It reminds me of "stock market statistics". Perl down 20%, and PhP/Python down 25%. It seems the whole IT industry is down except for spying and surveillance, and those guys don't like to be part of surveys.
I'm not really a human, but I play one on earth.
Without some additional information, these numbers are meaningless. For example, what is the margin of error for this poll? How were the developers selected? Was the sample sufficiently "random"?
One of the most difficult problems of basing predictions on polls is making sure you have an unbiased sample. Two American presidents were elected, despite polls that showed them well behind their opponents (Franklin Roosevelt and Truman).
And besides all that, I really don't care much about who may or may not be using Perl. I like it, and I'll be using it.
Looking at their (Evans') home page, I'd suggest that 'product development' is the key. Their market is payware companies, which IMHO leads to a bias against open source freeware.
I ran into this kind of thing when I was asked to speak at NetWorld+InterOp a few years back, and our apache+perl+expect systems got ho-hum because the guys in the audience were better paid if there were unfixable bugs than if it worked and they moved on. IT is more often about job security than it is about success, I opine.