47 
experimenter-absent; however, this was shown not to have an 
effect on recall. If extraneous recall criteria influenced 
the recall listings, these too must have been self-selected. 
Since the full sets of things from which subjects were 
recalling exemplars were considerably larger than 10, some 
additional criteria must have been involved in the selection 
process, and they may not have been related to ownership per 
se. For example, some subjects requested further 
restrictions on the recall instructions, asking, "What do 
you want here?". Others asked about listing jointly owned 
things or things acquired but not yet paid for. Subjects 
were left to self-select their own decision rules in these 
instances. Self-selection of recall conditions and of 
extraneous recall criteria appear not to have had 
differential effects on the recall listings. The eight 
tests comparing the recall lists by the two recall 
conditions did show that there was not a high statistical 
probability that the recall listings of the two recall 
groups were different. 
Much of the effects of the self-selection processes 
should have appeared as error variance, or "noise". That 
is, the effects of one subject's biases in one direction are 
compensated for by another subject's biases in another 
direction. Some of the effects, however, are probably 
systematic and can be measured to some extent. For example,