* *** POSTMORTEM_2014-11-04 TXT - 9 Nov 2014 15:37:24 - JKNAUTH Postmortem Report for Precinct 19-09 for 11/4/14 Election ---------- ------ --- -------- ----- --- ------- -------- Statistics ---------- We had a very large turnout, which was certainly good to have when compared to the very small turnout (53 voters) for the July election. Registered voters: 4070 (as of 11/1/14) M100 vs. ATV counts 1480 vs. 1479 (see below for discrepancy) Provisional ballots: 28 Emergency bin ballots: 0 Challenged ballots: 0 11/4/14 Absentee List: 611 Late mail-in absentees: 2 Write-in ballots: Around a dozen Curbside voters: 22 AutoMark voters: 2 Spoiled ballots: 9 2016 Photo ID Ack forms: 1 Observers: 0 Voter Complaints ----- ---------- Long lines and the time required to get thru it all, particularly the initial (Registration Table) lines (see below) Unwillingness of precinct officials (exactly following BOE instructions) to try to explain the constitutional amendment. A voter said it was written in one long, hard to understand sentence. She said she had probably figured it out, but wanted confirmation and was upset that we were not allowed to say anything. We explained why, but that did not seem to satisfy her. She said she would complain to the BOE. Unwillingness of precinct officials to allow a voter to keep the old (spoiled) ballot for reference when marking a new one. We explained why. Slowness of some personnel to process voters. Several M100 issues (see below) Problems Caused by Large Voter Turnout -------- ------ -- ----- ----- ------- The large voter turnout (over 1500 voters) was certainly much appreciated compared to the tiny one in July (only 53 voters). However it did cause some problems. The major one was getting people thru the Registration Table process. The lines sometimes went several hundred feet: thru the gym, out the inside door, thru the vestibule, out the outside door, down the steps, and into the parking lot or along the sidewalk. Fortunately the weather was good. We tried several different ways of splitting things alphabetically, but one line or the other then seemed to get favored and then some people complained that their side was too slow. Again appeared the many-times-reported problem of not having pollbooks split cleanly between letters. That and the large size of the pollbooks made it harder to pass around pollbooks to try to handle clumps of voters whose names concentrated on the several books being covered by just one official. The splitting also made Registration Table signs confusing; some voters complained about that. The BOE should finally solve the letter splitting problem. Also, at least for a large election, it would probably be better if there were more pollbooks, each with fewer voters. That would make it more likely that the required pollbook would be available for the next nearby voter, chosen by the line scheduler (see next paragraph). For a large election, we should probably have an official dedicated to Registration Table line scheduling, seeing which official is available and sending voters there with as little pollbook passing as possible, not requiring that pollbooks be dedicated to specific officials. We would also provide more Registration Table space to ease pollbook sharing. As it is now with all the signs, form pads, equipment, and things taped to the table as required by the BOE, it is not as easy as it should be to pass a pollbook from one official to another. Maybe we could have letter cards handed out to the entering voters to make it easier for the line scheduler to make the match of available pollbook/official to voter name. We need to come up with some good process by 2016 to solve these problems, which are not really bad in small elections, but are definitely significant in large elections. There will very likely be an even larger turnout for the 2016 elections. Also, the new photo ID processing may slow things down at the Registration Table in 2016, at least initially. Because of the time required to fill out the very long ballot, even after a voter finally got thru the Registration Table line and processing, they would often end up in another line (sometimes pretty long) waiting for a polling booth to free up. We had 13 booths this time, about 50% more than normal; that certainly helped a lot. Sometimes people chose not to wait; they just went to the side of the gym and sat on the benches there to fill out the ballot as best they could without a good writing surface. Earlier in the year I had asked our school contact if we could get some extra tables to use as overflow "voting booths". (The gym is large enough that we could easily set up some.) They asked around, even to other schools, but were unsuccessful. Certainly it is important to have all the officials well trained before election day, not appearing to learn on the fly. Also, if they have some physical difficulty with handling a job, e.g., poor hearing, that needs to be considered. In our location, the very loud HVAC system can make it hard for voters and officials to communicate, as well as to hear the anemic M100 alarm. The crowd noise from the big turnout made it even harder to hear. Once at the end of the day, the HVAC system shut off momentarily. It was funny to see the look of shock on peoples' faces when there was suddenly such quiet. What's going on?! Handling the large voter turnout made it