drug/alcohol testing policy: effective?

Casual debate he

Something like .1% of the pilots randomly tested for alcohol and drugs (one
was .5%, I believe) tested positive in 2004. That's one in a thousand. As
a result of this percentage, the random test rate will stay at 25% for drugs
and something similar for alcohol.

Meanwhile, commercial pilots and operators say that the cost of a
Part-135-type drug and alcohol testing program is nearly cost prohibitive,
so it can be argued that this sort of testing program hurts General
Aviation.

The discussion is, is the aviation community's drug and alcohol habit--or
lack thereof--influenced by drug testing policy; do pilots obstain because
of drug tests, or do they obstain because they're pilots? Would it be
better for the aviation community to test after accidents only, and do away
with the current random test practice and the associated expenses? 'Cause
if you have an accident, they're going to test you anyway, correct?

What are peoples' thoughts and experiences?

-c