Microsoft Office Tutorials and References
In Depth Information
Exhibit 3.14 Descriptive statistics of old website data
distribution), –0.22838, is slightly negative, indicating mild relative flatness. The
other measures are self-explanatory, including the measures related to samples: stan-
dard error and sample variance. We can see that these measures are more relevant to
cross-sectional data than to our time series data since the 100 teens are a randomly
selected sample of the entire population of visitors to the old website for a particular
period of time.
There are several other tools that are related to descriptive statistics—Rank and
Percentile and Histogram—that can be very useful. Rank and Percentile generates
a table that contains an ordinal and percentage rank of each data point in a data
set (see Exhibit 3.15). Thus, one can conveniently state that of the 100 viewers of
the old website, individuals number 56 and 82 rank highest (number 1 in the table
shown in Exhibit 3.15) and hold the percentile position 98.9%, which is the percent
of teens that are at or below their level of views (15). Percentiles are often used to
create thresholds; for example, a score on an exam below the 30th percentile is a
failing grade.
The Histogram tool in the Data Analysis group creates a table of the frequency
of the values relative to your selection of bin values. The results could be used
to create the graphs in Exhibit 3.10. Exhibit 3.16 shows the dialogue box entries
necessary to create the histogram. Just as the bin values used to generate Exhibit
3.10 are values from the lowest observed value to the largest in increments of one,
these are the entry values in the dialogue box in Exhibit 3.16—D2:D17. (Note the
Search JabSto ::

Custom Search