harder for the Chief Judge to do timely ATV and provisional envelope checking compared to smaller elections. The M100/ATV Match sheet was used by our Ballot Table much less than in previous elections. Handling the very large voter turnout gave officials less time to do the matching and also made it much harder to do because there were so many people voting in booths or on the gym benches and then going to the M100 while others were continuously being processed at the Ballot Table. So finding an acceptable sync point to do a match was pretty difficult and the matching was done sporadically, at best, thruout the day. When it could be done, a match usually seemed to exist, e.g., we matched (as best we could) with a count of 1300 at 6:30 PM. Note however that an hour and a half later (no more match checking was possible in the interim), the M100 tape count was one larger than the final ATV count. Given the number of in-flight components, I didn't have a lot of confidence in the accuracy of the matching process for this election. We closed the polls at 7:30 PM with no problem of anyone trying to get in line late. At that time we were able to get all the remaining lined-up voters into the building. We then closed the front door and posted a handwritten "Polls Closed" sign on it for any latecomers to see. We went thru the documented process of taking names from back to front of the line, but the line was moving quickly by then and only about a dozen names got listed by the time the name taker got to the front. At the time the polls closed there had been about 30 or 40 people either in some line or at the tables being processed or in the poll booths or at the M100. Things had quieted down very much from about 30 minutes before. All those voters completed voting with no problems. We then started the end-of-day procedures. The modem transmission went much more quickly than in past elections. I was surprised how few "V" voters we encountered -- fewer than 20, I think. From past election experience, I had thought we would see many more for such a large election. A significant part of the provisional voters were unreported moves, however. I think we had under half a dozen "ID" voters. In past large elections, when I worked in another precinct, the early voting percentage was much higher in that precinct, exceeding 50% of those who voted, if I remember correctly. In this election, for whatever reasons, the 19-09 early voting percentage was only about 30%. More people waited until Tuesday to vote, so we had longer lines at our polling place. Given their experience this election, maybe more 19-09 voters will vote early in 2016. Fortunately we had only one ballot style in our precinct. Split precincts probably had more hurdles to jump with the large turnout. M100 Alarm ---- ----- The alarm (beep) on the M100 is very faint and is easily missed. There is no way to adjust it. This problem has been reported for many years (see past postmortems). It is a serious issue. Often an error is not detected until the voter with the problem ballot has already left. It is usually found when the next voter tries to enter a ballot. Since the original voter is not available to correct the problem, a precinct official has to press Accept to clear the error and let the next voter enter a ballot. In our noisy environment, a big gym with a very loud HVAC system, the anemic beeps cannot be heard unless the monitoring official is standing right next to the M100. It is even worse when there is a lot of crowd noise as there was in the 11/4/14 election. But if the official stands too close to the M100, they can get accused of trying to view the voter's ballot. Of course this is a hardware and design problem, requiring a louder (adjustable) alarm or a flashing light or something along those lines, so is not easily fixable. I just want to be sure the problem is acknowledged and that future hardware upgrades will not have the same problem. In the meantime, the BOE should provide some recommendations about how to cope with this. M100 Anomalies ---- --------- Several peculiar things happened with the M100, possibly accounting for the reconciliation discrepancies described below. 1) A couple of times the M100 issued a message saying it could not read the back of the ballot. Just reinserting the ballot or turning it over and reinserting it seemed to cure that each time it occurred. 2) One lady said she inserted her ballot when the counter said some number (noted here as ####) and the M100 kicked the ballot back. She did not not recall seeing any error message or hearing an alarm, although they may have occurred. She reinserted the ballot and this time it was accepted. However she said the counter stayed at ####; it was never changed to #### + 1. She asked to see the Chief Judge. I wrote this up with her contact information and put the note in the blue bag along with comments from another judge. No one else during the day reported seeing anything similar. 