300 Fundamentals of Statistics
returned by the Pearson chi square might lead you erroneously to reject or
accept the null hypothesis. The StatPlus add-in will automatically display a
warning if this occurs in your sample data. This warning was not displayed
with the data from the survey; however, there was one cell that had a low
expected count of 4.43.
If you wanted to remove a sparse cell from your analysis how would you
go about it? You can either pool columns or rows together to increase the
cell counts or remove rows or columns from the table altogether.
For these data you’ll combine the counts from the Bus,Econ, HealthSc,
and SocSci departments into a single group and compare that group to the
counts in the MathSci department. Rather than editing the data in the work-
sheet, you can combine the groups in the PivotTable.To compare the MathSci department to all other departments:1 Return to the Survey worksheet and create another PivotTable on a
new worksheet with Department in the row area and Calculus in the
column area. Display the count of Department in the Values area.
2 Remove the blank Department entry from the row area of the Pivot-
Table and remove the blank Calculus entry from the Column area of
the table.
3 With the Ctrl key held down, click cells A5, A6, and A8 in the Pivot-
Table. This selects the three rows corresponding to the Bus,Econ,
HealthSc, and SocSci departments respectively.
4 Click the Group Selection button from the Group on the Options tab
of the PivotTable Tools ribbon. Excel adds grouping variables to the
PivotTable rows as shown in Figure 7-21.Figure 7-21
Grouping
rows in the
PivotTable