Microsoft Office Tutorials and References
In Depth Information
Exhibit 3.30 Change in each teen’s page views
Each of the randomly selected teens has two data points associated with the data
set: old and new website views. We begin with a very fundamental analysis: a cal-
culation of the difference between the old and new web-page views. Speciﬁcally,
we count the number of teens that increase their number of web-page views and
conversely the number that reduce or remain at their current number of views.
Exhibit 3.30 provides this analysis for these two categories of results. For the 100
teens in the study, 21 viewed fewer or the same number of web-pages for the new
design, while 79 viewed more. The column labeled Delta , column E, is the dif-
ference between the new and old website views, and the logical criteria used to
determine if a cell will be counted is > 0 placed in quotes. It is shown in the formula
bar as Countif (E3:E102, “>0”).
Again, this appears to be relatively convincing evidence that the website change
has had an effect, but the strength and the certainty of the effect may still be in
question. This is the problem with sampling—we can never be absolutely certain
that the sample is representative of the population from which it is taken.
Sampling is a fact of life, and living with its shortcomings is unavoidable. We
are often forced to sample because of convenience and the cost limitations associ-
ated with performing a census, and samples can lead to unrepresentative results for
our population. This is one of the reasons why the mathematical science of statis-
tics was invented: to help us quantify our level of comfort with the results from