Lies, damned lies, and statistics: artists don't see file sharing as a threat

I enjoy reading Good Morning Silicon Valley, usually.

But a few days ago, in a comment regarding music file sharing, John quoted a survey:

Only 28 percent of the 2,755 musicians surveyed saw file-sharing as a big threat to creative industries.

What would be interesting to know is the relationship between income and the perception of whether file-sharing hurts. Or perhaps the relationship between whether one has been signed to a media label and opinion on whether file sharing hurts.  The fact is, there are dramatic differences between subpopulations within this survey.

And there is admission that the survey was conducted by self-selection. How many big-name musicians do we imagine responding to such a survey?  I suspect only a few, at best, and so the survey speaks only about the attitudes of the disaffected artist. To the extent it speaks for any.

I can imagine one's attitude towards the publicity offered by file sharing might depend strongly on whether one has a record company contract providing promotion. Given the problems in survey design evidenced here, I feel perfectly within my rights to entirely explain away the result as being related to whether the artist had professional promotion. Even more, I am interested to see that even 28% see file sharing as a threat, when I imagine a survey population that is strongly (or even entirely?) biased towards the disenfrachised.  To my mind, this result is suggestive of the opposite conclusion to the one reported.

I simply cannot understand why anyone would conduct such a worthless piece of research, or publish the results when done.  Or then report on it (John!). Apart from to rant at the insanity, as I am doing.

Ok, ok, hyperbole aside, I can imagine why someone might conduct and report on a bad survey - activism. But not why a reporter would then report on the survey, as opposed to reporting on the activist intent. Or simply ignore it.

I hesitated to blog this one, because my feelings about filesharing technology are generally positive, and I didn't want my negative comments about the shoddy use of poor statistical methodology to be interpreted as I disagree with the point being made. I'm really only commenting here on the statistics.