
Patterns of Several Types of Points 611
data, if that were scientifically a ppropriate. For formal inference it is advisable to generate simul-
taneous envelopes by setting global=TRUE in the call to alltypes to avoid the ‘look elsewhe re
effect’.
Note th at CSRI will be accepted only when all of th e individual plot panels in Figure 14.25
indicate acceptance of CSRI. The null hypothe sis will be rejected if any of the panels indicate
rejection. The sign ificance level of this test is larger (i.e. weaker evidence) than the significance
level attached to each individual panel. It is advisable to set reuse=FALSE in the call to alltypes,
so that ea ch panel ...