Rounded up, 2 out of 7 or 4 out of 14 would get you 29%. That's far too small
a sample for accurate statistics. Many such statistics are derived from small
samples. They don't tell you that because they need to fill pages and sound
like they know what they are doing. Even if the sample is larger (say 100), it is
JW's that fill out such surveys that answered while the majority would simply
avoid answering, so it really would not represent JWism as a whole.
The article says that they contacted 186 JW's, which implies that the 29% who said they were registered to vote would have been about 54 individuals. Granted that's still too small a sample to reach any definite conclusions, it still strikes me as a high number. I can't imagine interviewing 100 publishers in a given congregation and finding that 29 of them were registered to vote. You may be right, though, that the sample itself may be skewed. Probably the more gung-ho JW's would avoid answering such a survey, especially if they knew it was an evangelical survey firm conducting it.