Quantitative Literacy
Quantitative Literacy: Thinking Between the Lines
Third Edition
Chapter 8
Voting and Social Choice
© 2018 W. H. Freeman and Company
Lesson Plan
Measuring voting power: Does my vote count?
Voting systems: How do we choose a winner?
Fair division: What is a fair share?
Apportionment: Am I represented?
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (1 of 31)
Voting Coalition: A group of voters who vote the same way
Winning Coalition: A set of voters with enough votes to determine the outcome of an election; otherwise it is a Losing Coalition
Quota: The number of votes necessary to win the election, in a voting system
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (2 of 31)
Example: Suppose there are three delegates to a county convention: Abe has 4 votes from his precinct, Ben has 3 votes, and Condi has 1 vote. A simple majority of the votes wins.
What is the quota?
Make a table listing all of the coalitions of voters. Designate which of them are winning coalitions.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (3 of 31)
Solution:
There are 8 votes, so the quota for simple majority is 5 votes.
The following table shows all possible coalitions and votes each coalition controls. The last column indicates whether the coalition controls a majority of the votes and is a winning coalition.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (4 of 31)
| Number of Votes: 4 | Number of Votes: 3 | Number of Votes: 1 | Total Votes | Winning Coalition? |
| Abe | Ben | Condi | 8 | Yes |
| Abe | Ben | 7 | Yes | |
| Abe | Condi | 5 | Yes | |
| Abe | 4 | No | ||
| Ben | Condi | 4 | No | |
| Ben | 3 | No | ||
| Condi | 1 | No |
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (5 of 31)
Critical voter: A member of a winning coalition is a critical voter if the coalition becomes a losing coalition when that voter is removed.
Example: The table below contains information from the three winning coalitions in the previous example.
| Number of Votes: 4 | Number of Votes: 3 | Number of Votes: 1 | Total Votes | Winning |
| Abe | Ben | Condi | 8 | Yes |
| Abe | Ben | 7 | Yes | |
| Abe | Condi | 5 | Yes |
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (6 of 31)
In the first coalition: Abe is the critical voter.
In the second coalition: both are critical voters.
In the third coalition: both are critical voters.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (7 of 31)
The previous information can be summarized in the following coalition table:
| Number of Votes: 4 | Number of Votes: 3 | Number of Votes: 1 | Total Votes | Winning Coalition? | Critical Voters |
| Abe | Ben | Condi | 8 | Yes | Abe |
| Abe | Ben | 7 | Yes | Abe, Ben | |
| Abe | Condi | 5 | Yes | Abe, Condi | |
| Abe | 4 | No | Not applicable | ||
| Ben | Condi | 4 | No | Not applicable | |
| Ben | 3 | No | Not applicable | ||
| Condi | 1 | No | Not applicable |
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (8 of 31)
Winning Coalition and Critical Voters
A set of voters with enough votes to determine the outcome of an election is a winning coalition. A voter in a winning coalition is critical for that coalition if the coalition is no longer a winning one when that voter is removed. We can summarize the essential information about coalitions in a coalition table.
Counting coalitions: Number of coalitions
For n voters, there are 2n - 1 possible coalitions (each of which includes at least one voter).
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (9 of 31)
Example: If there are 7 voters, there are 27 possibilities for voters to be in or not in a coalition. This includes the possibility of all the voters not being in any coalition, so there are 27 - 1possibilities.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (10 of 31)
Banzhaf power index: The number of times a voter is critical in a winning coalition divided by the total number of instances in which any voter is critical. Expressed as a fraction or percentage.
Example: Use coalition Table 8.1 to compute the Banzhaf index for each county convention delegate described in the previous example.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (11 of 31)
Solution: From the coalition table: overall there were 5 instances in which any voter was critical. So the Banzhaf power index of a voter is the number of times that voter is critical divided by 5.
Abe was critical 3 of the 5 times, so he has a Banzhaf power index of 3/5 or 60%.
Ben was critical I of the 5 times, so he has a Banzhaf power index of 1/5 or 20%.
Condi was critical I of the 5 times, so she has a Banzhaf power index of 1/5 or 20%.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (12 of 31)
Example: In the first round of voting at the 1932 Democratic National Convention, Franklin D. Roosevelt received 666.25 votes, Al Smith received 201.75 votes, John Nance Garner received 90.25 votes, and the other seven candidates combined received 195.75 votes. These are referred to as voting blocs. It is common for the nomination to be decided by negotiations among the various blocs. There were 770 votes required to win the nomination.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (13 of 31)
What is the quota?
Determine the winning coalitions.
Determine the critical voting blocs in each winning coalition.
