The Secretary of State’s office is posting totals to its blog from the signature verification of Referendum 71, and reports that the first day’s totals show the signatures to be relatively “clean” thus far.
During the first day of signature-verification for Referendum 71, over 5,000 voter signatures were scrutinized and the error rate was 11.34 percent.
The State Elections Division crew turned up 4,991 valid signatures out of the 5,646 they reviewed. A handful were duplicates or the signature didn’t match the voter registration card. Almost 600 petition signers were not found on the roll of registered voters.
The early error rate — the count could take the better part of a month at the current pace of checking by about 20 crew members – was running cleaner than the historic average of 18 percent. Sponsors, a campaign group called Protect Marriage Washington, submitted 137,689 signatures. That is roughly 14 percent more than the bare minimum, 120,577, required to secure a place on the November ballot.
Taken at face value that should be encouraging news to R-71 backers seeking to put our state’s recently expanded domestic partnership rights before a vote of the people. But, well… I’m not one to simply take such things at face value.
First of all, at 11.34%, the reported error rate is actually a lot closer to the threshold than it first appears, for while it is true that sponsors submitted roughly 14% more signatures than the bare minimum, the actual maximum allowable error rate is (137,689 – 120,577) / 137,689, or 12.43%.
On the basis of this first batch of signatures, R-71 would appear to be skating by on the low end of the 11% to 16% rejection rate typically seen on all volunteer signature drives, but the raw error rate on such a small sample is deceptive as it does not account for the exponential increase in duplicate signatures as the universe of data expands. That’s why when performing a statistical sampling of submitted petitions, the Secretary of State’s Office attempts to adjust for duplicates using a complex but straight forward algorithm as defined in WAC 434-379-010.
Breaking down the data, of the 655 signatures rejected in the first batch, only 7 were duplicates, yet even this small number plugged into the statistical sampling formula suggests that R-71 would not qualify for the ballot. If we assume that this first batch of signatures, roughly 4.1% of the total submitted, represents a random sampling (and this is not a safe assumption), then it appears that R-71 would likely fall one to two thousand signatures short.
Of course, that margin is still awfully close, and we’re dealing with a very small data set. But for the moment at least, I’m cautiously optimistic that R-71 will fail to qualify for the ballot.