R-71: Grand total on Thursday signature check ?>

R-71: Grand total on Thursday signature check

State election officials continue posting fresh numbers in their closely watched check of signatures for Referendum 71, with the latest complete numbers for Thursday now showing 6,483 checked and 935 rejected, for a cumulative daily error rate of 14.42 percent. The previously announced rejection rate for the day-shift check was running a bit higher, almost 15 percent.

Thursday was the first day the Elections Division used two shifts of checkers. At the new rate, the verification process could be completed in another week to 10 days.


With the evening crew’s work included, the updated numbers for Thursday: 6,483 were rejected, with 5,548 accepted and 935 rejected or awaiting an electronic signature from the voter’s home county. The day’s rejection rate was 14.42 percent. The signatures not accepted included 742 people who are not registered to vote in Washington, 45 whose signature appeared more than once, 128 whose signature did not match the one on file, and 20 who were missing a signature on the state voter registration database. Many, if not all, of the latter group will eventually be shifted over to the accepted pile once their counties report back with an electronic signature.

The new grand totals as the Friday check began: 29,940 signatures have been checked so far, 25,883 accepted and 4,057 rejected or deferred so far, for a cumulative error rate of 13.55 percent. That was the highest daily error rate so far, and above the 12.42 percent rate that sponsors will be able to absorb once all signatures have been counted. The Secretary of State’s Office has also used a 14.2 percent number to express the excess number of signatures — 17,112 — that sponsors submitted above and beyond the bare minimum required to nail down a fall ballot position, 120,577.  But the more important number to watch is the error rate — 12.42 percent. Sponsors will actually be able to sustain even lower than that rate going forward, because they have been exceeding the rate almost from the beginning. Election workers plan to check every single signature.

The numbers, which are posted on the special R-71 webpage , show that the signatures not accepted so far include 3,506 who are not registered to vote in Washington, 113 duplicates, 349 whose signatures don’t match the one on file, and 89 awaiting a checkable signature from the person’s home county.

26 thoughts on “R-71: Grand total on Thursday signature check

  1. With a quarter checked and a fairly uniformly random sample, it’s clear that the error rate is in the upper-13% to lower-14% range before duplicates, statistically speaking… but what’s up with the first two days? Is it reasonable to assume that it was likely due to an understandable burn-in period for the signature verification process where a few that would otherwise be rejected were accepted for lack of a tuned verification process?

    Or is it something that is better left in the unknown so we don’t come back and over-think analysis of the process?

  2. I heard someone from your office talking To Dave Ross on KIRO this morning. I’m sorry but I did not catch the man’s name. He stated the overall rejection rate was 11.?% and yet your site shows a much higher rejection rate.
    Which am I to believe or are there “apples and oranges” here?

  3. Are the numbers wrong in the top table on the R-71 webpage? They don’t match up with Table 1, and they don’t match up with your numbers above.

    The top table says there’s a rejection rate of just 11.63% even though the lowest daily rejection rate was 11.60% and it has been significantly higher each subsequent day.

  4. If you look at the R-71 Website too it also top says 29 county to be verified and then on the Daily snap shot it shows 89 pending. So have they already verified some of these signatures?

  5. You need to fix the page at http://wei.secstate.wa.gov/osos/en/initiativesReferenda/Pages/R-71SignatureStats.aspx or explain how we went from 4057 rejected to 3559 rejected overnight. Aside from the 89 signatures where the Secretary of State’s electronic records had no signture, there is no reason for a day-to-day drop EVER.

    It’s now 11:30 and the page has been up since 9 am. This is during business hours on a weekday. You guys have been good about getting information out–kudos to you for that. But this slip is serious, and it bothers me greatly that it’s not been fixed or explained.

  6. Also, please post cumlative numbers for each day and percentages. I can figure this stuff out, but it’s nice to just see it.


  7. Did your web guy hear “we just verified X amount” and do a little math instead of getting the breakouts? Because it sounds like in the 1 and a half hours of checking, he probably got a number for how many had been checked and assumed it was the amount of new “accepted” signatures.

    I’m just not figuring this math out. Is it as simple as the person updating the website writing down the wrong numbers?

  8. After is was found that the petition was loaded with out of state names, etc. it should have been thrown out and the people responsible for it arrested for perjury!!!

  9. David,

    Looking at the other page, you state that the rejected number of ballots (total as of this morning) is at 3,559 (11.63% error rate.) But if you add all the officially rejected ballots at the very bottom you get a total of 3,998 (13.04% error rate.)

    Can you clarify which numbers we should actually be looking at? It appears it is that latter, but those in favor of putting other people’s relationships up to a popular vote are going to milk the inconsistency for all it’s worth.


  10. I agree with David– the anti-family/anti-equality/anti-elderly side has already proven litigious, and with Eyman’s I-917 flap, this could head to the courts pretty quickly if the anti-family side doesn’t like the numbers at the end of the day.

    I’m all for transparency, but in such a brief process as this, having the numbers be questionable on a website for a good part of a day is just fuel for their fire.

  11. Dave @ 12:02. I’ve actually called the SOS office where real people answer the phone. Still waiting for a call back. You may want to try. Maybe you’ll luck out and the person you need to speak will not be on the phone with someone else.

  12. Ditto the comments above. I applaud transparency, but presenting different numbers in differnt formats with different cumulative percentages (11.63% vs. 13.55%) is more confusing than helpful.

  13. Can someone please explain this wording in the above post? I have become very confused. “Sponsors will actually be able to sustain even lower than that rate(12.42) going forward, because they have been exceeding the rate almost from the beginning. Is this a typo? It would seem that they would be UNABLE to sustain a lower rate going forward because they have been exceeding the rate from the beginning.


  14. Dan,

    I think that he meant that sponsors will HAVE to sustain a lower rate.