Determine the Banzhaf index of each bloc.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (14 of 31)
Solution:
Because 770 votes were required to win the nomination, that is the quota. Note that this number is considerably more than a simple majority of the 1154 votes cast.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (15 of 31)
The following table lists the 24 – 1 = 15 possible coalitions.
| 666.25 | 201.75 | 195.75 | 90.25 | Total votes | Winning coalition |
| R | S | O | G | 1154 | Yes |
| R | S | O | 1063.75 | Yes | |
| R | S | G | 958.25 | Yes | |
| R | S | 868 | Yes | ||
| R | O | G | 952.25 | Yes | |
| R | O | 862 | Yes | ||
| R | G | 756.5 | No | ||
| R | 666.25 | No |
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (16 of 31)
| 666.25 | 201.75 | 195.75 | 90.25 | Total votes | Winning coalition |
| S | O | G | 487.75 | No | |
| S | O | 397.5 | No | ||
| S | G | 292 | No | ||
| S | 201.75 | No | |||
| O | G | 286 | No | ||
| O | 195.75 | No | |||
| G | 90.25 | No |
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (17 of 31)
The accompanying table lists the winning coalitions only, along with the critical voting blocs in each case.
Winning Coalitions Only
| Votes: 666.25 | Votes: 201.75 | Votes: 195.75 | Votes: 90.25 | Total votes | Critical voters |
| R | S | O | G | 1154 | R |
| R | S | O | 1063.75 | R | |
| R | S | G | 958.25 | R,S | |
| R | S | 868 | R,S | ||
| R | O | G | 952.25 | R,O | |
| R | O | 862 | R,O |
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (18 of 31)
There are 10 instances in which a voting bloc is critical:
Roosevelt is critical in 6 of these 10 instances. Therefore, this Banzhaf index is 6/10 = 60%.
Smith is critical in 2 of these 10 instances. Therefore, this Banzhaf index is 2/10 = 20%.
"Other" is critical in 2 of these 10 instances. Therefore, this Banzhaf index is also 2/10 = 20%.
Garner is not critical in any of these 10 instances. Therefore, this Banzhaf index is 0%.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (19 of 31)
Banzhaf power index: A voter’s Banzhaf index is the number of times that voter is critical in some winning coalition divided by the total number of instances in which any voter is critical. The index is expressed as a fraction or as a percentage.
Swing voter: Suppose the voters vote in order and their votes are added as they vote. The swing voter is the voter whose vote makes the total meet the quota and thus decides the outcome. Which is the swing voter depends on the order in which the votes are cast.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (20 of 31)
Example: Members of the European Union have votes on the Council determined roughly by a country’s population but progressively weighted in favor of smaller countries. Ireland has 7 votes, Cyprus has 4 votes, and Malta has 3 votes. Suppose that these three countries serve as a committee where a simple majority wins, so the quota is 8 votes. Make a table with all the permutations of the voters and the swing voter in each case.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (21 of 31)
Solution: There are n! different permutations of n objects, where
𝑛!=𝑛×(𝑛−1)×…×1
So the 3 objects have 3!=3×2×1=6 permutations. The following table lists the permutations and swing voter in each case.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (22 of 31)
| Order of Voters | Order of Voters | Order of Voters | Swing Voter |
| Ireland (7) | Cyprus (4) | Malta (3) | Cyprus |
| Ireland (7) | Malta (3) | Cyprus (4) | Malta |
| Cyprus (4) | Ireland (7) | Malta (3) | Ireland |
| Cyprus (4) | Malta (3) | Ireland (7) | Ireland |
| Malta (3) | Ireland (7) | Cyprus (4) | Ireland |
| Malta (3) | Cyprus (4) | Ireland (7) | Ireland |
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (23 of 31)
The Shapley-Shubik power index: is calculated as the fraction (or percentage) of all permutations of the voters in which that voter is the swing.
Example: Compute the Shapley-Shubik power index for the committee of Ireland, Cyprus, and Malta from the previous example.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (24 of 31)
Solution: There are six permutations of the voters. Ireland is the swing in 4 of the 6 cases so the index for Ireland is 4/6 = 2/3 or about 66.67%.
Cyprus is the swing in 1 of the 6 cases so its index is 1/6 or 16.67%.
Malta also is the swing in 1 of the 6 cases so its index is 1/6 or 16.67%.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (25 of 31)
Example: A few states in the 2016 U.S. presidential election were especially important in determining the outcome. They are referred to as “swing states.” Four of these states were Virginia with 13 electoral votes, Wisconsin with 10, Colorado with 9, and Iowa with 6. Assume that a simple majority vote from only these four states would determine the election.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (26 of 31)
How many permutations of these four states are there?
What is the quota?
Make a table listing all of the permutations of the states and the swing voter in each case.
Find the Shapley-Shubik index for each state.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (27 of 31)
Solution:
The number of permutations of four items is
4! = 4 × 3 × 2 × 1 = 24
There are 38 votes, and it requires a majority to win, so the quota is 20.
