to be able to determine how the impact of voters as far as over-0 vote, under-a vote, who shows up on election day, what are they choosing to participate in. there are two main sources of data, the department of elections has a lot of data on their website. a lot of their data in recent years is much more informative than what they provided an older years, so it is harder to compare rank choice voting out compared with the old december runoff's we used to have because the data available back then is not quite as in debt as the data is today. what i am doing is trying to figure out the best way to present the information, but still have available to us. we have four types of voting systems, rank choice voting, used for all city offices, plurality voting, used for most other offices, the multi- candidate voting, which is the school board, and humidity

college board, and then the simple yes/no voting, which are used on ballot measures and some judges racist. we try to figure out -- and some judge races. we're trying to compare the current systems, how people vote in november, compared with how many people voted under the old run off system, which is where i first target. the information is not as thorough as the old days. before rank choice voting got put into place, the department of elections only tracked if somebody voted and the ballot was counted, and all of the people who did not have their ballots counted either because they under voted or over voted. i put together the data. that should be noted that some of the races are won candidate fields. when i am comparing elections, i always ignore one candidate fields. people look at that and say, i

am not voting in this race, there is only one person to vote for, so why bother checking the box. you get a very depressed turnout, and that shows up in the data. a couple of things that become, apparent in the next slide, in the mayor's race, there was the 99 election for the mayor where the right in canada it was the second-highest voter. up and lot of people were doing right-in, and their ballot was not counted, or somebody said there is a very popular mayor, not really anybody else on the ballot, so you had multiple factors. that is one thing that skewed the results. when i get to processing that, i look at it in both ways. the assessor reporters race is usually done an even number years, but there was a special election to fill a vacancy of an appointed incumbent. that was done in an eye on your

election, so those numbers are also skewed a little bit. -- that was actually done in an odd year election, so those numbers are also skewed a little bit. you kind of gett column what is going on at the city wide level. one thing i want to note, the next row over, if you compare open seats to open seats, there is nothing to compare, because there is only been one of open seat election, and that was the mayor's choice election, under the rank choice voting system. there was no way to really do with their comparison. so you cannot compare those, but i did compare rank choice voting, incumbent verses incumbent. rank choice voting is a better differential. those who are shutting out to vote and get their vote counted have a higher percentage of that amount.

something i want to point out, the city attorney's race, very high in the first column run off, 12%, partly because there was an open seat in the runoff, and you have never had a net and seat in the rank choice voting, so its use those numbers. i did the exact same thing with all of the supervisor races. the 2000 elections, there were some minor changes because of redistricting. there was not very much, it was not very large, so i did not view that as a major issue. the same type of comparisons, all races, regardless of run off, incumbents, or open seats, and then just compare open seats. it is interesting that all of the open seat races that you have the ability to compare, all of them are rank choice voted in favor. even incumbent against

incumbent, rank choice voting when possible is the way to get a better level of participation in the election. then what i want to after that was looking at some of the discussion that has been going on in the city. there is a piece of legislation introduced at the board of supervisors about eliminating rank choice voting from the citywide races. so possibly creating a scenario where you have a primary, and the get a certain percentage of the vote, 65%, you get an automatic win, you don't have to run in november. the allotted to look like that. . i wanted to look at that. the reality is we have never had a primary general in the way that is described for city offices. so i just looked at the general statewide and federal offices. what you get is a primary general election. no place in the country could i find 65% rule winner. it is either 50% or the top two

vote getters move on to the general election when it is a non-partisan seat like ours. when you look at the chart in front of you, if you had a 50% rule, how many of those people would have been the winning person, what percentage would have gotten in november? the highest is 1998, were you got almost 40% of the vote, but you would have gotten 50%. under the 65% rule, there is only one election since 1990 when you have a 65% winning margin. one of the other interesting tidbits, in 2008, we had a presidential primary in february, the regular primary in june. there is a little differential. as soon as you take the presidency of the election, it dramatically drops the number of people who are voting in the election and the primary. having that top ticket item to draw people out is what impacts

how many people showed up more than the local offices. chairperson campos: a question on that? vice chairperson avalos: so you are inferring that 65% threshold is more arbitrary? >> it does not exist anywhere else that i could find, any other major city across the country. i cannot say that it does not exist anywhere because there are a lot of elections and i have no ability to check every single one. vice chairperson avalos: so we are comparing the% of the general foote, 65% of the winner in the primary, is equal in almost all cases, except one, 1998, something much less than the majority? >> correct. vice chairperson avalos: so this whole idea about 65% being the sweet spot number that says

