Sir, - The statistics and conclusions which Assistant Commissioner Ian
Holyoake supplied to Parliament's health select committee regarding
marijuana and homicide (August 6) would be howlingly funny if their
potential to affect law and government policy were not involved.

Nine of 67 murders in one year out of a population of 3.6 million is
what we in the statistics trade call a bizarrely small statistical
sample.

With such numbers, randomness and poor data collection are far more
likely culprits than pattern. No professional statistician would dare
suggest a causative link between any two such phenomena, let alone
cast the shadow of such a suggestion on legislation.

There is an extremely confident, long-established statistical link
between substance abuse and homicide - when the substance is alcohol,
which Mr Holyoake or The Dominion article neglected to mention.

These 67 homicides involving alcohol should have been reported for the
obvious comparison of inference it would have yielded.

Another revealing statistic is the percentage of violent-crime arrests
and prisoners with a history of alcohol abuse.

Anecdotally, as manager of an emergency winter shelter for the
homeless in the United States, the single most identifiable factor in
bringing us our guests is alcohol abuse.

Well behind are heroin abuse and psychiatric illness, or a mix of
these three.

We have never encountered a guest brought in by marijuana
abuse.

It's just not that kind of substance.

Robert Merkin
Massachusetts