The following table shows the different permutations and the swing vote state in each case.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (28 of 31)
| States | States | States | States | Swing |
| VA (13) | WI (10) | CO (9) | IA (6) | WI |
| VA (13) | WI (10) | IA (6) | CO (9) | WI |
| VA (13) | CO (9) | WI (10) | IA (6) | CO |
| VA (13) | CO (9) | IA (6) | WI (10) | CO |
| VA (13) | IA (6) | WI (10) | CO (9) | WI |
| VA (13) | IA (6) | CO (9) | WI (10) | CO |
| WI (10) | VA (13) | CO (9) | IA (6) | VA |
| WI (10) | VA (13) | IA (6) | CO (9) | VA |
| WI (10) | CO (9) | VA (13) | IA (6) | VA |
| WI (10) | CO (9) | IA (6) | VA (13) | IA |
| WI (10) | IA (6) | VA (13) | CO (9) | VA |
| WI (10) | IA (6) | CO (9) | VA (13) | CO |
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (29 of 31)
| States | States | States | States | Swing |
| CO (9) | VA (13) | WI (10) | IA (6) | VA |
| CO (9) | VA (13) | IA (6) | WI (10) | VA |
| CO (9) | WI (10) | VA (13) | IA (6) | VA |
| CO (9) | WI (10) | IA (6) | VA (13) | IA |
| CO (9) | IA (6) | VA (13) | WI (10) | VA |
| CO (9) | IA (6) | WI (10) | VA (13) | WI |
| IA (6) | VA (13) | WI (10) | CO (9) | WI |
| IA (6) | VA (13) | CO (9) | WI (10) | CO |
| IA (6) | WI (10) | VA (13) | CO (9) | VA |
| IA (6) | WI (10) | CO (9) | VA (13) | CO |
| IA (6) | CO (9) | VA (13) | WI (10) | VA |
| IA (6) | CO (9) | WI (10) | VA (13) | WI |
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (30 of 31)
Looking at the first row, Virginia’s 13 electoral votes are not enough to make 20, but Wisconsin’s 10 added to it exceeds 20. Therefore, Wisconsin is the swing voter in this order. Looking at the fifth row, we see that Virginia’s 13 electoral votes plus Iowa’s 6 are not enough to make 20, but Wisconsin’s 10 added to that exceeds 20. Therefore, Wisconsin is the swing voter in this order too.
The Shapley-Shubik Index of a given voter is calculated as the fraction of all permutations of the voters in which that voter is the swing.
© 2018 W. H. Freeman and Company
8.1 Measuring voting power: Does my vote count? (31 of 31)
Virginia: 10/24 or about 41.67%; Wisconsin: 6/24 or 25%; Colorado: 6/24 or 25%; and Iowa: 2/24 or about 8.33%.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (1 of 28)
Voting System: A set of rules under which a winner in an election is determined.
Plurality Voting: The system of voting in which the candidate that receives more votes than any other candidate is the winner.
Example: Of four candidates and 100 votes, what is the smallest number of votes needed to win?
Solution: If the four candidates have an equal number of votes, they would have 100/4 =25 each. So a candidate could have a plurality with as few as 26 votes.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (2 of 28)
Spoiler: A candidate who has no realistic chance of winning but whose presence in the election affects the outcome.
Example: In the 1996 presidential race, the three major candidates were Bill Clinton, Robert Dole, and H. Ross Perot. Here is the outcome for the state of Florida.
| Candidate | Votes |
| Clinton | 2,546,870 |
| Dole | 2,244,536 |
| Perot | 483,870 |
| Others | 28,518 |
| Total votes cast | 5,303,794 |
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (3 of 28)
What percentage of all votes cast in Florida were for Clinton? Did any candidate achieve a majority of votes cast?
Florida determines the winner of a presidential election by plurality, so Clinton was declared the winner of all 25 of Florida’s electoral votes. Let’s suppose that the vote was conducted with only the top three candidates and that all their supporters continued to vote for them. Would it be possible for any of the three to achieve a majority if all the votes cast for “Others” were awarded to Clinton?
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (4 of 28)
It was speculated that by far most Perot voters would have voted for Dole if Perot were not in the race. Let’s suppose that 83% of Perot voters would have voted for Dole and 17% for Clinton. How many votes would Clinton have had and how many would Dole have had? Who would have won the election? Would you call Perot a spoiler in this case?
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (5 of 28)
Solution:
Of the 5,303,794 votes cast, Clinton received 2,546,870 votes, which is 2,546,870/5,303,794 or about 48% of the votes. This is not a majority. Clinton won more votes than any other candidate, so no candidate received a majority of votes cast.
