R-71: 5,815 more signatures checked
Election officials have processed another 5,815 Referendum 71 signatures, rejecting 14.4 percent.
Rounding out the third day of R-71 checks, 17,317 signatures have now been processed, with a total of 15,067 accepted so far. It takes 120,577 valid voter signatures to gain a place on the statewide November ballot. Sponsors, a conservative social-action group called Protect Marriage Washington, turned in137,689 names on July 25, and the state Elections Division is doing an every-signature check to see if the people are on the state voter rolls and that the signature on the petition and the voter registration card match.
The error rate was the highest of any seen during the first three days of scrutiny. The day’s count included 22 duplicates, 69 where the signatures did not match, 732 where the person does not appear on the voter rolls, and 12 where the voter roll shows the person but doesn’t provide a signature to compare. That was a total of 835 signatures that were not accepted on Tuesday.
All told, there have been 45 duplicate signatures, 150 n0-match, 2,006 non-voters, and 49 where the state is checking with the county for a signature. Some or all of those may be added to the accepted list if the signatures come back and compare OK.
The sponsors are trying to force a public vote on the state’s new “everything but marriage” expansion of rights and responsibilities for state-registered domestic partners. The bill, Senate Bill 5688, ordinarily would have taken effect on July 26, but is on hold when the referendum by foes is pending.
The state Elections Division has just created a special Web page with all things R-71. Starting Wednesday, the check updates will also live there.
17 thoughts on “R-71: 5,815 more signatures checked”
Mr. Ammons:
First, I wanted to thank you for these updates. This is what government should be like – open and responsive. Thank you.
Second, a question: based on your post above, “some or all” of the disqualified signatures could be restored if the counties report back that they are valid. It would be helpful to know how likely it is that the counties will alter the outcome. Do you know historically what % of initially disqualified signatures later get restored as the result of the counties’ reporting back? Have the counties ever saved a referendum that was about to be rejected by the SoS?
Even a ballpark as to the county impact would be useful to get a sense as to how this thing is going.
This might be a dumb question, but duplicates will most likely go up the more signatures are checked, correct?
First, thanks for the regular updates, David. Many of us really appreciate knowing about the process and the count (especially those of us who it could directly and personally affect) .I had a few more questions, if you can indulge me 🙂
1) Have any signatures occured more than twice, and are fraud charges ever considered in such cases (especially the more extreme ones?) Are duplicates completely discounted, or is one of each duplicate allowed, if it is otherwise valid?
2) If someone is deceased now, but was alive when the petition was signed, does their signature count?
3) Could you describe how the verification works? Do those hard working students in the picture above manually type each name and address from the petition into the computer and it brings up a signature which they visually verify against the petition? Or does something else happen?
4) Follow up: If the address doesn’t exactly match the address on record, does it still count, or how much leniency is given?
Thank you!
You write, “there have been 45 duplicate signatures, 150 n0-match, 2,006 non-voters, and 49 where the state is checking with the county for a signature. Some or all of those may be added to the accepted list if the signatures come back and compare OK.”
Those with duplicate signatures and those who are non-voters cannot later be added to the accepted list. Is this correct?
Those where the state is contacting the county for for the signature possibly later could be added to the accepted list if the signatures match. Is this correct?
The final listed category is those whose signature does not match. Is this rejected determination final or is it subject to challenge and later could be added to the accepted list?
Yes, the duplicate rate increases as more signatures are checked. (Technically, that really only applies to random samples, but it appears to be happening so far with these counts.) Now, the duplicate rate is 0.26% (45 / 17,317). To estimate the total duplicates in the sample, that rate is divided by the percent of signatures checked so far, 12.6% (17,317 / 137,689). 0.26% / 12.6% = 2.1% expected total duplicates.
Dave can correct me if I’m wrong, but I think I can answer some of these questions.
David @1… we’re only talking about the 49 signatures thus far in which there is no signature electronically on file in the statewide voter database. Those are the signatures for which the SOS is contacting the counties, and those are the only signatures that can be added back into the mix later.
Jake @2… not a dumb question, as the media has thus far ignored the issue, but as I’ve discussed extensively on my blog, yes, the percentage of duplicates goes up exponentially as the sample size expands. That’s why we’ve seen 7, 16, and 22 duplicates respectively in successive batches of similar sizes, and that’s why the SOS adjusts its formula accordingly when conducting a statistical sample. Currently the duplication rate pencils out to over 2 percent of the total number of signatures submitted. Added 2 percent to the current invalidation rate, and you get a clearer projection of the final number.
