Microsoft Office Tutorials and References
In Depth Information
Exhibit 3.33
t-test: Paired two sample for means results
lower mean suggesting that the number of new site views is lower than before. The
critical value (1.9842
This indicates that we can reject the notion that the means for the new and old page
views are equal. Thus, outcomes for both the one-tail and two-tail tests suggest that
we should believe that the web-site has indeed improved page views.
Although this is not the case in our data, in situations where we consider more
than 2 means, and more than a single factor in the sample (currently we consider
a visitor’s status as a teen as a single factor), we can use ANOVA (Analysis of
Variance) to do similar analysis as we did in the t-tests. For example, what if we
determine that gender of the teens might be important and we have an additional new
website option? In that case, there are two alternatives. We might randomly select 3
groups of 100 teens each (50 men and 50 women) to view three websites—the old
website, a new one from web designer X, and a new one for web designer Y. This is
a very different and more complex problem than our paired t-test data analysis, and
certainly more interesting. ANOVA is more sophisticated and powerful statistical
test than t-tests and they require a basic understanding of inferential statistics. We’ll
see more of these tests in later chapters.
Finally, we might wonder if most teens are equally affected by the new website—
Is there a predictable number of additional web-pages that most teens will visit
while viewing the new site? Our initial guess would suggest no because of the wide
distribution of the histogram in Exhibit 3.31. If every teen had been influenced to
view exactly 4 more web-pages after viewing the new site, then the histogram would
indicate a value of 4 for all 100 observations. This is certainly not the results that we
) in this case is also much smaller than the t-Stat (9.843
Search JabSto ::

Custom Search