Microsoft Office Tutorials and References
In Depth Information
Exhibit 3.32
t-test: Paired two sample for means dialogue box
value of alpha that is selected. Often, an alpha of 0.05, 0.01, or 0.001 is appropri-
ate, and we will choose 0.05. Once the data is provided in the dialogue box, a table
with the resulting analysis appears. See Exhibit 3.32 for the dialogue box inputs and
Exhibit 3.33 for the results.
The resulting t-Stat , 9.843008, is compared with a critical value of 1.660392
and 1.984217 for the one-tail and two-tail tests, respectively. This comparison
amounts to what is known as a test of hypothesis . In hypothesis testing, a null
hypothesis is established: the means of the underlying populations are the same and
therefore their difference is equal to 0. If the calculated t-stat value is larger than
the critical values, then the hypothesis that the difference in means is equal to 0 is
rejected in favor of the alternative that the difference is not equal to 0. For a one-tail
test , we assume that the result of the rejection implies an alternative in one direc-
tion. In our case, we might compare the one-tail critical value (1.660392) to the
resulting t-Stat (9.843008), where we assume that if we reject the hypothesis that
the means are equal, we then favor that the new website mean is in fact greater than
the old . The preliminary analysis that gave us a 4.29 page increase would strongly
suggest this alternative. The one-tail test does in fact reject the null hypothesis since
9.843008 is greater than (>) 1.660392. So the implication is that the difference in
means is not zero.
If we decide not to impose a direction for the alternative hypothesis, a two-tail
test of hypothesis is assumed. We might be interested in results in both directions:
a possible higher mean suggesting that the new website improves page views or a
Search JabSto ::




Custom Search