In answer to the questions about adding signatures back in, Dave is only talking about the 49 thus far in which there is a voter record, but no signature electronically on file in the statewide database. They go back to the county to try to get a copy of the signature to see if it matches. I’m not sure what would happen if missing signatures were enough to make up the margin.
And by the way, I second the big thanks to Dave for providing such prompt and thorough updates.
David Johnson– Many thanks for the comment. Re the signatures with a missing signature on the registration database, yes, many of those will be shifted to the “accepted” list as the counties report back with the missing signature. Presumably elections departments have the signature locally and didn’t remember to include it in their submission to the statewide database. Thus far, this is a small number, 49 statewide, and I wouldn’t think it would make-or-break the total tally.
Jake– no such thing as a dumb question. i continue to ask tons of questions every day of my life. “lifelong learning” and all that. The answer, explicated by Rob on post No. 5, is that duplicates most likely will increase the more signatures we check.
Matthew– working on answers to yr good questions …
Charles– Re dupes and nonvoters, you are correct, these will never be shifted over to the accepted list, but many and possibly all of those where we’re checking with the counties for missing sigs can be transferred to the “accepted” stack. Re non-matching signature, our process is final as far as we are concerned once such a determination is made. there are observers from both camps watching the operation.
David G.– tnx much for your post, for your comment, and for your interesting and informative blog coverage of the math of this whole thing.
Matthew — I have answers for you and others who were interested in the same topics, courtesy of Shane Hamlin, assistant state director of elections, whose responsibilities include the initiative and referendum operation:
1) Have any signatures occurred more than twice, and are fraud charges ever considered in such cases (especially the more extreme ones?) Are duplicates completely discounted, or is one of each duplicate allowed, if it is otherwise valid?
No, at the time of this writing we have not had a case where a signature appeared on petition sheets more than twice.
Only the first occurrence of a valid signature is counted. Additional occurrences are rejected as duplicates. Finding a duplicate signature does not disqualify the initial, valid signature.
Yes, we will send any suspicious signature activity to the appropriate law enforcement authorities. In the past, we have sent suspect petition sheets to the Washington State Patrol.
The signature checkers look for anomalies that might indicate fraudulent activity, e.g. all the signatures on a single petition sheet are in alphabetical order or all the signatures appear to have been made by the same hand.
One more set of comments for this response, if I may.
Some of us in the Elections office have postulated that during a long signature gathering cycle – like an initiative to the people – where signature collecting may take place across five or six months it is likely some people may sign a petition for the same cause a couple of times. If someone signed a petition the first month it is circulated, it is not inconceivable that this same person might not remember she signed it when asked four months later. This is even more likely if multiple petitions are being circulated or if the individual simply wanted to sign the petition to get the signature gatherer to stop bothering them.
On the other hand, during short signature gathering periods perhaps fewer people are likely to sign a petition for the same cause more than once – primarily because there is less time to forget signing the first time.
We do not have any empirical data to support these scenarios and assumptions, but each seems quite plausible.
2) If someone is deceased now, but was alive when the petition was signed, does their signature count?
Yes. This is the same rule used for counting ballots. If a voter mailed in his or her ballot prior to Election Day and then passed away, the ballot is still counted (assuming the signature on the return envelope matches the signature on file).
3) Could you describe how the verification works? Do those hard working students in the picture above manually type each name and address from the petition into the computer and it brings up a signature which they visually verify against the petition? Or does something else happen?
The short answer is yes. As you requested, though, here is a brief outline of the process:
Signature verification is a manual process.
The petitions are scanned and saved as digital images to create a back-up copy of the documents.
The scanned images are not used for the verification process.
Petition sheets are then stacked and bound together to create what we call volumes. Each volume is numbered and inventoried.
In the case of Referendum 71, where we are doing a full signature check, the verification staff look at each signature on each petition page in a volume. Using a computer application we built the staff enter address and name information taken from the petition forms. This info is used to search the state voter registration list and find the voter registration info for a particular signer.
When the signer’s voter registration record is found a digital image of the signature on the voter’s registration card is displayed on the screen. The staff person compares the image from the voter registration file to the signature on the petition. If it matches, the signature is accepted.
If the staff person determines the signature does not match, the signature is sent to a senior signature verification staff member for a final determination.
We will go through this process for every single signature submitted for Ref. 71.
4) Follow up: If the address doesn’t exactly match the address on record, does it still count, or how much leniency is given?
