I wouldn't put any stake in this survey as any sort of representation of the field in any way whatsoever. The people with SO accounts who even are even aware that SO is conducting a survey is so self-selecting that its in zero way any sort of representation of the industry as a whole.
For example, in SO's survey 80% of respondents say they code as a hobby. If this survey was not self selecting I'd bet the farm that it would be the reverse - 80% didn't code as a hobby.
Is "coding as a hobby" intended to be mutually exclusive with "coding professionally"? If not, I don't think I'd be that surprised by the 80% number even if it came from a source that everyone agreed was statistically significant.
For example, in SO's survey 80% of respondents say they code as a hobby. If this survey was not self selecting I'd bet the farm that it would be the reverse - 80% didn't code as a hobby.