
Classical approaches to data analysis 379
Now, let us consider the ratio s
2
B
/s
2
A
.
F
0.975(12,6)
= 5.37 (12.32)
F
0.025(12,6)
=
1
F
0.975(6,12)
=
1
3.73
= 0.268 (12.33)
If the s tatistic s
2
B
/s
2
A
is lower than 0.268 or higher tha n 5.37, we will reject
the null hypothesis. If the statistic s
2
B
/s
2
A
is in between the two values, we will
not be able to reject the null hypothesis:
1
3.73
≤
s
2
B
s
2
A
≤ 5.37 (12.34)
An examination of equations 12.31 and 12.34 reveals that the same condi-
tions are used in both cases, independently of the choice of the populations.
For instance, if s
2
A
is to be significantly higher than s
2
B
, the statistic s
2
A
/s
2
B
has to satisfy:
s
2
A
s
2
B
> F
0.975(6,12) ...