Invalid reasoning appears to be very widespread. This is even true
with respect to scientific reasoning among published research
scientists, but is most prevalent and damaging within the media.
Here is one example from a recent item produced by World Science
(http://world-science.net) a newsletter of currently topical
science-related stories to which I am subscribed and about which I
sent the following response:

In "Pot could boost psychosis risk later in life, study finds", World
Science July 27, 2007
(http://www.world-science.net/othernews/070727_cannabis.htm) based on
study published in Lancet (http://pmid.us/17662880), it is stated:

"In the new re­search, The­re­sa Moore of the Un­ivers­ity of
Bris­tol, U.K., and Stan­ley Za­m­mit of Car­diff Un­ivers­ity, Wales,
an­a­lyzed 35 past stud­ies on ma­ri­jua­na up to last year. They
found that peo­ple who had used can­na­bis were 41 per­cent more
likely than those who had nev­er used it to have any form of
psy­cho­sis. The risk rose with dos­age, the re­search­ers added. They
cal­cu­lat­ed that about 14 per­cent of psy­chot­ic episodes in young
adults in the UK would not oc­cur if can­na­bis weren't con­sumed."

The last sentence in this report (particularly the phrase "would not
occur if cannabis weren't consumed") was *not* a conclusion of the
authors of the report and would not be a conclusion of anyone who
understands scientific method.

The good evidence found for correlation between the use of cannabis
and later psychotic episodes, does *not* imply the causation of such
episodes by the use of cannabis. Because of that fact of scientific
epistemology, it is even incorrect for the researchers to refer to use
of cannabis as a "risk" for psychotic episodes. The increased
likelihood of psychotic episodes among those using cannabis could just
as easily be because those people more likely to later have psychotic
episodes are also more likely to choose to use cannabis.

One way to resolve this question would be to use psychological
personality index testing on a group of cannabis users and a group of
matched non-cannabis users to see whether or not those using cannabis
have a personality that is more likely to lead to psychosis. If this
has been done, there is certainly no mention of it in the Lancet
abstract (I do not have access to the full paper).

--Paul Wakfer

MoreLife for the rational - http://morelife.org
Reality based tools for more life in quantity and quality
The Self-Sovereign Individual Project - http://selfsip.org
Self-sovereignty, rational pursuit of optimal lifetime happiness,
individual responsibility, social preferencing & social contracting

------------EOM-----------

[The practice of assigning causation to association is very widespread in many
aspects of human activity, including in health related science writing. This
particular article being part of an email subscription was convenient for making
a comment to the editor - and to one of the author's after a quick search at
PubMed for an email address - and so it was done. We have no particular interest
in the use of cannabis other than that the usage of any substance properly
should be the decision of the individual only, with all responsibility for its
effects also that of the same individual (unless that individual has been
defrauded in regard to the substance).

We encourage readers to be aware of articles and/or reports they read (or hear)
where causation is drawn from the mere statistical correlation of factors or
events. Consider making the time to respond to writers and/or editors of
articles where this takes place pointing out the fallacy of such statements. If
you can think of a method of helping determine true causation for the associated
factor, you could also include this as a suggestion as Paul did above. Even if
only several individuals respond to major articles in this way - especially if
some are published - the point is unlikely to be ignored. Those who do submit
such comments to writers/editors (and are qualified to post) are welcome to also
send them here to MoreLife Yahoo. **Kitty]