data point (130) and the fourteenth data point (140). This yields 136.
The PERCENTILE.EXC function computes the
k
th percentile as the (n+1)p ranked item in the data
set. PERCENTILE.EXC computes the ninetieth percentile of the data as (15+1)(.9), which is the 14.4
ranked item. That is, the ninetieth percentile (assuming again data is sorted in ascending order) is
computed to be 40 percent of the way between the fourteenth data point (140) and the fifteenth data
point (300). This yields (0.60)(140) + (0.40)(300) =204. You can see that the two functions return
drastically different answers. If you consider the data to have been drawn by sampling from a large
set of data, you might assume
that given the data you’ve seen, there is much more than a 10 percent
chance that a piece of data would be more than 136. After all, two of the 15 data points are more than
130, so it does not seem reasonable to say that the ninetieth percentile of the data is only 136.