Suppose that all of the 28,518 “Other” votes went to Clinton. Then he would have received 2,575,388 votes. But half of 5,303,794 votes is 2,651,897, so he still would not have had a majority. Certainly the same is true of the other two candidates.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (6 of 28)
If 83% of Perot’s votes went to Dole, then Dole would have received 83% of 483,870 or an extra 401,612 votes, for a total of 2,646,148. Clinton would have received 17% of 483,870 or an extra 82,258 votes, for a total of 2,629,128. Therefore Dole would have won with 17,020 more votes than Clinton. If this were indeed true, then Perot would have been a spoiler.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (7 of 28)
Preferential Voting System: Systems in which voters express their ranked preferences between various candidates, usually with a ranked ballot that is used to avoid several rounds of voting and the voter lists his or her candidate preferences. Two examples follow.
Top-Two Runoff System: If no candidate receives majority, there is a new election with only the two highest vote-getters.
Elimination Runoff System: If no candidate receives majority, the lowest vote-getter is eliminated and a vote is taken again among those who are left. This repeats until a majority is reached.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (8 of 28)
Example: Consider the following ranked ballot outcome for 10 voters choosing among three candidates:
| Rank | 4 Voters | 4 Voters | 2 Voters |
| First Choice | Alfred | Gabby | Betty |
| Second Choice | Betty | Alfred | Gabby |
| Third Choice | Gabby | Betty | Alfred |
Determine the winner under the elimination runoff system.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (9 of 28)
Solution:
No candidate has first-choice majority. Betty has the least so she is eliminated from the first round.
| Rank | 4 Voters | 4 Voters | 2 Voters |
| First Choice | Alfred | Gabby | Betty |
| Second Choice | Betty | Alfred | Gabby |
| Third Choice | Gabby | Betty | Alfred |
With Betty eliminated, the table is now as follows:
| Rank | 4 Voters | 4 Voters | 2 Voters |
| First Choice | Alfred | Gabby | Gabby |
| Second Choice | Gabby | Alfred | Alfred |
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (10 of 28)
Now the first-choice votes are for Alfred and 4 + 2 = 6 for Gabby. In this runoff, Gabby has majority and is the winner.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (11 of 28)
Borda count: A method of ranked balloting that assigns for each ballot: 0 points to the choice ranked last, 1 point to next higher choice, and so on. The Borda winner is the candidate with the highest Borda count.
Example: To decide on food, five friends mark ranked ballots by preference, using the table:
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (12 of 28)
| Pizza | Tacos | Burgers | |
| Ballot 1 | 2 | 1 | 0 |
| Ballot 2 | 2 | 1 | 0 |
| Ballot 3 | 2 | 1 | 0 |
| Ballot 4 | 0 | 2 | 1 |
| Ballot 5 | 0 | 2 | 1 |
Did one of the foods receive a majority for first-choice?
Use the Borda count to determine which food should be ordered.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (13 of 28)
Solution:
First-place votes are indicated by the number 2. There were three first-place votes for pizza, which is a majority of the five first-place votes.
The Borda count is:
for pizza: 2+2+2+0+0 = 6
for tacos: 1+1+1+2+2 = 7
for burgers: 1+1=2
According to the Borda count, the group should order tacos. This is true in spite of pizza receiving majority.
This lends to our understanding that no voting system is perfect with three or more candidates.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (14 of 28)
Example: The three finalists for the 2016 Heisman Trophy follow:
| Player | First-place votes | Second-place votes | Third-place votes |
| Lamar Jackson (Louisville) | 526 | 251 | 64 |
| Deshaun Watson (Clemson) | 269 | 302 | 113 |
| Baker Mayfield (Oklahoma) | 26 | 72 | 139 |
Determine the Borda counts and the Borda winner.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (15 of 28)
Solution:
| Player | First-place votes | Second-place votes | Third-place votes |
| Lamar Jackson (Louisville) | 526 | 251 | 64 |
| Deshaun Watson (Clemson) | 269 | 302 | 113 |
| Baker Mayfield (Oklahoma) | 26 | 72 | 139 |
Borda count for L. Jackson=(526×2)+(251×1)+(64×0)=1303
Borda count for D. Watson =(269×2)+(302×1)+(113×0)=840
Borda count for B. Mayfield =(26×2)+(72×1)+(139×0)=124
Lamar Jackson has the highest Borda count and is the Borda winner.
He was the winner of the Heisman Trophy in 2016.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (16 of 28)
Example: Consider the following ranked ballot outcome for 100 voters choosing from options A, B, C, D:
| Rank | 28 votes | 25 votes | 24 votes | 23 votes |
| 1st Choice | A | B | C | D |
| 2nd Choice | D | C | D | C |
| 3rd Choice | B | D | B | B |
| 4th Choice | C | A | A | A |
Who wins under plurality voting?
Who wins under the top-two runoff system?
Who wins under the elimination runoff system?
Who wins under the Borda count system?
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (17 of 28)
Solution:
Under plurality voting only first-choice picks are considered. In that case, candidate A has the most first-choice votes with 28 out of 100.
