Aliens On Earth.com Resources for those who are stranded here UFOs | Paranormal | Area 51 People | Places | Random Top 100 | What's New Catalog | New Books Search... for keyword(s)   in Page Titles in Book Title/Author in Contents of Pages ---other searches--- Metacrawler Yahoo FAST Search Google Altavista Dogpile Inference Find Northern Light IWon.com MSN Search Las Vegas Sun Our Bookstore is OPEN Mothership -> UFO -> Updates -> 1996 -> Dec -> Here

From: "Steven J. Powell" <sjpowell@access.digex.net>
Date: Mon, 30 Dec 1996 08:59:16 -0500
Fwd Date: Mon, 30 Dec 1996 11:23:52 -0500
Subject: Re: Philosophy of Science and UFOs

> From: jan@cyberzone.net (Jan Aldrich)
> Subject: Philosophy of Science and UFOs

> However, I would like to throw out some more half-baked ideas.
> 1.  LARGE NUMBER OF REPORTS:
> The actual number of ufo reports is unbelievably huge.  Few, I think,
> realize the sheer number of reports.  Vallee, Don Johnson, Barry
> Greenwood, and some of our French colleagues probably have an idea.

Something like 10,000 a year worldwide.

But this is just the rawest data, no investigation whatsoever, and is
therefore a wholly meaningless number.

Obviously if the sightings have passed even basic investigation then
they should not be included in a database upon which statistical
analysis is to be performed.

> Most reports become IFOs, but it is easy to become bogged down in
> these IFOs. Reading some of Ruppelt's papers it can be seen, by
> Sep-Oct 1952 he was very discouraged about the great amount of useless
> data, contradictory eyewitness testimony, and just plain foolishness
> that he wanted to chuck the whole thing and go to an instrumentation
> only study.

There's a catch-22 here.  Obviously everything starts with collecting
data, then at some point moves to analyzing that data.  The purpose of
analysis is to find significant patterns and significant lack of
patterning in the data.  The third basic step can go two ways.  One way
is to take the general population patterns back to specific cases and
confirm matching.  The other angle is to begin to make predictions,
collect additional data to test those predictions.

The catch-22 is that we should technically be collecting _all_ the data
so as not to risk jeopardising the database and therefore the results.

I think the safe way out of this catch-22 is to create selection
criteria that _don't_ affect the observation, just the
observing/observer.

For example, remove from the database _all_ single person observations.

>  However, the surprising thing is that Battelle did not use this data
> as a check against the Air Force files.  The USAF data was bias
> because many of the reports were from official sources around military
> and official installations which were forced to report ufos by
> regulation while public reports were voluntary.  If the same analysis
> were done using clipping service reports, what patterns would appear?

Actually, Battelle _should_ have performed two in-paralel series of
analysis on both sets of data to confirm/disprove the assumption of
bias.  I don't think (personal opinion) that the assumption is valid or
meaningful.

> 2.  RELIABILITY AND CREDITABILITY
> Vallee, Berliner, Hynek and others have addressed this in different
> ways. The posting from Tom Rice discussed the military intelligence
> system. In any database, this necessarily subjective determination,
> should be taken into account.  Don Berliner told me that when the UFO
> EVIDENCE was written about 6000 reports were screened first for
> reliable witnesses. The reports selected were then checked for
> patterns.

I don't know of an objective way for determinimg a reliable witness.  I
don't think its safe to cull the data based on observer-related
criteria.  I think it is safe to separate the data based on
observation-related criteria.  If we're looking for a pattern, and if
there are patterns, then those patterns should be present or at least
not removed from the data simple by selecting data based on
observation-related criteria.  If a patter-type happens to be Silver
Discoid then that should be reflected in both the multi-observer
sightings _and_ the single-observer sightings -OR- there's a major
problem with all the data.

> 3.  The early discussions and papers of the Condon committee seemed to
> indicate that they recognized they knew they were dealing with a small
> residuum of reports that contained the problem.
>   The Condon report suffers from this lack of focus, focus on that
> which earlier discussion had decided was the problem" the residuum.
> Everything but the kitchen sink is discussed in the report: crackpot
> items, trivial cases and truly puzzling reports.

At least today we know that the bulk of the cases, something like 80% to
95%, are rather quickly explained.  We also know that we will always
have that bulk of easily explainable data in any given year in any given
decade.  We don't _need_ to explain this dataset or even perform
statistical analysis on it.  Maybe it would be interesting for the
social science to perform statistical analysis on the
misidentification data but that isn't our focus.

> 3.  USAF, MOD, DND, ETC., STATISTICS
>    Just about everyone is impressed by statistic.  In the CIA analysis
> of the 1952 wave, they saw the AF claim of "only" 20% unexplained as
> evidence that the AF was on the right track.  The CIA thought however,
> that the AF did not pay enough attention to possible patterns in the
> ufo data.  The small percent of unidentifieds is always a reason to
> reject any further study.

That was silly then and it is silly now.  Given the amount of complexity
that is inherent in eyewitness observations of non-identified aerial
objects we _should_ expect to have unidentifieds that remain that way
_after_ investigation.  The purpose of performing statistical analysis
on that dataset is to _further_ the stalled investigation of those
observations.

After all, do we need to collect data on how many times eyewitnesses
observe the full moon and then perform statistical analysis on that data
to determine that the Moon does indeed exist <grin>?

My personal opinion is that earlier gov't-based and military-based data
analysis was designed to explain as many cases as possible and use that
dataset of explained cases as a statistical basis for ignoring the
unexplained remainder.  I think they properly assummed that the opposite
approach, to collect and analyze the unexplained cases, was likely to
yeild data patterning too closely consistent with existing classified
aerial projects, or worse, too closely consistent with the media-based
definition of "alien spaceships."  (They were probably equally worried
about both potential outcomes.)

Well, we already know that the bulk of the unexplained cases 'seem' like
"alien spaceships" because says so <grin>.  It might be worthwhile to
compare the isolated single observer sightings with the multi-observer
sightings to see which and how many patterns exist in both datasets.

--

Thanks, take care.
John.

([]][][][][][][][][][][][][][])
[                             ]
[  sjpowell@access.digex.net  ]
[                             ]
([]][][][][][][][][][][][][][])