Microsoft Office Tutorials and References
In Depth Information
3.4.2 Cross-Sectional Data—Visual Analysis
Now, let us consider another set of data that is collected by a web-based e-tailer
(retailers that market products via the internet) that specializes in marketing to
teenagers. The e-tailer is concerned that their website is not generating the num-
ber of page-views (website pages viewed per visit) that they desire. They suspect
that the website is just not attractive to teens. To remedy the situation they hire a
web designer to redesign the site with teen’s preferences and interests in mind. An
experiment is devised that randomly selects 100 teens that have not previously vis-
ited the site and exposes them to the old and new website designs. They are told
to interact with the site until they lose interest. Data is collected on the number of
web-pages each teen views on the old site and on the new site.
In Table 3.2 we organize page-views for individual teens in columns. We can see
that teen number 1 (top of the 1st column) viewed 5 pages on the old website and
14 on the new website. Teen number 15 (the bottom of the 3rd column) viewed 10
pages on the old website and 20 on the new website. The old website and the new
website represent treatments in statistical analysis.
Our ﬁrst attempt at analysis of this data is a simple visual display—a graph. In
Exhibit 3.10 we see a frequency distribution for our pages viewed by 100 teens,
before and after the website update. A frequency distribution is simply a count
of the number of occurrences of a particular quantity. For example, if in Table 3.2
we count the occurrence of 2 page views on the old website, we ﬁnd that there
are 3 occurrences—teen 11, 34, and 76. Thus, the frequency of 2 page views is
3 and can be seen as a bar 3 units high in Exhibit 3.10. Note that Exhibit 3.10
counts all possible values of page views for old and new websites to develop the
distribution. The range (low to high) of values for old is 1-15. It is also possible
to create categories of values for the old, for example 1–5, 6–10 and 11–15 page
views. This distribution would have all observations in only 3 possible outcomes
and appear quite different from Exhibit 3.10.
Table 3.2 Old web site pages visited
Old Website
56241482046587523956
47176590680641885841
76281560414616861866
52557725067528975866
7
7 10 10
6 10
6
10
8
9 14
6 13 11 12
9
7
4 11
5
New Website
14 51819101111121510 9 911 91011 8 521 8
10
10 16 10
14 15
9
12 16 14 20
5 10 12 21 12 16 14 17 15
12
12 17
7
9
8
11
12 12 12
8 12 11 14 10 16
8
5
6 10
5
16
9
9
14
9
12
11 13
6 15 11 14 14 16
9
7 17 10 15
9
13 20 12
11 10
18
9 13 12 19
6
9 11 14 10 18
9 11 11

Search JabSto ::

Custom Search