For a runoff with only the top two candidates, C and D are eliminated from the table, as below:
| Rank | 28 Votes | 25 Votes | 24 Votes | 23 Votes |
| Adjusted 1st Choice | A | B | B | B |
| Adjusted 2nd Choice | B | A | A | A |
In this runoff, B has 72 first-choice votes; this is clearly a majority, so B is the winner.
© 2018 W. H. Freeman and Company
50
8.2 Voting systems: How do we choose a winner? (18 of 28)
In the first round of elimination runoff, D has the fewest votes and is eliminated. The table follows:
| Rank | 28 votes | 25 votes | 24 votes | 23 votes |
| 1st Choice | A | B | C | C |
| 2nd Choice | B | C | B | B |
| 3rd Choice | C | A | A | A |
Now B has the fewest votes and is eliminated:
| Rank | 28 votes | 25 votes | 24 votes | 23 votes |
| 1st Choice | A | C | C | C |
| 2nd Choice | C | A | A | A |
C wins in this runoff by a majority of 72 votes.
© 2018 W. H. Freeman and Company
51
8.2 Voting systems: How do we choose a winner? (19 of 28)
The Borda count incorporated into the original table:
| Rank | Borda Value | 28 votes | 25 votes | 24 votes | 23 votes |
| 1st Choice | 3 | A | B | C | D |
| 2nd Choice | 2 | D | C | D | C |
| 3rd Choice | 1 | B | D | B | B |
| 4th Choice | 0 | C | A | A | A |
So the Borda count for each candidate is:
A=28×3+25×0+24×0+23×0=84
B=28×1+25×3+24×1+23×1=150
C=28×0+25×2+24×3+23×2=168
D=28×2+25×1+24×2+23×3=198
Hence D is the winner based on Borda count.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (20 of 28)
Common Preferential Systems
Plurality: The candidate with the most votes wins.
Top-two runoff: If no one garners a majority of the votes, a second election is held with the top two vote-getters as the only candidates.
Elimination runoff: Successive elections are held where the candidate with the smallest number of votes is eliminated. This continues until there is a majority winner.
Borda count: Voters rank the candidates first to last. The last-place candidate gets 0 points, the next 1 point, and so on. The candidate with the most points wins.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (21 of 28)
A Condorcet winner is a candidate who beats each of the other candidates in a head-to-head election.
Example: Suppose in an election there are seven voters and three candidates, A, B, and C; the voters’ preferences follow:
| Preferences | 3 voters | 2 voters | 2 voters |
| 1st Choice | A | C | C |
| 2nd Choice | B | B | A |
| 3rd Choice | C | A | B |
Is there a Condorcet winner? If so, which candidate?
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (22 of 28)
Solution:
Find the results of each head-to-head contest:
Consider A and B: there are 3+2=5 voters who rank A over B, and only 2 who rank B over A. So A wins versus B.
The results of the other head-to-head contests are:
A and C: C wins by 1 (4 to 3)
B and C: C wins by 1 (4 to 3)
There is a Condorcet winner―it is candidate C.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (23 of 28)
Example: In an election there are seven voters and candidates A, B, C, and D. The tally of ranked ballots follows:
| Voter: | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| 1st Choice | A | A | B | C | D | A | C |
| 2nd Choice | B | D | A | B | B | D | B |
| 3rd Choice | C | B | C | A | A | B | A |
| 4th Choice | D | C | D | D | C | C | D |
Who wins the plurality system?
Who wins the top-two runoff system?
Who wins in the elimination runoff system?
Who wins the Borda count?
Is there a Condorcet winner? If so, which candidate is it?
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (24 of 28)
Solution:
In a plurality voting, only first choices are considered. Candidate A has 3 votes, B has 1, and C has 2 votes. A has the most, so A wins.
The first-place winner is A. Candidate C is second with 2 votes. In a runoff with A and C, A wins with 5 votes to 2.
Because B and D get only one vote each, they are eliminated and A wins.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (25 of 28)
The Borda count is as follows:
A=3×3+1×2+3×1=14
B=1×3+4×2+2×1=13
C=2×3+2×1=8
D=1×3+2×2=7
So, A wins the Borda count.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (26 of 28)
There is a Condorcet winner. The head-to-head outcomes are as follows:
B beats A (4 to 3)
B beats C (5 to 2)
B beats D (4 to 3)
So, B is the Condorcet winner.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (27 of 28)
The Condorcet winner criterion says that if there’s a Condorcet winner, then he or she should be the winner of the whole election.
The condition of Independence of irrelevant alternatives states: Supposing candidate A wins an election and B loses, and another election follows in which no voter changes their preference concerning A and B, B should still lose to A no matter what happens concerning the other candidates.