    Actually, none of the posted rates reflect the number we should be watching. The 89 deferred signatures should be assumed to be eventually accepted. We know they are neither non-registered or duplicates so the only chance to reject would be if after the state get’s the signature from the county it differs. The total rejected are the duplicates, the non-registered, and the bogus signatures, a total of 3,506.

    Of the total 29,940 inspected, the real fail rate is 13.25%.

    At present 22% of signatures have been reviewed. In order for this petition to be valid, the remaining signatures have to have a fail rate below 12.21%.

  15. Unfortunately, it looks like the revised numbers are correct as about 400 signatures have been “reversed” from invalid to valid.

    From http://knowthyneighbor.blogs.com/home/2009/08/why-did-wa-ref-71-invalid-signatures-drop-from-4057-to-3559-today.html

    “Surprised by the change, I spoke with David Ammons at the Secretary of State’s office (+1-360-902-4140) and found that there is an adjudication process for questionable signatures that has resulted in 409 “reversals” (i.e. an invalid signature being declared a valid signature) after review by more senior staff.”

  16. Thanks for the info Matt. It sure would have been nice to be given that information directly from the SoS’s office earlier.

  17. How is the even possible? Up until this morning, there were only 349 invalid signatures. So how could there have been 409 reversals? And of those 349, one would expect that only a handful of those would be reversed anyway. Are they saying that some of the voters not found in the rolls are suddenly “found” now?

  18. As for more than the 349 signatures being reversed, I believe that sometimes the state does not have a voter’s signature on file, but the county may have it (possibly due to the voter recently registering). Once the state checks with the county, they could verify the voter’s signature.

  19. Don, the 349 signatures are from the “signatures don’t match” column. If they don’t match, they don’t match. There is no reason to send those to the county. There is a column specifically designed for the “no signature on file.”

  20. Sorry for the delay in responding to a variety of questions and comments, mostly in regards to the Elections Division using a new reporting method that resulted in a lower calculation by them of how the cumulative error rate is going. It IS a lower number than we’ve seen and is lower than the daily snapshots, and reflects mostly an acknowledgement that the master-checkers, or supervisors, are accepting signatures previously rejected for not matching the signatures on file with the statewide voter registration database. I’m told this is 409 shifts so far. Previously discussed in this space is that missing signatures usually are supplied by the voters’ home county and they move from rejected to accepted.

  21. David, that doesn’t make sense either. In the columns there were 349 that were rejected due to the signature “not matching.” Where did the 409 number come from?

  22. Kerry– yes, the counties are regularly responding by supplying the missing electronic signatures for their voters, and then the person moves from rejected to accepted. the shift could be several days after they were first flagged.

    paul johns– the categories ebb and flow as subsequent judgments are made by the master-checkers and as the counties respond to requests for missing electronic signatures. this whole process is not always linear, but is more organic. typically, we are able to use random sampling and the process is really compressed and we know the answer within a few days about whether the measure is on the ballot or not.

    AJ–today, the webpage was updated with some Friday morning count and decisions by the master-checkers. That was later information than the blog this morning. sorry the numbers weren’t parallel.

    James–out of state addresses would never even have been checked–they would have been eliminated by the checkers before the check began. if the signer has a good washington address, the checker looks to see if they appear on the state voter database and whether signatures match those on record.

    David — the webpage will typically have the freshest info, although we are trying to keep pace with this blog. i would read both — or wait until a final determination is made!

    Paul B — can’t disagree, but i hope it’s now clear why they changed their reporting format and assumptions. it would have been better, obviously, to do so from the get-go, but we’re all learning how to do social-media updates and web updates as the check is under way — and in a way that accurately describes what’s happening. it’s more complicated than i ever knew.

    Dan Hinkley — i was trying to say that if they exceed their allowable error rate of 12.42 percent on very many days, their subsequent petition sheets need to be even cleaner for the remainer of the check. the average historic average for error rate is 18 percent of those submitted.

  23. David Ammons,

    I am still confused by the numbers. From what I can tell, the WA SoS is saying:

    1. There were 642 unreported inspections, 100% of which were valid.
    2. Of the deferred rejections, 60 were validated by counties.
    3. All 349 of the invalid signatures (100%) were overridden by supervisors and were deemed valid
    4. An addition 60 signatures which were either duplicates or unregistered voters were either declared not to be duplicates or were declared to be a registered voter.

    The net result is increasing the valid signature count from 25,883 to 26,994 for a total of 1,111 new valid signatures (642 + 60 + 349 + 60 = 1111)

    To be honest, some of this seems to strain credibility. I hope this is simply a confusion or error.

  24. So David, for 409 signatures to have been moved from rejected no sig on file and rejected to no match, with 29 signatures not on file still pending, EVERY SINGLE signature originally rejected as not matching the signature on file would have to have been accepted by the master checkers. That would be a 100% rate of misidentifing mismatched signatures. Is this normal?

  25. The numbers still don’t add up. On the official site, http://wei.secstate.wa.gov/osos/en/initiativesReferenda/Pages/R-71SignatureStats.aspx, it says the total # of signatures checked was 35296. But if I added the numbers of signatures checked column in the daily snapshot table, it should be 35863. Where did the 567 signatures checked go? Why were they not included in the top table?

    If I take the 31199 accepted from the top table, divided by the total checked 35863 (#’s added from the daily snapshot table), the accepted rate was only 87%, which means the error rate was 13%.

    This is very frustrating how the SOS is reporting the #’s. 4 days ago, my partner and I were full of hope that this idiotic thing would fail. And Friday came, voila. Welcome aboard, Death Train 71, non stop to hell. How could an average of error rate of 13.5% drop two full percentages overnight to 11.5%?

Comments are closed.

Comments are closed.