:
You get:
Now we try a more formal goodness-of-fit test for normality:
Select Fit Distribution, then Normal from the red triangle menu next to Percent.
One should see:
Scroll to the right to see the following output:
Select Goodness of Fit from the red triangle pull down menu:
One gets:
The smaller the number Prob<W, the less likely it is that
the data comes from a normal distribution.
In this case - since Prob<W is equal to 0.0002 - we see that
the data is not likely to come from a normal distribution.
Let's exclude Puerto Rico by selecting row 39 and going to the Rows menu and selecting Exclude/Unexclude:
and repeat the analysis above, one gets:
and
and now the large Prob<W value - close to one - indicates that it is likely that the data excluding Puerto Rico comes from a Normal distribution.
Notice that we had a good reason for excluding Puerto Rico
since it is different from the states.
One should have a good reason such as this for excluding a data
point.
Data points should not be excluded arbitrarily.