© 2018 W. H. Freeman and Company
8.2 Voting systems: How do we choose a winner? (28 of 28)
Arrow’s impossibility theorem: If there are three or more candidates, there is no voting system (other than a dictatorship) for which the Condorcet winner criterion and the Independence of irrelevant alternatives hold.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (1 of 32)
Divide-and-Choose Procedure: One person divides the items into two parts and the other person chooses which part he or she wants.
The Lone-Divider Method applies this procedure to parties of 3:
Person 1 divides the assets into three parts, persons 2 and 3 then choose between the piles.
If they choose differently, person 1 gets the remaining pile.
If they choose the same, person 1 chooses which pile they want and persons 2 and 3 mix the remaining piles and perform the divide-and-choose procedure.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (2 of 32)
Adjusted winner procedure: Two people assign points to bid on each item, assigning a total of 100 points.
Initial division of the items gives each item to the person offering the highest bid.
The division is then adjusted based on the ratio of the bids for each item so that ultimately each person receives a group of items whose bid totals are the same for each person.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (3 of 32)
Example: Suppose my sister and I want to divide an inheritance. The assets consist of a guitar, a jewelry collection, a car, a small library, and a certain amount of cash. To start the procedure, each of us takes 100 points and divides those points among the assets. In a division of assets, point values of two siblings follow:
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (4 of 32)
| My Points | Item | Sister’s Points |
| 35 | Guitar | 10 |
| 10 | Jewelry | 10 |
| 20 | Car | 40 |
| 15 | Library | 10 |
| 20 | Cash | 30 |
Initial round: I get the guitar and the library for a total of 50 points; the sister gets the car and the cash for a total of 70 points.
The tied item, the jewelry, goes to the point leader, my sister.
So she is 80% satisfied while I am only 50% satisfied. The points must be adjusted.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (5 of 32)
To make the values even, my sister must share some of her property. Each item my sister won can be calculated as:
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (6 of 32)
Note my sister’s bid goes on top because she is the point leader.
Arrange the ratios in increasing order:
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (7 of 32)
Items are then transferred until an item changes the point leader.
First the jewelry is transferred to me: my expended points is 60 and my sister’s is 70.
The next item, the cash, is the Critical Item; it changes the point leader. Just enough of the critical item is transferred to make the score come out even.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (8 of 32)
The equation for dividing the critical item is as follows:
My score+𝑝 percent of my cash bid=Sister’s score –𝑝 percent of her cash bid
This gives the equation: 60+20𝑝=70–30𝑝 or 𝑝=0.20.
So I get the guitar, the library, the jewelry, and 20% of the cash.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (9 of 32)
My sister gets the car and the remaining 80% of the cash.
The value scores are:
My total points: 35+10+15+20×0.20=64
My sister’s total points: 40+30×0.80=64
Both are 64% satisfied.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (10 of 32)
The Adjusted Winner Procedure: Each of two people makes a bid totaling 100 points on a list of items to be divided.
Step 1: Initial division of items: Each item goes to the higher bidder. Tied items are held for now. The higher score is the point leader, the lesser score is the trailer.
Step 2: Tied items: Tied items go to the leader.
Step 3: Calculate leader/trailer ratios: For each item belonging to the leader, calculate the ratio:
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (11 of 32)
Step 4: Transference of some items from leader to trailer: Transfer items from leader to trailer in order of increasing ratios as doing so does not change the lead. The item to change the lead is the critical item.
Step 5: Division of the critical item: Give p percent of the critical item to the trailer from leader with the following equation:
Trailer’s score + 𝑝 × Trailer’s bid = Leader’s score – 𝑝 × Leader’s bid
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (12 of 32)
Example: Divide the following items from an inheritance:
| John | Item | Faye |
| 50 | Vacation condominium | 65 |
| 20 | Red 1962 GT Hawk | 15 |
| 20 | Family silver set | 15 |
| 10 | Dad’s Yale cap/gown | 5 |
Solution: Initially John gets the Hawk, the silver, and the cap and gown. Faye gets the condominium.
Faye is the leader; her condo is the critical item so is considered for the transfer.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (13 of 32)
John’s score+𝑝×John’s bid=Faye’s score –𝑝×Faye’s bid
John takes 13% of the condo from Faye.
Faye gets 87% ownership of the condo.
John gets the Hawk, the silver, the cap/gown, and 13% ownership of the condo.
The divided ownership of the condo could be satisfied by splitting use of the condo up with John getting 7 weeks and Faye getting 45 weeks each year.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (14 of 32)
Example: Use the adjusted winner procedure to fairly divide the following items:
| Anne | Item | Becky |
| 50 | Laptop | 40 |
| 5 | DVDs | 5 |
| 6 | MP3 player | 5 |
| 31 | Tablet | 10 |
| 8 | TV | 40 |
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (15 of 32)
Solution: Initially Anne gets the laptop, the MP3 player, and the tablet for 87 points. Becky gets the TV for a point total of 40.
