8.6Hypothesis Tests in Bernoulli Populations 325
should be rejected. Now
P.02{X≥ 10 }= 1 −P.02{X< 10 }
= 1 −
∑^9
i= 0
(
300
i
)
(.02)i(.98)^300 −i
=.0818 from Program 3.1
and so the manufacturer’s claim cannot be rejected at the 5 percent level of significance. ■
When the sample sizenis large, we can derive anapproximatesignificance levelαtest
ofH 0 :p≤p 0 versusH 1 :p>p 0 by using the normal approximation to the binomial. It
works as follows: Because whennis largeXwill have approximately a normal distribution
with mean and variance
E[X]=np, Var(X)=np(1−p)
it follows that
X−np
√
np(1−p)
will have approximately a standard normal distribution. Therefore, an approximate
significance levelαtest would be to rejectH 0 if
X−np 0
√
np 0 (1−p 0 )
≥zα
Equivalently, one can use the normal approximation to approximate thep-value.
EXAMPLE 8.6b In Example 8.6a,np 0 =300(.02)=6, and
√
np 0 (1−p 0 )=
√
5.88.
Consequently, thep-value that results from the dataX=10 is
p-value=P.02{X≥ 10 }
=P.02{X≥9.5}
=P.02
{
X− 6
√
5.88
≥
9.5− 6
√
5.88
}
≈P{Z≥1.443}
=.0745 ■