520 Chapter 12:Nonparametric Hypothesis Tests
1.0
0.8
0.6
0.4
0.2
0.0 1
x
0 23456
y
FIGURE 12.2 A symmetric density: m= 3.
f(x)=
{
max{0, .4(x−3)+
√
.4} x≤ 3
max{0,−.4(x−3)+
√
.4} x> 3
I 1 =1. Similarly,I 2 =1, andI 3 andI 4 equal 0. Hence, the value of the test statistic is
T= 1 + 2 =3. ■
WhenH 0 is true, the mean and variance of the test statisticTare easily computed.
This is accomplished by noting that, since the distribution ofYj=Xj−m 0 is symmetric
about 0, for any given value of|Yj|— say,|Yj|=y— it is equally likely that eitherYj=y
orYj=−y. From this fact it can be seen that underH 0 ,I 1 ,...,Inwill be independent
random variables such that
P{Ij= 1 }=^12 =P{Ij= 0 }, j=1,...,n
Hence, we can conclude that underH 0 ,
E[T]=E
∑n
j= 1
jIj
=
∑n
j= 1
j
2
=
n(n+1)
4
(12.3.1)