Microsoft Office Tutorials and References
In Depth Information
3. Can the results for our sample of one hundred teen subjects be generalized to the
population of all possible teen visitors to our website?
4. How secure are we in our generalization of the sample results to the population
of all possible teen visitors to our website?
As with our time series data, there are many other questions we could ask, but
these four questions are certainly important to our understanding of the effective-
ness of the new website design. Additionally, as we engage in the analysis, other
questions of interest may arise. Let us begin with a simple examination of the data.
Exhibit 3.29 presents the descriptive statistics for the new and old website data.
Notice that the mean of the old website is 7.54 and the new website mean is
11.83. This appears to be a signiﬁcant difference, an increase of 4.29 pages visited.
But the difference could also be a matter of the sample of 100 individuals we have
chosen for our experiment; that is, the 100 observations may not be representative
of the universe of potential website visitors. Yet, in the world of statistics, a random
sample of 100 is often a relatively substantial number of observations. The website
change in views represents an approximately 57% increase from the old page views.
Can we be sure that a 4.29 page change is indicative of what will be seen in the
universe of all potential teen website visitors? Fortunately, there are statistical tools
available for examining the question of our conﬁdence in the outcome of the 100
teens experiment. We will return to this question momentarily, but in the interim, let
us examine the changes in the data a bit more carefully.
Exhibit 3.29 New and old website descriptive statistics