Article/Document:

Condon Report Proves Accuracy of Witnesses

Brad Sparks, UFO Updates Mailing List, Apr 2001

original source |  fair use notice

Summary: As much as they like Condon's conclusions about the alleged unreliability and inaccuracy of eyewitnesses, debunkers have never done any in-depth study of the data on witness accuracy that is contained in the Condon Report.

Recently on this List there has been a lot of comment on the
alleged accuracy or ianccuracy of eyewitnesses.

As much as they like Condon's conclusions about the alleged
unreliability and inaccuracy of eyewitnesses, debunkers have
never done any in-depth study of the data on witness accuracy
that is contained in the Condon Report. The Condon Committee
itself suppressed any statistical analysis of the witness
accuracy data that its own project investigators had compiled.
Instead they left the data sitting undigested, some of it in the
final report, much of it left unpublished. What was published
did include some very interesting compilations of witness
testimony on IFO's which the Committee evidently refused to
analyze because it ran counter to the conclusion the Committee
debunkers wanted -- which was that eyewitnesses are very
unreliable so please forget the whole UFO phenomenon. The data
in the Condon Report are -- conveniently for the debunkers --
just not put together.

The reason for this failure by the Condon Committee to
statistically analyze its own sighting data on witness accuracy
is evidently because the witnesses came out TOO ACCURATE and it
did in passing comment grudgingly on that accuracy in at least
one case (Case 18). Here are some statistics I compiled by
laboriously scrutinizing the case writeups in the Condon Report,
focussing on the multiple-witness IFO cases with the largest
numbers of witnesses because it's easier to cross-check multiple
witnesses against each other and it's easier to compile the
statistics:

Case 18 -- approx. 99% accurate (about 150 data points).
Case 23 -- approx. 99% accurate (about 66 data points).
Case 36 -- approx. 90% to 98% accurate (about 42 data points).
Zond 4 Rocket Re-entry -- approx. 97% accurate (about 86 data
points).

TOTAL: Approx. 97% to 98% accurate (about 344 data points total).

I did a partial analysis back in 1977 of the alleged Zond 4
rocket reentry sightings which showed that contrary to
Hartmann's disparaging remarks most o f the witness observations
were accurate (I still am not convinced it was a reentry, and
still think that it was a meteor fireball instead, but for
simplicity I'll keep calling it the "Zond 4 Rocket Reentry";
Hartmann very suspiciously omitted all time duration data that
would help decide the issue). He had padded his list with
erroneous witness opinions and conclusions instead of filtering
those out and analyzing the pure observational data. When one
does the latter, filter out opinions and conclusions, the actual
observational data turn out to be 97% accurate.

For some examples out of the Condon Report where there were
large numbers of witnesses, Case 18 turned out to be several
hot-air balloons launched as a prank experiment. However, the 9
mostly multiple-witness reports from 14 observers turned out
"highly consistent with one another," according to the Condon
Report with differences "no greater than would be expected from
situational and perceptual differences." The "degree of
similarity between the reports" was considered "noteworthy." The
investigators Low and Wadsworth commendably decided not to focus
on witness "speculations" about the nature of the objects but to
be "confined to observational data." They did mention that "Many
small discrepancies could be pointed out, especially with regard
to estimates of distance and direction, but these are not great
enough to affect the overall impression of the event." (CR p.
305) But my examination of the data shows no such "small
discrepancies" at all. It would be dishonest to criticize
witnesses for distance estimates -- really opinions rather than
observations -- when a competent scientific investigator knows
it was impossible to estimate distance. But that is always the
debunker trick -- attack and discredit the witnesses for errors
actually committed by the investigators.

I could find only _one_ error in all of the data points (which
are unfairly reduced in number by conflating the 14 multiple
witnesses into 9 reports one for each single site location). An
astronomer made a time error and thought the sighting was at
10:40 PM whereas the hoaxers and other observers more accurately
put the time at about 10:00-10:20 PM. The astronomer also was
the only one who didn't see the three hot-air balloons that had
been launched together but only saw one light but he was also
the most distant of the observers and two of the balloons may
not have been visible from his position. So out of roughly 150
data points on duration, angular size, shape, color, formation,
motion, initial elevation and direction, final elevation and
direction, and actions such as dropping sparks, only 1 or
possibly 2 points were in error, making witness observational
data about 99% accurate in this case. No wonder they didn't do a
statistical analysis of witness accuracy -- it would have turned
out too good!