3) Another lady reported that she had stood behind a voter who entered a ballot. She said the M100 kicked out *two* ballots for the voter. The lady said she thought the voter was trying to enter two ballots and had gotten an extra ballot when the first one was spoiled. (Our Ballot Table officials do NOT allow any voter to ever have two ballots at the same time.) The official monitoring the M100 thought what had happened was that the M100 was in the process of rejecting a ballot from the previous voter (who had already left) when the second voter too quickly tried to enter her own ballot. So it appeared that second voter had gotten back two ballots and seemed to be trying to voting twice. Another possibility is that a voter had been given two ballots stuck together. Our Ballot Table official often commented that the ballots seemed to be very thick and she was concerned that some voter might accidentally be given two ballots. She was very careful when giving out each ballot to check this. We never found such a double ballot (I had seen some in an election years ago.) I counted some of the ballot packs and always fan bent the stack of ballots to spread them out for counting. That probably should have split any double ballots; I never saw any in this election. However a double ballot would explain the discrepancy between the M100 count and the ATV count: M100 = ATV + 1. 4) I think some ballots were rejected as unreadable, but were accepted when re-entered. I know one voter retried a couple of times, using the same ballot without properly remarking it; it was eventually accepted. Others properly remarked their ballots with black ovals for the retry. Reconciliation Discrepancies -------------- ------------- 1) As noted above, our M100 final count was one greater than our final ATV number. I had checked each ATV during the day and thought the numbering was correct. At the end of the day, in an attempt to resolve the discrepancy, two other people went thru the ATVs one more time and found no numbering errors. Our Help Table area was isolated from the M100 and the provisional voting booth was set in the back of that. The Help Table officials did not think any provisional ballot could have gotten inserted in the M100. The possibility of two ballots stuck together was discussed above; maybe the M100 saw two where the Ballot Table official saw one. We did attempt to use the M100/ATV Match sheet during the day. However we were able to do this much less than in prior elections. See comments above in the "Problems Caused by Large Voter Turnout" section. 2) When a ballot pack is opened, we always immediately count the ballots. This time one ballot pack had 99 votes and one had 98; the other fourteen opened had 100. If this short count is ignored, the M100 count and the ballot count do match for the 11/4/14 election. The reason for the discrepancy is unknown, unless a miscount was done. However most times the counts were done multiple times, e.g., that's what I did for the 99 pack (counted it three times, in fact) and the count stayed the same. So if that 99 count was right, then one of the other "100" counts was wrong to make up the difference. If that is so, maybe there were enough wrong "100" counts to produce the difference between what was calculated at the Ballot Table and what the M100 reported. In all prior elections I have worked at, such counting did produce an exact reconciliation and helped resolve some problems, so I have much faith in that process. Maybe the M100 miscounted, possibly due to one of the anomalies described above. Ballot Bin Emptying ------ --- -------- We called a halt and emptied the left-side ballot bin twice during the day. Each time was after another 500 or so ballots had been entered. This went pretty quickly, taking only two or three minutes each time, and the voters seemed to understand the preceding announcement. Actually the ballots fell into the bin pretty well and didn't appear to be close to where a jam would occur. That surprised me, given the 17" ballot, which often has a stacking problem. I did try to joggle the ballot box occasionally during the day to deter pyramid stacking. Ballot Problems ------ -------- Of course the number of races to be voted on slowed down things a lot, contributing much to the length of the line waiting for polling booths. Several people said they were confused by all the races with only one candidate and asked what that was all about. One person said he could interpret the last race on the front side as being continued to the back side of the ballot (the ballot actually says "continue voting next side" *in the race block* on the front side). The next race on the back side had exactly the same title and he thought it might have been a continuation of the front-side race. He asked and we clarified. He then requested that we report the confusing ballot format. One person complained about the amendment wording (see above). There is nothing we could do about any of this except to provide answers where we were allowed to. Hopefully the people putting together future ballots will take such things into consideration. Please pass on these comments to those people. There were very few spoiled ballots, given the likelihood that voters could accidentally mark more than one entry in a race. Of our 30 races, 29 allowed only one entry to be marked. I know a few voters chose to just press Accept when the M100 rejected a ballot that had more than one entry marked. However that didn't happen often. Considering that and and that there were only nine spoiled ballots, the voters were being careful in their ballot marking. AutoMark -------- Two people used the AutoMark. One was visually impaired and one was unable to use his hands for writing. One of our judges showed each how