something does not really say as much as it intends to? >> based on the city's primary system that has been in existence since 1990, i would say yes. some cities do something similar. some of our collections are down on off years, so voter turnout is different. -- some of our collections are done on off years, so voter turnout is different. let me get to that and that my answer part of your question. vice chairperson avalos: what if we were to make the threshold 80%? would that help us get to a 50% majority? >> i would assume that there may be some that might creep over the 50% mark. vice chairperson avalos: i just don't know if the 65% signed makes any sense, based on this data, at all. >> correct. getting to the city offices, out

of the top 20, we looked at the top 20 cities to determine whether there are systems for voting for mayor. three of the top 20 outside of san francisco had enough data that we couldn't collect it and inform ourselves, los angeles, detroit, and charlotte. in los angeles, which you will notice, they have 50% and you win in the primary, and you do not go on to the second collection, and they are actually in a march, april, may, june that scenario, separate from any other time or anything else on the ballot. in those cases, the 50%/65% rule could work out because their voter turnout tends to be lower and there is a much core group of people showing up, and the fact that you could win at wrightwood push more people to shop. detroit, there is a non-partisan race, similar to last august,

november, and in need of their races is there a 65% rule in the august winning numbers to dictate a winner in november, using the same number of folks. and in charlotte, to have a partisan seat, so their primary is very much reduced. they did not get anywhere close to having a 65% winner. chairperson campos: has any reasoning been articulated as to 65% being proposed? >> i have not asked any of the board members why they picked that number. chairperson campos: okay, thank you. >> moving on from that, there is also discussion about there's a lot more over votes in certain neighborhoods than others under rank choice voting. one of the things i wanted to do was look neighborhood by neighborhood comparisons. the data is available to go back

to 2008 to do a quick and reasonable analysis. we went back to the last three years and compared the four different styles of footage we had in the city. i did ballot voting in 2011. i have not gone further back than that at this point. errors occur in every election. one of the other things that i want to know is the state assembly and u.s. representatives, in order to make the city wide to cover all of the neighborhood's equally, i emerged those elections together for san francisco voters because they are elected in the same year, so voting patterns should be basically the same. i could have done that in the board of supervisors race, but i did not because it crossed years and i did not feel that was appropriate because you have different types of turnouts in those years that could impact stuffed differently and skewed numbers in a way that was not viable for the system. that is where we get to the next number, where we have how many

errors occur on the ballots by neighborhood. even in a ballot measure, yes/no issue, while there is a small percent of the vote, there is still a% of the vote that messes up and does not have their ballots counted. i then went out into the plurality races, supporting statewide from local races. just a bit of an idea of what was going on. there are errors that increase. city-wide it, rank choice voting races have 6.5% of the vote is actually an error on average among the four races available to us. what is interesting, that is not the race the ec the most number of errors. school boards and community college board elections actually produce the highest number of errors on the citywide average. in this case, there were only three elections -- the committee college board and one of the years had three candidates running for three seats to fill.

i looked at that as a one candidate, one seat race to fill. the numbers were so low, there were still some over votes, but it was not a high number, so i left that out under the same reasoning as one can it also does not the same type of voting patterns in the system. then, in order to say 0.19%, 0.48%, what does that mean in comparison? i took that as a percent, the citywide average of that neighborhood. for example, bayview hunters point, the first on the list, under rank choice voting, they have 82.75% error rate higher than the city average, but under the plurality system for local offices, they are at 108% above averages. statewide there are 79.2%. for college board races, they

are 160%. you go down the list and determine where they are in comparison to different types of election styles, where they are. there are more over votes that occur in the rank choice voting races than the plurality races. so it is not the exact same number, but it is a way of trying to do the comparison. our people messing up at a higher rate than other collections? and then doing a quick analysis , the department of elections has 26 neighborhoods, and all of these there were 10 that were above average in r.c.v. races and in looking at those 10, i found that six of them, the rank choice voting was closer to 0 than the other races occurred and this was stuff i figured out before the meeting so i don't have it in a slide. there were a couple of races

where they were in the middle and there were two races where they were lower than the plurality races. so rank choice voting, their percent is closer to average under rank choice voting than it is the other systems. now, the ones below average, those get closer to the average so their error rate as a comparison goes up in some ways. so that is the entire presentation. happy to take any questions you may have. >> i think it's important to provide the additional analysis in terms of what happens in terms of the over voting in rank choice voting as opposed to other systems because i think the way it's been presented here, it's to suggest that there is something inherently different about the rank choice voting system that over-voting happens in certain communities. the fact is that you have over-voting happening irrespective of what system you

