Two phenomena concerning displacement--the reinforcement effect and the
displacement trend--have been only meagerly investigated with virtually no
further published studies of them since they were first reported decades ago.
Because there appears to be increasing interest in displacement, and because
both of these proposed phenomena or effects have specific implications for
processes underlying displacement, it is important to establish their validity.
Even if the effects turn out to be nonexistent, the process of elimination
alone would be informative. The purpose of this investigation was to seek
further evidence of the two effects, using recently collected data.
REINFORCEMENT EFFECT
The reinforcement effect involves a certain pattern of targets that may enhance
between-trials displacement. A displacement hit is scored whenever a subject's
call corresponds either to the next target (+ 1) or the preceding target (-1).
The reinforcement effect refers to the case where displacement scoring is
significantly higher when the immediately preceding and immediately following
targets are the same than when they are different.
When the preceding and following targets are the same, the pattern will be referred to as a sandwich. A nonsandwich bracket refers
to the case where the two targets are different.
The reinforcement effect was first investigated by Soal and Goldney in 1936
(Soal & Goldney, 1943). Greville (1951, 1954) developed the statistical
analysis that Pratt (1951) applied to four sets of
data collected by Soal and Goldney. Pratt found significant evidence for the
reinforcement effect in two of the three data sets provided by Basil
Shackleton, but not in data collected from Gloria Stewart. Apparently, the only
other published study on this effect was conducted by West (1953).
Unfortunately, West used a different method of analysis, which was later
discredited (Greville, 1954). Consequently, although the reinforcement effect
is occasionally mentioned in the literature, the evidence for it seems to come
from only one subject, and with relatively few runs.
Method
Description of studies. Because the reinforcement effect involves
something in addition to "normal" displacement effects, it would not
be expected to occur in data that show no evidence of above-chance
displacement. Failure to find the effect in any set of data could simply mean
that displacement did not occur in the study. Because
there is no evidence that displacement is ubiquitous, such a finding
would be of no interest. However, if significant displacement were
to occur in the absence of the reinforcement effect, this would cas
doubt on the existence of the effect.
The data to be analyzed here were selected to give the best possible
opportunity for the reinforcement effect to occur. The 10 studies (Crandall,
1987, Studies 1 and 2; Crandall, 1988, Studies 1
to 7; Crandall & Kanthamani, 1987) made use of different kinds of
targets (ESP symbols and English nouns) and procedures (clairvoyance,
precognition, and GESP). They did not provide evidence of
reliable, nonchance displacement in subjects who ,scored above
chance on direct targets. However, missers in all studies showed
above-chance displacement, on average, when tested under favorable conditions
but not when tested under less favorable conditions. Therefore, the present
analyses were restricted to the data of missers, tested under relatively
favorable conditions. The 10 studies involved 213 missers and 259 runs of 25
trials each.
Displacement effects in the ten studies. Analysis of displacements
must take into account the potentially artifactual effects of different
numbers of call repeats. Appropriate correction formulas, provided
by Burdick and Broughton (1987), were applied to the data. To determine the
significance of displacement in the 10 studies combined, I used Stouffer's
(Wolf, 1986) procedure for combining the results
of independent samples, which resulted in CR = 3.993, p < .00004.
Consequently, the 10 studies should provide ample opportunity for the
reinforcement effect to manifest itself.
Planned analysis. Greville's (1951, 1954) analysis involves a comparison
of the number of displacement hits and misses occurring in the middle of
sandwiches with those occurring in the middle of non-
sandwich brackets. The analysis takes into account the fact that
there is a bias in favor of greater displacement scoring on sandwiches, even in
the absence of the reinforcement effect.
Results
In five of the studies, there were more displacement hits in the
middle of sandwiches and fewer in the middle of nonsandwich
brackets than would be expected; in the other five studies, the opposite trend
occurred. None of the differences was significant. The
study that was most nearly significant (p = .073) went against the
reinforcement effect. Of the three studies that independently
showed significant displacement, two yielded negative results for the
reinforcement effect. Considering this split in the results, one sees
that there is more than a problem here of low power because of
small sample sizes. Nevertheless, Stouffer's procedure was applied
to the results to see if there might be a significant trend over the
combined studies. The result was CR - 0.313, p = ;.754 (two-tailed). That the results were in the opposite direction from the reinforcement
effect is far from encouraging. It seems that the effect,
if it exists, must be regarded as either quite weak or unreliable.
Bindrim's Displacement Trend
We turn now to another early aspect of displacement research
that has received scant attention. Bindrim (1947) described a different method
of analyzing data for displacement effects. He concluded that this method was
more sensitive than the usual one (testing deviations from chance), because
with it he obtained significant
results in data that showed no other significant effects. It is unfortunate
that, as far as I can find, no one has made further tests of this effect.
His method addresses the matter of consistency of displacement
orientation, either forward or backward. Bindrim recognized that
orientation might easily fluctuate from day to day, or even from run
to run. Consequently, he looked for consistency of orientation between the two
halves of a run.
Subjects who show more displacements of one kind than the
other in both halves of the run are described as showing the d/s-placement trend, or consistency of displacement orientation.
Inconsistent subjects are those who show greater forward displacement in
one half but greater backward displacement in the other half. If a
tie occurs in either half, the data are discarded. Analyzing data provided by Gertrude Schmeidler, Bindrim found significant (p = .007)
evidence for the displacement trend in subjects who scored 3 or
fewer direct hits in runs of 25 trials.