to use the system. The AutoMark worked well for both voters and they were happy with it. Voter Parking Conflicts with School Bus Parking ----- ------- --------- ---- ------ --- ------- In prior elections at this polling place, we have had big issues with voter parking vs. school bus parking. (See previous postmortem reports.) For this election we dedicated a person to do both curbside processing and traffic control, particularly during the afternoon bus time. He did an excellent job, also working with the school and bus personnel. Shortly before the buses arrived, one of our judges suggested that we see if the school had any traffic cones. The school then provided a number of very large gymnasium cones, which were definitely not standard traffic cones. They helped immensely to prevent voters from trying to use parking slots that the buses would either need or would block off. A serious problem is that the main school parking lot (the one we direct voters to) quickly fills up with voters' cars, so the next voters to arrive try to come to the gym lot, where they would want to park anyway since it is right by the voting place. However because it could take a long time for a voter to vote, they could still be parked in the gym lot while trying to vote when the buses start to arrive. The buses essentially fill up the gym lot for part of the afternoon. A bus might have to be where the voter has parked, or might trap the voter's car so it could not be driven away until the buses finally have left 30-45 minutes later. The Wake County election BOE should try to get the Wake County education BOE to make the 2016 November election day be a school holiday, or a teacher workday, or something like that. Supplies and Pick-ups/Drop-offs -------- --- ------------------ There were no problems this time for the Saturday supplies pick-up, Monday pollbook pick-up, or Tuesday drop-off. One pair of scissors fell apart almost immediately while cutting string for the voting booths. One tape dispenser had little tape on it as delivered. It was great to get two blue tape rolls; in prior elections a roll was often needed in two places at the same time. We ran out of "I Voted" stickers for about an hour (displeasing/disappointing some voters); our coordinator made a special warehouse run to get more for us and the many other polling places that had the same problem. The Ballot Table checklist specifies that the M100/ATV Match Sheet is to be taped to the table during the Monday set-up. We couldn't find the sheet after searching quite a while. It wasn't in the CJ binder or the gray bin (or listed in their inventories) and wasn't in the pollbook boxes. It turned out that this sheet is packed in one of the ballot boxes, which of course cannot be opened until Tuesday. The Ballot Table checklist needs to note this and not imply the sheet is available at the Monday set-up. The "palm card"/"push card" was not well documented in pre-election material, but we liked it when we saw it. Good in particular was the part about what is needed to get a DMV ID card, which is information missing from the other 2016 handouts. Our Registration Table people said they gave out a number of cards to people who had their own acceptable IDs already, but said they would give the cards to other people who did not have such IDs. That seemed like a great way to achieve what the BOE wants to get done. We didn't need the extra ATV bag. Although the turnout was very large for this election, the material our polling place created did fit easily in one bag. We left the other bag in the CJ binder. We did have some write-ins and put the write-ins bag in the blue bag, as instructed by the BOE e-mail received just before election, overriding the blue bag checklist marked-out line. Pollbooks --------- In addition to the undesirable letter splitting and pollbook size problems mentioned in the "Problems Caused by Large Voter Turnout" section above, our Registration people had another problem with the pollbooks. They often accidentally tore out an extra ATV and then had to tape it back. We didn't have this problem in prior elections with this format of the pollbooks. Perhaps because of the large turnout there was so much more page turning of the books this time that the pages might have gotten weakened. Other than being annoying and further slowing down the Registration Table processing, this wasn't a serious problem, i.e., no voter information was lost. Help Table Education ---- ----- --------- Our first-time Help Table Official said his Help Table class had been cut short by an hour. Some of the scenarios he encountered on election day were not covered in the class. I should have emphasized again to all the officials that they needed to go thru the flowcharts the day before the election and ASK QUESTIONS if they were not clear on everything. The current flowcharts are in very good shape. Documentation ------------- Ballot Table Checklist, front page: 1) Should note that the M100/ATV Match Sheet is in the first ballot box and cannot be accessed until election day. 2) Under "During Election Day", last check bullet has a bad reference: "Registration Table procedures" should be "Ballot Table procedures". ------------------------------------------------------------------------ Report written by Jeff Knauth, Chief Judge for 19-09 with input from the other officials for this precinct