Anne is the leader so she gets the tied item, the DVDs, bringing her points to 92.
Anne’s items are compared against Becky’s:
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (16 of 32)
The DVDs and MP3 player are transferred. Anne’s score is now 81 and Becky’s is 50. The laptop is the critical item.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (17 of 32)
Becky gets the DVDs, the tapes, the MP3 player, the TV, and 34% of the laptop. Anne gets the tablet and the remaining 66% of the laptop.
This is as fair as it can be for a party of two, considering of course that neither party changes their mind about point allocation.
There are some problems with this procedure, for example, parties of three or more, critical items being non-dividable, etc.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (18 of 32)
The Knaster procedure is a method for dividing items among several parties, and it does not require dividing ownership of items. The procedure is based on having the parties assign monetary value to each item, their bid. The bidding is without knowledge of others’ bids.
Example: Assume there is one item to be divided, an automobile. The four people involved are each entitled to 1/4 of the car. Each person places a bid and each will end up with a value of at least 1/4 of his bid. Suppose the bids are as follows:
| Person | Abe | Ben | Caleb | Dan |
| Bid | $30,000 | $23,000 | $28,000 | $25,000 |
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (19 of 32)
Example:
| Person | Abe | Ben | Caleb | Dan |
| Bid | $30,000 | $23,000 | $28,000 | $25,000 |
The highest bidder, Abe, gets the car.
But that is the entire estate and he is only entitled to 1/4, which is $7500. So he puts the remaining 3/4, $22,500, into a kitty.
Each of the others withdraws 1/4 of what they assigned as the value of the car:
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (20 of 32)
Ben takes 1/4 of $23,000=$5750
Caleb withdraws 1/4 of $28,000=$7000
Dan takes 1/4 of $25,000=$6250
The total withdrawn is $5750+$7000+$6250=$19,000
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (21 of 32)
This leaves a kitty of $3500 to be divided equally among the four people. Each person receives ¼ of the value he assigned to the car plus .
The following table summarizes the transaction:
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (22 of 32)
| Person Bid | Abe $30,000 | Ben $23,000 | Caleb $28,000 | Dan $25,000 | Kitty |
| Car award | Car | 0 | |||
| To kitty | -22,500 | 22,500 | |||
| From kitty | 5750 | 7000 | 6250 | 3500 | |
| From kitty | 875 | 875 | 875 | 875 | 0 |
| Final share | Car-less $21,625 | $6625 | $7825 | $7125 | $0 |
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (23 of 32)
The Knaster Procedure, or method of sealed bids:
Step 1: Each person bids a secret dollar amount for each item to be divided (sealed bids).
Step 2: For each item separately:
The item is given to the person who bids the most.
The winner contributes the difference between their vote and his or her share of the item to a kitty.
Those who don’t win the item withdraw their share of each of their bids from the kitty.
Step 3: The money left in the kitty is divided equally among the bidders.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (24 of 32)
Example: A family farm is to be divided; parts of the farm are to be bid on by three family members.
| Bid | Julie | Anne | Steve |
| Farmhouse | $120,000 | $60,000 | $75,000 |
| Wheat fields | 210,000 | 450,000 | 390,000 |
| Cattle operations | 300,000 | 750,000 | 600,000 |
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (25 of 32)
Determine the fair division of the farm using the Knaster procedure.
Solution: Each part of the farm is given to the highest bidder.
Julie gets the farmhouse.
Anne gets the wheat field and the cattle operations.
Steve will not get any of the farm, but he will get cash.
The divisions of each piece of the farm and its cash value follow.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (26 of 32)
Farmhouse: Julie gets the farmhouse. She bid $120,000 but is entitled to 1/3 of that, $40,000, so she puts $80,000 in the kitty. Anne and Steve are entitled to 1/3 of their respective bids.
Anne takes 1/3 of $60,000=$20,000.
Steve takes 1/3 of $75,000=$25,000.
The total removed from the kitty is $20,000+$25,000=$45,000.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (27 of 32)
| Bid | Julie | Anne | Steve | Kitty |
| Farmhouse | $120,000 | $60,000 | $75,000 | |
| Farmhouse award | Farmhouse | $0 | ||
| To kitty | -$80,000 | $80,000 | ||
| From kitty | $20,000 | $25,000 | $35,000 |
This leaves in the kitty.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (28 of 32)
Wheat fields: Anne gets the wheat fields. She bids $450,000 but is entitled to 1/3 of that, $150,000, so she puts $300,000 in the kitty. Julie and Steve are entitled to 1/3 of their respective bids.
Julie takes 1/3 of $210,000=$70,000.
Steve takes 1/3 of $390,000=$130,000.
The total removed from the kitty is $70,000+$130,000=$200,000.
