RESPONSE TO HYMAN

• Document Type: CREST
• Collection: STARGATE
• Document Number (FOIA) /ESDN (CREST): CIA-RDP96-00789R003200120001-3
• Release Decision: RIFPUB
• Original Classification: U
• Document Page Count: 3
• Document Creation Date: November 4, 2016
• Document Release Date: October 27, 1998
• Sequence Number: 1
• Content Type: BULL

Approved For Release 2000/08/08 : CIA-RDP96-00789R003200120001-3
: l.'g+cal Bulletin
4 s, I. 115, No. 1, 25-27
Copyright 1994 by the American Psychological Association, Inc.
0033-2909/94/$3.00
Response to Hyman
Daryl J. Bem
R. Hyman (1994) raises two major points about D. J. Bern and C. Honorton's (1994) article on the
psi ganzfeld experiments. First, he challenges the claim that the results of the autoganzfeld experi-
ments are consistent with the earlier database. Second, he expresses concerns about the adequacy of
the randomization procedures. In response to the first point, I argue that our claims about the con-
sistency of the autoganzfeld results with the earlier database are quite modest and challenge his
counterclaim that the results are inconsistent with it. In response to his methodological point, I
present new analyses that should allay apprehensions about the adequacy of the randomization pro-
cedures.
I am pleased that Ray Hyman, one of parapsychology's most
knowledgeable and skeptical critics, concurs with Charles Hon-
orton and me on so many aspects of the autoganzfeld experi-
ments: the soundness of their methodology, the clear rejection
of the null hypothesis, and, of course, the need for further repli-
cation. I hope this brief response will further augment our areas
of agreement.
Hyman raises two major points about our article. First, he
challenges our claim that the results of the autoganzfeld studies
are consistent with those in the earlier database. Second, he ex-
presses concerns about the "incomplete justification of the ade-
quacy of the randomization procedures" and speculates that
inadequate randomization may have interacted with subject or
experimenter response biases to produce artifactual results.
Consistency With the Earlier Database
The earlier ganzfeld database comprised studies whose meth-
ods and results were quite heterogeneous. Consequently, one
cannot justify any strong claims that some subsequent finding
is either consistent or inconsistent with that database. For this
reason, Honorton and I were careful not to make such claims.
With regard to the major finding, we simply observed that ear-
lier studies had achieved an overall hit rate of about 33% (25%
would be expected by chance) and noted that the autoganzfeld
experiments achieved approximately the same effect size. End
of claim.
In general, the earlier database served primarily to suggest the
kinds of variables that needed to be examined more systemati-
cally or more rigorously in the new studies. For example, previ-
ous ganzfeld studies that had used multi-image View Master
slide reels as target stimuli obtained significantly higher hit rates
I am grateful to Richard Broughton of the Institute for Parapsychol-
ogy in Durham, North Carolina, for going through the original auto-
ganzfeld computer files with me to unearth the data necessary for the
additional analyses presented in this response.
Correspondence concerning this article should be addressed to Daryl
J. Bent, Department of Psychology, Uris Hall, Cornell University, Ith-
aca, New York 14853. Electronic mail may be sent to d.bem@cor-
nell.edu.
than did studies that had used single-image photographs. This
finding prompted Honorton and his colleagues to include both
video film clips and single-image photographs in the autoganz-
feld experiments to determine whether the former were supe-
rior. They were. Our only claim about methodological compa-
rability was the modest observation that "by adding motion and
sound, the video clips might be thought of as high-tech versions
of the View Master reels."
But Hyman argues at length that video clips are not really
like View Master reels. Surely this is a matter of interpretation,
but does it really matter? Usually in psychology, successful con-
ceptual replications inspire more confidence about the reality
of the underlying phenomenon than do exact replications. I be-
lieve that to be the case here.
An example of a variable selected from the earlier database
for more rigorous reexamination was sender-receiver pairing.
Previous ganzfeld studies that permitted receivers to bring in
friends to serve as senders obtained significantly higher hit rates
than did studies that used only laboratory-assigned senders. But
as we emphasized in our article, "there is no record of how
many participants in the former studies actually brought in
friends," and hence these studies do not provide a clean test of
the sender-receiver variable. Moreover, the two kinds of studies
differed on many other variables as well.
In the autoganzfeld studies, all participants were free to bring
in friends, and it was found that sender-receiver pairs who were
friends did, in fact, achieve higher hit rates than did sender-
receiver pairs who were not friends (35% vs. 29%). But the reli-
ability of this finding is equivocal. In the archival publication of
the autoganzfeld studies, Honorton et al. (1990) presented this
finding as a marginally significant point-biserial correlation of
.36 (p = .06). In our article, however, we chose to apply Fisher's
exact test to the hit rates themselves. Because this yielded a non-
significant p value, we thought it prudent simply to conclude
that "sender-receiver pairing was not a significant correlate of
psi performance in the autoganzfeld studies."
But to Hyman, "this failure to get significance is a noteworthy
inconsistency." (In part, he makes it appear more inconsistent
than it is by erroneously stating that the earlier database yielded
a significant difference in performance between friend pairs and
nonfriend pairs. As noted earlier, this is an indirect inference at
best.)