use. >> correct. >> there might be differences but i think it's good to provide that information because, you know, to the extent that there is over-voting that takes place, it goes beyond what system you use so it's not necessarily unique to rank choice voting and i think it's important to note that. >> that is very correct. getting into the voting, how people actually vote actually does also make a difference, as well. there are systems out there that we're not currently using in the city but are potentially available and some need approval at the state level from the. secretary: but you could potentially reduce voting errors at least for those that show up day of voting by having a different machine you use at the polling place. instead of having them fill out a card, if you have a touch screen, you could have zero errors because your touch screen would not allow the error at the polling location. >> i don't know if my colleagues have comments or questions but in terms of over-voting,

comparing what happens in places like oakland, is that less of a problem in those -- in oakland than it is here? >> i haven't had as much time to look at outside data from other areas. i know corey cook did an analysis on this in his report of the 2010 race in comparing stuff, you know, between the oakland mayor's race. they had a little lower error rate but i don't know what their overall error rate is for other types of elections, i haven't gone to that level to determine if they're making errors similar to us. if there are errors occurring in the system, the question is, is there different ways of educating folks on how to vote correctly in the election and are their ways to try to catch and stop the over-vote from occurring. at the polling location, you change your machines to a touch screen, you can virtually get rid of all errors from at least your polling location voters. they can't make an error because your touch screen, if you set it up properly, will stop them from making that error. one of the other things i

haven't put in a slide yet is also the issue of the three choices that you get. the more candidates you have in the field, what you will actually find is the more people have their ballot disenfranchised or thrown out because they voted for three people but didn't vote for one of the top two so in the mayor's race if i remember correctly off the numbers off the top of my head, i think it was 16% of the people that voted had their vote thrown out because they voted for three people but they did not vote for the top two finishers. chairperson campos: commissioner avalos? vice chair avalos: thank you. using the word disenfranchised, though, is problematic for me. they actually voted. >> correct. vice chair avalos: the vote was counted. >> correct. vice chair avalos: just the person they voted for didn't get elected. >> what i would argue is the system didn't allow them to choose as many people as they may have wanted to. vice chair avalos: in some cases

people didn't want to choose more? >> there are cases where you have people who only chose one person. that was their choice to vote for only one and therefore no one would consider that an error or a problem within the system but there are some people and i'm one of them, i voted for multiple candidates in a race and my ballot -- i'm not going to say who i voted for but my ballot may or may not have been thrown out based on the fact that i only had three choices and if i had four, six, seven or more, i may have gotten to the final two and had a final choice in the matter so maybe disenfranchise is a harsh term and one i will watch using in the future but i will say that the system makes it so people can have their ballots, if not given enough choices, could have their ballot not counted in the election. there are ways to make sure more people choose who is the winner in the end is what i'm trying to get to. vice chair avalos: based on your findings and your research, do you see that you have any way of

explaining the assertions that were made throughout last year and in the media, as well, often reported in the media, columnists and reporters and sometimes editorials stating that rank choice voting is confusing based on your research, could you explain why people might say that it's confusing? >> i couldn't necessarily get into why people would think it's confusing because this is looking at a statistical analysis of who's shown up and voted and cast a ballot, not getting into the mental state of is someone not voting because they're confused or is someone voting wrong because they're confused? it gets into the voting wrong because we can look at the over-votes because there's another way to look at over-votes which is not included in what i'm doing but i'm planning on looking at it, as well, is how many over-votes are there somewhere on the ballot,

you theoretically could have chosen correctly in column one but had an over-vote in column two but because column one was eliminated, your over-vote is not acknowledged in the way the system is currently reported on but corey cook studies how many people made an error. if memory serves me, it's 1.2 person of all people cast a ballot in the 2012 mayor's race, had an over-vote somewhere on their ballot. not all those people have their ballot disqualified for the over-vote. they may have voted for the top person in their first choice and never had the over-vote shown in the numbers. vice chair avalos: in your findings, i'm not sure if you can say it or corey cook, either of you say that number of over-votes is statistically significant? >> it depends on your level -- do you consider 1% of the vote being statistically significant or not. vice chair avalos: comparing rank choice voting to other types of voting? >> if you're looking -- if you go back to the raw data of the

average number of over-votes per race, you know, in the plurality local races you get .09% over-vote on average since 2008. in the statewide races, you get .12% and in the r.c.v. races you get 2.6%. so you're getting twice as many as your state races are getting so the question is, is that statistically relevant? that's coming into a determination, what level is statistically relevant or not? i personally don't want to make that call right now saying that you could say twice as many and that's statistically relevant or you could say it's .14%, is that going to change any election in the system? a very small percentage could have had a major impact on how one of the supervisor's recent races went because it was a very close race all the way through but would .14% of the turnout made a difference in who came out and became the winner? chairperson campos: thank you.