Another case that the Condon Committee wrote up but didn't
statistically analyze because it showed the witnesses were too
accurate, was a daytime fireball meteor in Case 36, though the
project didn't really investigate it at all but relied on the
reports sent in by a university professor. Project staffer
Wadsworth commented that "The discrepancies in distance and size
are hardly significant because such estimates are
characteristically inaccurate," but that the _ratio_ of size to
distance (= angular size) was consistent. (CR pp. 366/8) One of
the four independent witnesses apparently got the time wrong
(said it was 10 AM instead of 9:05). That witness and two others
reported the object as "cylinder" shaped which may or may not be
inaccurate since a fireball burns a tubular path through the air
and that may be what they were trying however imperfectly to
describe (if the case had actually been investigated this could
have been clarified) -- they did agree it was a luminous or
glowing blue-green and/or yellow. Out of roughly 42 data points
in this _uninvestigated_ case, the time was wrong once, and it
is debatable whether the three "cylinder" descriptions are
errors. The time-error witness also said it appeared "metallic"
-- we know from the Iowa Fireball case Phil Klass loves to
recite that daytime fireballs can apparently seem to be
metallic, so is that an error? Something about the intense
flashing luminosity can appear to seem like a metallic sheen.
The size/distance ratios were accurate so it seems unfair to
score the witnesses for error in reporting the unobservable,
i.e., distance or size taken separately. So at worst call the
one time and three "cylinder" shapes as 4 data points in error,
or at best the time error alone for 1 data point in error, out
of roughly 42, for an accuracy rate of about 90% to 98% and had
it actually been investigated by the Condon Committee instead of
being assigned a Case Number as if it had, the seeming errors
might have been clarified into accurate observational detail.

Here's another, Case 23, where they maligned the 7 witnesses for
supposedly having "widely different versions" of the event, a
prank by Navy flyers with a searchlight, when in fact they
nitpicked and split hairs over a fairly accurate set of
accounts. The Condon Committee investigators alleged the
accounts "differed substantially as to motion, appearance,
duration of sighting, and the object's identity." (CR p. 324)
But the "object's identity" is not an observation but a
conclusion, an opinion of the witness drawing upon all the
details he/she observed and how it relates to past experience.
The "details" that "differed widely" included the fact that 5
witnesses said the object came from the NW but one said the N
(that's nitpicking!), and only the one saying the E was in error
unless perhaps that was the direction it disappeared in. The
durations ranged from about 20+ seconds to 2-3 minutes, but that
isn't necessarily due to error because the witnesses did not all
see the aircraft at exactly the same time and some focused on
the appearance of the searchlight which was briefer in duration
-- the witnesses were in two groups 1/2 mile apart. There were 3
who said the object flew straight but 2 who said it made a 90º
turn before it departed, but the investigators couldn't say who
was right or wrong about this and depending on completeness or
incompleteness of the accounts all 5 could easily be right (same
about the number of running lights seen or not seen). The
witness who guessed the object was about 50 feet in size was of
course just about exactly correct depending on the type of
twin-engine Navy plane used. In any case there was really only
one clear error, the E direction, out of about a total of about
66 data points, for an accuracy rate of almost 99%.

There were actually two tables conveniently compiling much of
the witness data in two of the cases, Tables 4 and 5 (CR pp.
304, 367). No one on the Condon Committee bothered to
statistically analyze this data to get an assessment of witness
observational accuracy. Yet three entire chapters of the report
were devoted to this subject of witness perception and accuracy
(CR pp. 559-598) without a single analysis of the project's own
case investigations of IFO's for calibrating witnesses'
reliability, even though two chapters lambasted witnesses over
the Zond 4 rocket reentry case which had NOT been investigated
by the Condon Committee but rather perfunctorily by Project Blue
Book. (CR pp. 561, 571-577, 585-588) As I said earlier, when
properly analyzed so that witnesses are not blamed for the
investigators' errors, the Zond 4 rocket reentry case comes out
about 97% accuracy rate for eyewitness observation.

Brad Sparks

Read more articles on this topic:

Unsorted Documents 2