There are at least two ways in which the displacement trend
could occur. One would involve a fairly consistent orientation either
toward the + 1 or toward the -1 target. On the other hand, there
could be an equally consistent orientation away from both the intended target and either the + 1 or - 1 target. Bindrim's results appear to
have involved avoidance, not approach. The data of his
missers showed nonsignificant missing on displacements. Consequently, there is
a possibility that the displacement trend is associated with displacement
missing, not hitting. Bindrim (1947, p. 220)
concluded that the combination of displacement missing and the
displacement trend indicated a spread of a "rejection reaction."
Thus, although the avoidance of the intended target was believed
to be a confusion of aim, not a motivated effect, Bindrim suggested
that the avoidance could also spread consistently either forward or
backward from the target.
Method
Planned analyses. Bindrim's analysis was applied to the same data
just described. The way the data are sorted for his analysis precluded the
possibility of any study-by-study analysis. There would
not be enough usable runs in each study to make such analysis
meaningful. Consequently, the analysis was applied to the combined
data from the 10 studies. The data were analyzed separately for
subjects who scored above and below chance on displacements. Subjects who
scored at chance were discarded. Bindrim's results suggest
that the displacement trend may occur only for subjects who are
below chance on displacements.
A supplementary analysis was designed to detect a more fine-grained consistency
of orientation between trials rather than between halves of the run. For this
analysis it was noted, for each successive pair of displacements, whether the
second displacement was
in the same or a different direction from the first. The number of
repeats and changes of orientation were then compared.
Results
Bindrim found evidence for the displacement trend only among
subjects who scored three or fewer direct hits. In the present case,
the results for subjects with four or fewer direct hits were virtually
identical to the results for subjects with three or fewer hits. Only the
former will be reported because it involved larger samples.
There was no evidence for the displacement trend in subjects
who scored either above or below chance on displacements with
either analysis. Disregarding ties, the former group had 48% consistent runs,
and the latter group had 50% consistent runs. The
analysis of successive pairs yielded 52% consistency for the former
group and 53% consistency for the latter group. Bindrim had found
58% consistency in Schmeidler's data.
It is difficult to account for the difference between the present
results and those found by Bindrim. Because it was not feasible to
apply the analyses on a study-by-study basis, we cannot tell whether
certain procedures were more conducive to the displacement trend
than others. It should also be noted that Bindrim had more usable
runs (212, disregarding ties) than were available here (151). Consequently, the
present results do not refute the displacement trend;
they merely provide no support for it. Similar to the conclusion concerning the
reinforcement effect, it appears that above-chance displacement does not
require any consistency of displacement orientation.
Implications for Processes Underlying Displacements
The question now is what, if anything, the preceding analyses
tell us about possible underlying processes that contribute to above-chance
displacements. Even though such processes are far from clear at this point,
perhaps we can narrow down the possibilities.
The reinforcement effect, involving what amounts to a double
signal, could be produced in either of two ways. First, assume that
psi orientation fluctuates forward and backward from the intended
target on a given trial and that it picks up partial information about
each adjacent target. In the case of a sandwich, such fluctuations
would produce an augmented signal of the same target. In the case
of a nonsandwich bracket, the fluctuations could produce interference or a
cancellation effect, as suggested by Pratt (1951, p. 108). Second, it is
possible that instead of the above there is an orientation
away from the intended target in both directions simultaneously,
again resulting in a strengthened signal in the case of a sandwich.
The absence of support for the reinforcement effect in the present
data indicates that neither of these processes was operating frequently enough
to demand our attention. Rather, it seems more likely that, in the majority of
missers who showed above-chance displacement, this was due to orientation away
from the intended target in one direction only, on any given trial.
Whether subjects were above or below chance on displacements,
they were as likely to change displacement orientation from one trial
to the next, or from the first to the second half of a run, as they
were to maintain the same orientation. Above-chance scoring on displacements
did not require any consistency of orientation between trials. Although some
may regard the combining of + 1 and - 1 displacements in other analyses of
displacement effects as a case of mixing apples and oranges, the present
results suggest otherwise.
References
BINDRIM, E. (1947). A new displacement effect in ESP. Journal of
Parapsychology, 11, 208-221.
BURDICK, D. S., 8c BROUGHTON, R. S. (1987). Conditional displacement
analysis. Journal of Parapsychology, 51, 117-123.
CRAND^LL, J. E. (1987). Effects of cognitive style and type of target on
displacements. Journal of Parapsychology, 51, 191 - 2 ! 5.
CRANDALL, J. E. (1988). Psi-missing, displacement, and artifacts: A reanalysis
of recent data. Journal of the American Society for Psychical Research,
82, 115-127.
CRANDALL, J. E., & KANTHAMANI, H. (1987). Further evidence of the relation
of displacement effects to favorability of ESP testing conditions
with a discussion of possible artifacts. In D. H. Weiner and R. D. Nelson
(Eds.), Research in parapsychology 1986 (pp. 66-69). Metuchen, NJ:
Scarecrow Press.
Greville, T. N. E. (1951). A method of evaluating the reinforcement effect.
Journal of Parapsychology, 15, 118-121.
Greville, T. N. E. (1954). A reappraisal of the mathematical evaluation of
the reinforcement effect. Journal of Parapsychology, 18, 178-183.
PRATT, J. G. (1951). The reinforcement effect in ESP displacement. Journal
of Parapsychology, 15, 103-117.
SOAL, S. G., & GOLDNEY, K. M. (1943). Experiments in precognitive
telepathy. Proceedings of the Society for Psychical Research, 47,
21-150.
WEST, D. J. (1953). Home-testing ESP experiments: An examination of
displacement effects. Journal of the Society for Psychical Research, 37,
14-25.
WOLL F. M. (1986). Meta-analysis: Quantitative methods for research
synthesis. Beverly Hills, CA: Sage.
