To: Aspen City Council
From: Millard Zimet
Re: Bullet Voting in the May 2009 Election
Date: March 15, 2010
________________________________________________________________________
The purpose of this memo is to observe the actual voting data generated by Aspen’s May 2009 Mayoral and City Council IRV elections with regards to bullet voting, and to observe whether bullet voting impacted the outcomes of those elections. This memo is expressly not to be construed as an attack upon any person, elected official, candidate, government official or entity, political party, or political persuasion.
The data in this memo is derived from the strings, which I have previously sent to you in Excel format. If you want to replicate the data contained in this memo, please take those strings and use the “filter” function in Excel to sort the data.
This was an election to fill the position of Mayor, and there were four candidates. There were 2528 ballots cast, and 541 ballots (21.4%) named only one candidate and left the rest of the choices blank. So about one out of every five voters cast their ballot using the bullet voting pattern.
Table 1 shows how those 541 bullet votes were distributed among the candidates, and shows each candidate’s share of the bullet voting pool.
Table 1
Bullet Votes Share
Ireland 322 59.5%
Marks 136 25.1%
Erspamer 58 10.7%
Kole 18 3.3%
Write-In 7 1.3%
==== ======
Total 541 100%
Table 1 shows that about six out of every ten bullet voters in the Mayoral election bullet voted for Ireland. Ireland’s margin of 186 bullet votes over Marks by itself exceeded the number of bullet votes received by any other candidate (Mayor or Council races) in the May 2009 election.
The strings also allow one to observe the knowledge level of each candidate’s supporters (i.e., which supporters knew to bullet vote). For this purpose I'm defining a "supporter" as a voter who gave a given candidate that voter's first place vote, and I’m measuring the percentage of each candidate’s first place votes that were derived from bullet votes; for purposes of this memo I’m going to call that percentage a given candidate’s “Bullet Voting Percentage”, and it shows which supporters knew to bullet vote.
Table 2 shows each candidate’s number of first place votes, number of bullet votes, and Bullet Voting Percentage.
First Place Votes Bullet Votes Bullet Voting %
Ireland 1089 322 29.57%
Marks 877 136 15.51%
Erspamer 421 58 13.78%
Kole 126 18 14.29%
Table 2 shows that Ireland’s Bullet Voting Percentage of 29.57% is approximately twice as high as that of the other three candidates, indicating that his supporters were much more likely to know to bullet vote.
The strings also allow one to observe how each precinct produced bullet votes. Table 3 shows, for each precinct, the number of ballots cast, the number of ballots that were bullet voted, the percentage of ballots within each precinct that were bullet voted, and that precinct’s share of the total bullet voting pool.
Table 3
Precinct Total Ballots BV Ballots % BV in Precinct Share
1 291 75 25.8% 13.9%
2 305 54 17.7% 10.0%
3 314 46 14.6% 8.5%
4 585 121 20.7% 22.4%
Early 794 198 24.9% 36.6%
Mail 239 47 19.7% 8.7%
===== ==== =====
Total 2528 541 21.4% 100%
Table 3 shows that about one out of every four voters at Precinct 1 and at the early voting station at City Hall knew to bullet vote. It also shows that the City Hall ballot box accounted for 36.6% of the bullet voting pool.
Table 4 shows how the bullet votes were distributed among the candidates in each precinct.
Table 4
Precinct 1 Precinct 2 Precinct 3 Precinct 4 Early Mail Total
Ireland 39 32 36 65 130 20 322
Marks 20 12 6 34 47 17 136
Erspamer 13 4 4 17 16 4 58
Kole 1 5 0 3 4 5 18
Write In 2 1 0 2 1 1 7
=== === === === === === ===
=== === === === === === ===
Total 75 54 46 121 198 47 541
Table 4 shows that at the City Hall early voting station Ireland received 130 bullet votes, or 65.7% of all bullet votes cast at that precinct. Ireland received 374 first place votes from the early voting station, and so Ireland had a 34.8% Bullet Voting Percentage at that precinct.
II: The Council Election
This was an election to fill two available Council seats, and there were nine candidates. There were 2487 ballots cast, and 518 ballots (20.8%) named only one or two candidates and left the rest of the choices blank; this was the bullet voting pattern predicted by FairVote in their analysis dated January 20, 2009. So once again about one out of every five voters cast their ballot using the bullet voting pattern.
For purposes of this section of this memo, a “bullet voted ballot” means a ballot that was cast for only one or two choices and left all other choices blank. Likewise, a “bullet vote” means a vote cast in the #1 or #2 position on a given bullet voted ballot.
Table 5 shows each candidate, the number of bullet votes they received, how the 931 total bullet votes were shared by percentage among the candidates, and the frequency with which a given candidate was named on any given bullet voted ballot (ie bullet votes per candidate divided by 518 bullet voted ballots).
