This is quite insightful. My own selection bias: I assumed it was an
email harvesting scam:-)

Yes, that's a good point.

I doubt whether the ideal is achievable, though.  E.g. the test has presumably
been validated against a population that speaks English as its first language,
but I doubt if that's representative of the people who might follow a link from
a Java newsgroup.

Other skews include:

   - the email (mis)handling eliminates both irritable purists (such as
myself) and suspicious buggers (such as John).

   - pointing the test at a Java population will (I'd guess) tend to include a
fairly high proportion of would-be programmers who do not, in fact, "combine
mixed problem solving and learning strategies with a passion for accurate
communication".  (Yes, that's a value judgement about Java programmers'
demographics compared with the programming community in general.  It may be as
wrong as it's unflattering -- it's only a personal opinion anyway -- but there
are legitimate grounds for /questioning/ how representative, statistically,
Java programmers are (or, if you prefer, what they are representative of)).

   - A not-uncommon attitude to such tests among programmers is the "hacker"
attitude: "I'm taking this test because I'm interested in how the test is
constructed, not in what it might 'tell' me about myself".  That's the attitude
I approached it with, and I suspect that was your attitude too (plus, perhaps,
a willingness to help someone who needed to collect data).  The test may
attract disproportionately many people with that attitude.

and so on...

   -- chris