168
Chapter 9: Field Intercomparison
Table 9.1: The characteristics oE 1000 pp or BP estimates oE the 5% critical
points oE ei th er D 2 or k based on sampIes of size 10 from multivariate normal
distributions of dimension m. The second column contains the proportion oE
estimates greater than the true critical point. The last column contains the
proportion of tim es the null hypothesis was erroneously rejected using the
indicated procedure. (From Zwiers, 1987).
BP/D 2
Prob ability to
PP/D 2
Probability to
estimate reject
estimate reject
m
> X~.05
Ho
m
2
> XO.05
Ho
2
0.50
0.05
2
0.50
0.04
4
0.47
0.04
4
0.40
0.05
8
0.64
0.06
8
0.38
0.08
12
0.79
0.05
12
0.39
0.07
24
0.98
0.02
24
0.39
0.05
BP/k
Prob ability to
PP/k
Prob ability to
estimate reject
estimate reject
m
> kO.o5
Ho
m
> kO.05
Ho
2
0.35
0.05
2
0.49
0.05
4
0.02
0.09
4
0.47
0.09
8
0.61
0.06
8
0.37
0.07
12
0.17
0.07
12
0.43
0.08
24
0.54
0.05
24
0.38
0.07