commissioner olague? commissioner olague: i was wondering if you could -- if you might analyze some of these findings against the measure that -- >> i looked at it a little bit but if you want me to go more deeply into the measure i'm happy to do that and report next month. commissioner olague: yeah, i would. because at the time i thought, well, some of these changes might be considered, i didn't have the advantage of having this information in front of me nor was i aware at the time that there was the study that was in play or in progress, so i would like to, you know, to see -- >> i'm more than happy to do that. with the commission's permission, what i might suggest doing is having me go speak -- if the supervisors are available that co-sponsor, to get some understanding on what's in the measure and why they did it the way they did, go talk with them, share with them what i found so far and see if there's anything they might have questions on.

chairperson campos: i think that would be helpful. commissioner olague: that would be useful, yeah. >> i did notice, the way the ballot measure is written, there is one issue that i don't think was an intended consequence that might occur if the ballot measure stays the way it is and that's, there are two elections for statewide office that are elected in even number years, the recorder and the public defender. the way the ballot measure, the way i read it right now, is that there would be a september, november election, in even numbered years, you already have a june primary so we could potentially be seeing a june primary, a september primary for two offices and then a november election for to make the final decision on everything so i don't know if that was an unintended consequence where they were thinking we had 2011 and there was no primary earlier on in the year. that's one of the things i wanted -- a red flag that raised in my head, i don't think that the intent was to have three

elections but obviously i will bring that up with the co-sponsors of that legislation if i get a chance to talk with them. chairperson campos: thank you very much. by the way, one note that i will make is that i don't think that the data that we have before us backs up the claims that have been made in terms of rank choice voting disenfranchising certain communities. that's why i think what commissioner avalos was saying is very true which is we have to be careful about how we use that term because the fact is that over-voting and mistakes happen in just about any system you use and they happen more in some communities than others, and so i think we have to be very careful when we make those kinds of claims and that's why, you know, i think that i always felt that having an in-depth analysis of the data is the best way to deal with these kinds of

assertions because i don't think in my view, i don't think the data supports that, but i do think that the further analysis would be helpful. >> ok. is there anything -- i mean, i get the wanting to look at the legislation and anything that might be in there and double checking that and seeing if there's anything but is there anything else off the top of your heads at least right now that comes that you would want to see me look at in more depths than what i've already done so i can incorporate that into the stuff i'm already working on chairperson campos: i would leave it up to the commissioners. commissioner avalos? vice chair avalos: i'm not sure if it's possible but are you able to compare errors of voting on particular voters, maybe there's a pattern of particular voters that are prone to -- in terms of like a single voter repeatedly is doing something wrong, we could look at perhaps one person hasn't been explained the proper way of voting or how

the system works and that could lead us to, you know, understanding of greater outreach and public education on how to vote would be more effective. >> in the 2011 election cycle and corey cook looks into this a little bit so some of his data provides this for us. prior to 2011, every individual office was not attached to any other individual office so if someone made an error in one race, you didn't know what other race they may have made an error in. in 2011, if are whatever reason, the way the date was a released, while we don't know the individual that voted so we don't know, demographically speaking, who they are, although you can look at the precinct where is that's occurring and look at the demographics of that precinct and make a reference off of that but you can tell in the 2011 cycle how many people made an error anywhere on their ballot as far as the three races that occurred for sheriff because there's some way that the system attaches them together. prior to 2011, i don't believe that data actually exists. you can only look at the

individual race and say there were x number of errors in x number of precincts but you don't know if they were made across across the board and cory cook's information gets into that a little bit so i can flesh that out for you. >> there are experts in this field we can have talk to you, like, corey is one or even personnel from the department of elections about the kind of data that they can produce and what kind of information you can statistically receive from that. vice chair avalos: great. just getting to the point of, do we -- if we looking at remedies to make voting more accurate, are we able to understand what they could be in terms of proper education, what type of information people need, what type of explanation. >> if you want me to get into that, i'm happy to do so. my original instructions were to stay more on the statistical side than on the analytical side

of the issue. >> and i would second that, that, really, in terms of his position here, the expertise -- that's what i'm talking about, expertise of someone that actually does that kind of work for a living, that's the kind of -- because other than bringing you the statistic, you're asking for a personal opinion. if you prefer to have somebody that actually deals with that for a living and can tell you what market research has shown or things, perhaps we can do that because i don't want to the put mr. fried in a position of trying to give you an opinion about what the data is saying other than just providing the data to you. vice chair avalos: ok. i'll follow up with mr. fried about that, too. thank you. chairperson campos: and one suggestion is that depending on where the additional information that mr. fried collects gets us, it may be that, commissioners, that you decide we want to go and get an expert to come out and analyze the