Approved For Release 2000/08/08 : CIA-RDP96-00789R003200120001-3
 Approved For Release 2000/08/08 : CIA-RDP96-00789R003200120001-3
I submit that Hyman is using a double standard here. If the
successful replication of the relation between target type and psi
performance is not analogous to the earlier finding with the
View Master reels, then why is this near miss with a method-
ologically cleaner assessment of the sender-receiver variable a
"noteworthy inconsistency"?
Hyman cannot have it both ways. If the heterogeneity of the
original database and the methodological dissimilarities be-
tween its variables and those in the autoganzfeld studies pre-
clude strong claims of consistency, then these same factors pre-
clude strong claims of inconsistency.
Randomization
As we noted in our article, the issue of target randomization
is critical in many psi experiments because systematic patterns
in inadequately randomized target sequences might be detected
by subjects during a session or might match their preexisting
response biases. In a ganzfeld study, however, randomization is
less problematic because only one target is selected during the
session and most subjects serve in only one session. The primary
concern is simply that all the stimuli within each judging set be
sampled uniformly over the course of the study. Similar consid-
erations govern the second randomization, which takes place
after the ganzfeld period and determines the sequence in which
the target and decoys are presented to the receiver for judging.
In the 10 basic autoganzfeld experiments, 160 film clips were
sampled for a total of 329 sessions; accordingly, a particular clip
would be expected to appear as the target in only about 2 ses-
sions. This low expected frequency means that it is not possible
to statistically assess the randomness of the actual distribution
observed. Accordingly, Honorton et al. (1990) ran several large-
scale control series to test the output of the random number
generator. These control series confirmed that it was providing
a uniform distribution of values through the full target range.
Statistical tests that could legitimately be performed on the ac-
tual frequencies observed confirmed that targets were, on aver-
age, selected uniformly from among the four film clips within
each judging set and that the four possible judging sequences
were uniformly distributed across the sessions.
Nevertheless, Hyman remains legitimately concerned about
the adequacy of the randomizations and their potential interac-
tions with possible receiver or experimenter response biases.
Two kinds of response bias are involved: differential preferences
for video clips on the basis of their content and differential pref-
erences for clips on the basis of their position in the judging
sequence.
Content-Related Response Bias
Because the adequacy of target randomization cannot be sta-
tistically assessed owing to the low expected frequencies, the
possibility remains open that an unequal distribution of targets
could interact with receivers' content preferences to produce
artifactually high hit rates. As we reported in our article, Hon-
orton and I encountered this problem in an autoganzfeld study
that used a single judging set for all sessions (Study 302), a prob-
lem we dealt with in two ways. To respond to Hyman's concerns,
I have now performed the same two analyses on the remainder
of the database. Both treat the four-clip judging set as the unit
of analysis, and neither requires the assumption that the null
baseline is fixed at 25% or at any other particular value.
In the first analysis, the actual target frequencies observed are
used in conjunction with receivers' actual judgments to derive
a new, empirical baseline for each judging set. In particular, I
multiplied the proportion of times each clip in a set was the
target by the proportion of times that a receiver rated it as the
target. This product represents the probability that a receiver
would score a hit if there were no psi effect. The sum of these
products across the four clips in the set thus constitutes the em-
pirical null baseline for that set. Next, I computed Cohen's mea-
sure of effect size (h) on the difference between the overall hit
rate observed within that set and this empirical baseline. For
purposes of comparison, I then reconverted Cohen's h back to
its equivalent hit rate for a uniformly distributed judging set in
which the null baseline would, in fact, be 25%.
Across the 40 sets, the mean unadjusted hit rate was 31.5%,
significantly higher than 25%, one-sample 1(39) = 2.44, p = .01,
one-tailed. The new, bias-adjusted hit rate was virtually identi-
cal (30.7%), t(39) = 2.37, p = .01, td;R(39) = 0.85, p = .40, indi-
cating that unequal target frequencies were not significantly in-
flating the hit rate.
The second analysis treats each film clip as its own control by
comparing the proportion of times it was rated as the target
when it actually was the target and the proportion of times it was
rated as the target when it was one of the decoys. This procedure
automatically cancels out any content-related target prefer-
ences that receivers (or experimenters) might have. First, I cal-
culated these two proportions for each clip and then averaged
them across the four clips within each judging set. The results
show that across the 40 judging sets, clips were rated as targets
significantly more frequently when they were targets than when
they were decoys (29% and 22%, respectively), paired t(39) =
2.03, p = .025, one-tailed. Both of these analyses indicate that
the observed psi effect cannot be attributed to the conjunction
of unequal target distributions and content-related response bi-
ases.
Sequence-Related Response Bias
Hyman is also concerned about the randomization of the
judging sequence
because we can expect strong systematic biases during the judging
procedure. The fact that the items to be judged have to be presented
sequentially, when combined with what we know about subjective
validation . .. would lead us to expect a strong tendency to select
the first or second items during the judging series.
Hyman's hypothesis is correct: As shown in Table 1, receivers
do display a position bias in their judgments X2(3, N = 354) _
8.64, p