Yes. Only the signatures need to match. However, address information is helpful in finding a registration signature if the petition signature is very hard to read.
People do not always update their voter registration information. We encourage voters to keep all their voter registration information up to date. Voters can complete a change of address on our website or with their County Auditor.
Let me join the chorus of thank yous for the prompt and informative updates on the signature count. One question I haven’t seen addressed here is what happens with a signature count if at some point it becomes clear that there are not going to be enough valid signatures to qualify the referendum?
Let’s say a ballot measure required 120,000 valid signatures and petition gatherers turn in 130,000 signatures. Does the count stop when 10,001 signatures are found invalid? Or do you complete the count anyway? I realize that the count for R-71 is likely to be quite close either way, but this would be interesting information for future reference.
Thank you, you are a model of open government.
Tom — thanks for the kinds remarks. it is our pleasure, and it’s a real teachable moment for us all about how the process works. i’ll soon have a fresh post that shows today’s error rate to be 14.2 percent and the cumulative rate to be 13.31 percent. In regards to your question, the Elections Division tells me they will count every last signature. They’re adding a second shift starting Thursday and hope to wrap up by week after next. If the sponsors more than eat up their “pad” of 17,112, it will be mathematically impossible to qualify. At the moment, they have 20,335 valid signatures accepted; they need almost exactly another 100,000.
Thank you very much David. We will all be experts by the time this is over 🙂
Has any consideration been given to using OCR (optical character recognition) software on the addresses and names, and some similar technology to compare signatures? It would probably take care of a large portion of the signatures and then the workers would just have to check the others manually.
Or maybe from a cost perspective it’s not worth it, since there are just a handful of petitions submitted each year?
So where are today’s numbers?
I thought you were going to post numbers every day around 5:30.
Thanks
Question – Can a person who signed ck. with the Office to see if all was verified and in order? In a close ending, non verified sigs. that can be corrected as valid might tip the balance.
In many ways the Gregoire squeaker election taught us that in a close call, the most minute situations can be very important. Remember there was a court ordered access for Dems. to allow voters to furnish proof of identity to allow signature matches that were questioned during ballot counting in King county. Hence, with help from civic activists and media messages, folks checked to see if there were questions about their ballot and then fixed so called non matches, allowing their ballot to count.
Will that happen here?
And, why the seemingly high number of registered voters with no signature on file? Certainly not the fault of the voter … voters have no role in maintaining data or records at any point after they think they are “safely” registered. Why then the penalty of not allowing their participation in a ref. petition if the data problem is not of their creation?
Matthew — thanks for yrs. This reply is from Shane Hamlin, assistant state elections director who oversees the I&R operation:
“Good questions. Thanks for asking.
Several Washington counties have worked with the Elections Division to explore automated signature verification technology to assist in processing mail ballot signatures.
Through these efforts the Elections Division conducted testing of an automated signature verification software called VoteRemote, currently available for use with a voting system here in Washington. While our office did not explore this technology for use in the ballot measure petition verification process, the test findings are applicable to your question.
Although our office approved the use of the VoteRemote automated signature verification product, we decided against implementing such technology for petition signature verification at this time. Counties initially interested in using automated signature verification software also opted to not use it as well.
Costs were a major factor in our decision, especially the cost per signature verified by the automated software. The vast majority of ballot measures (which are few anyway) are verified using a random sample method. The sample is small enough that the cost of automated software is higher than hiring a team of human checkers.
We remain interested in automated signature verification solutions, but are holding off making investments in these tools until the costs come down.”
George Bakan–thanks for the Q. There is currently no system for allowing a check-back to track an individual signature and how it was adjudicated. During a typical initiative, there will be over 300,000 individual names on like 15,000 petition sheets (referenda about half that, typically). The names are not in any alphabetical or geographic order, so it would a very labor-intensive job to paw through the stacks to find one specific person and how it was judged. With ballots, of course, we can track the ballot with a bar code, but can’t, and shouldn’t, track the person’s actual votes and how they were recorded. You have any ideas on this?
Re the question about the missing signature for comparison, the Elections Division says that’s a “transmission error” between the counties who sign up the voter and the state voter registration database, or between the Department of Licensing motor-voter program and the state database. The person is legit as a voter, but for some reason, the checkable signature wasn’t transmitted, and the state checkers have to contact the counties to see what’s up. It’s a pretty small number, 69 statewide so far, and most likely will be added to the “accepted” pile. They are technically rejected for the time being only, and probably could be in a separate category, neither accepted nor invalidated.
Comments are closed.