Table 5
Bullet Votes Share Frequency
Torre 177 19.0% 34.2%
D. Johnson 175 18.8% 33.8%
J. Johnson 137 14.7% 26.4%
Frisch 129 13.9% 24.9%
Behrendt 119 12.8% 23.0%
Kasabach 92 9.9% 17.8%
Wampler 62 6.7% 12.0%
Speck 21 2.3% 4.1%
Lasser 19 2.0% 3.7%
=== ====
Totals 931 100%
Table 5 shows that the two candidates who received the most bullet votes and were named most frequently on bullet voted ballots were also the two winning candidates.
Turning now to Table 6, and Bullet Voting Percentage, one can observe which candidate’s supporters knew to bullet vote. For purposes of this analysis a “supporter” is a voter who gave a given candidate their first place vote.
In Table 6 the first column shows the total number of first place votes cast for each candidate. The second column shows the number of each candidate's first place votes that came from bullet voted ballots. The third column shows the percentage of each candidate's first place votes that came from bullet voted ballots, or Bullet Voting Percentage.
Table 6
1st Place Votes 1st Place Bullet Votes BV%
D. Johnson 465 87 18.71%
J. Johnson 456 87 19.08%
Frisch 418 98 23.44%
Torre 385 107 27.79%
Behrendt 354 64 18.08%
Kasabach 249 41 16.47%
Wampler 94 22 23.40%
Speck 38 6 15.79%
Lasser 27 6 22.22%
This table shows that Torre decisively benefited from bullet voting, as his supporters knew to bullet vote by a significant margin over the supporters of the other candidates. Another interesting observation is that Derek Johnson received 88 second place bullet votes, which was more than his number of first place bullet votes; Jackie Kasabach also received more second place bullet votes than first place bullet votes.
Regarding precinct distribution, Table 7 shows, for each precinct: (i) the number of ballots cast at that precinct; (ii) the number of bullet voted ballots cast at that precinct; (iii) the % of ballots at a given precinct that were bullet voted; and (iv) the given precinct’s share of total bullet voted ballots (518).
Table 7
Precinct Ballots Bullet Voted % Ballots that were BV's Share
1 290 75 25.9% 14.5%
2 300 56 18.7% 10.8%
3 307 46 15.0% 8.9%
4 571 121 21.2% 23.4%
Early 794 178 22.4% 34.4%
Mail 225 42 18.7% 8.1%
==== === ===
Total 2487 518 20.8% 100%
The data in Table 7 is comparable to the data in Table 3. Once again, voters at Precinct 1 were the most likely to bullet vote. And once again by far the most bullet voted ballots were produced by the early voting station at City Hall.
Table 8 shows how bullet votes (a #1 or #2 choice on an otherwise blank ballot) were distributed in each precinct to each candidate.
Table 8
Precinct 1 Precinct 2 Precinct 3 Precinct 4 Early Mail Total
Torre 30 20 24 42 48 13 177
D. Johnson 22 20 16 53 58 6 175
J. Johnson 19 11 12 30 55 10 137
Frisch 21 13 5 33 43 14 129
Behrendt 17 15 6 19 49 13 119
Kasabach 14 7 7 18 39 7 92
Wampler 4 5 9 15 23 6 62
Speck 7 2 0 4 7 1 21
Lasser 2 4 4 3 6 0 19
=== ==== ==== ==== === === === ===
=== ==== ==== ==== === === === ===
Totals 136 97 83 217 328 70 931
Table 8 shows that at the key precinct, the City Hall ballot box, all the major candidates received a substantial number of bullet votes. So while Table 4 shows that in the Mayor’s race there was a concentrated distribution of bullet votes from the City Hall ballot box, Table 8 shows a wider distribution in the Council races.
III: Conclusion
About one out of every five voters in each election bullet voted. The winning candidates in each election were the ones who received the most bullet votes. The early voting ballot box at City Hall in each election produced by far the most bullet voted ballots. The level of bullet voting knowledge of candidate supporters in each election was decidedly uneven, as some supporters knew to bullet vote a lot more than other supporters. One must conclude that bullet voting impacted the outcome of each election.
While the Mayoral election was a single seat election and the Council election was a two seat election, and so the metrics are a bit different for each election, there are three common measurements that can be applied to all the candidates, namely: (i) number of bullet votes; (ii) percentage share of bullet votes a given candidate received in his/her election; and (iii) Bullet Voting Percentage (a/k/a supporter level of bullet voting knowledge). Table 9 shows those three measurements for all of the candidates.
Table 9
Bullet Votes Share BV%
Ireland 322 59.5% 29.57%
Torre 177 19.0% 27.79%
D. Johnson 175 18.8% 18.71%
J. Johnson 137 14.7% 19.08%
Marks 136 25.1% 15.51%
Frisch 129 13.9% 23.44%
Behrendt 119 12.8% 18.08%
Kasabach 92 9.9% 16.47%
Wampler 62 6.7% 23.40%
Erspamer 58 10.7% 13.78%
Speck 21 2.3% 15.79%
Lasser 19 2.0% 22.22%
Kole 18 3.3% 14.29%
No comments:
Post a Comment