Pass or fail? Binomialrelated hypothesis testing and confidence intervals using independent samples - Improving the User Experience through Practical Data Analytics - page 98

FIGURE 4.2
Excel tables and chi-square test command.
you irst for the observed frequencies, then the expected frequencies. Select the appropri-
ate cells to complete the CHITEST command, per Figure 4.2 .
By typing “E4:F5” in the CHITEST command, Excel knows the data values are
where they are. Similarly for “E10:F11.” If we now click on enter, we obtain the
p -value of the test, as shown in Figure 4.3 (see arrow pointing out the p -value).
The p -value is 0.00617, well less than 0.05, and we reject H0 and conclude that
the true pass rates are different for task 1 using the two search engines.
4.3.2 SPSS
Unfortunately, you cannot enter tables such as those in Figure 4.1 directly into SPSS.
Instead, you need to enter the data as in Figure 4.4 . In one column, you use a “1”
for the Novix search engine and “2” for the Behemoth search engine. In an adjacent
column, you enter a “1” for each pass and a “0” for each fail.
Note that search engine 1 (Novix) has 3 passes out of 10, while search engine 2
(Behemoth) has 9 passes out of 10.
We now pull down Analyze and go to “Descriptive Statistics” and to submenu
“Crosstabs,” as shown in Figure 4.5 . See arrows in Figure 4.5 .
We now see the Crosstabs dialog box in Figure 4.6 .
We bring over one of the variables to the “Rows” box (see dashed arrow
in Figure 4.6 ) and the other variable to the “Columns” box (see bent arrow in
Figure 4.6 ). It does not matter which variable goes to which box. We then click
on “Statistics”—see solid (straight) arrow in Figure 4.6 . This gives us Figure 4.7 .