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (29 of 32)
| Bid | Julie | Anne | Steve | Kitty |
| Wheat fields | $210,000 | $450,000 | $390,000 | |
| Wheat fields award | Wheat fields | $35,000 | ||
| To kitty | -$300,000 | $335,000 | ||
| From kitty | $70,000 | $130,000 | $135,000 |
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (30 of 32)
Cattle ops.: Anne gets the cattle operations. She bids $750,000 but is entitled to 1/3 of that, $250,000, so she puts $500,000 in the kitty. Julie and Steve are entitled to 1/3 of their respective bids.
Julie takes 1/3 of $300,000=$100,000
Steve takes 1/3 of $600,000=$200,000
The total removed from the kitty is $100,000+$200,000=$300,000
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (31 of 32)
| Bid | Julie | Anne | Steve | Kitty |
| Cattle operations | $300,000 | $750,000 | $600,000 | |
| Cattle ops. award | Cattle ops. | $135,000 | ||
| To kitty | -$500,000 | $635,000 | ||
| From kitty | $100,000 | $200,000 | $335,000 | |
| Kitty distribution | $111,667 | $111,667 | $111,667 |
© 2018 W. H. Freeman and Company
8.3 Fair Division: What is a fair share? (32 of 32)
After the item distribution, there is $335,000 left in the kitty to be divided among the three.
335,000/3=$111,667 each
Julie gets the farmhouse and cash totaling:
−$80,000+$70,000+$100,000+$111,667=$201,667
Anne gets the wheat fields, the cattle operations, and cash totaling:
$20,000−$300,000 –$500,000+$111,667=−$668,333
That is, Anne needs to pay out that much.
Steve gets no property but cash totaling:
$25,000+$130,000+$200,000+$111,667=$446,667
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (1 of 39)
Apportionment: To find the size of the ideal district for a House representative you use the formula:
The ideal district size is also known as the Standard divisor. This number is used to determine each state’s quota, or its share of the House of Representatives:
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (2 of 39)
Example: For the U.S. population of 6,584,255 in 1810, with the House having members:
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (3 of 39)
Each state’s population is divided by the ideal district size to calculate its state’s quota.
Connecticut had a population of 261,818
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (4 of 39)
The rest of the states follow:
| State | Population | Quota | Rounded quota |
| Connecticut | 261,818 | 7.197 | 7 |
| Delaware | 71,004 | 1.952 | 2 |
| Georgia | 210,346 | 5.782 | 6 |
| Kentucky | 374,287 | 10.289 | 10 |
| Maryland | 335,946 | 9.235 | 9 |
| Massachusetts | 700,745 | 19.263 | 19 |
| New Hampshire | 214,460 | 5.895 | 6 |
| New Jersey | 241,222 | 6.631 | 7 |
| New York | 953,043 | 26.199 | 26 |
| North Carolina | 487,971 | 13.414 | 13 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (5 of 39)
| State | Population | Quota | Rounded quota |
| Ohio | 230,760 | 6.344 | 6 |
| Pennsylvania | 809,773 | 22.261 | 22 |
| Rhode Island | 76,888 | 2.114 | 2 |
| South Carolina | 336,569 | 9.252 | 9 |
| Tennessee | 243,913 | 6.705 | 7 |
| Vermonth | 217,895 | 5.99 | 6 |
| Virginia | 817,615 | 22.476 | 22 |
| Total | 6,584,255 | 181 | 179 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (6 of 39)
A problem arises over what to do with the fractional parts, so methods have been devised to deal with them either by ranking or rounding them somehow.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (7 of 39)
Hamilton’s solution: Ranking the fractions
Step 1: Calculate the quota for each state.
Step 2: Give to each state the number of representatives corresponding with the whole number part of the quota.
Step 3: Allocate the leftover House seats, representing the fractions left over, as follows: Rank the states by the size of the fractional part of their quota, from greatest to least, and give one leftover member to each state in that order until the leftovers are exhausted.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (8 of 39)
| State | Quota | Quota rounded down: Initial seats | Fractional part of Quota | Added seats | Final seats |
| Connecticut | 7.197 | 7 | 0.197 | 7 | |
| Delaware | 1.952 | 1 | 0.952 | 1 | 2 |
| Georgia | 5.782 | 5 | 0.782 | 1 | 6 |
| Kentucky | 10.289 | 10 | 0.289 | 10 | |
| Maryland | 9.235 | 9 | 0.235 | 9 | |
| Massachusetts | 19.263 | 19 | 0.263 | 19 | |
| New Hampshire | 5.895 | 5 | 0.895 | 1 | 6 |
| New Jersey | 6.631 | 6 | 0.631 | 1 | 7 |
| New York | 26.199 | 26 | 0.199 | 26 | |
| North Carolina | 13.414 | 13 | 0.414 | 1 | 14 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (9 of 39)
| State | Quota | Quota rounded down: Initial seats | Fractional part of quota | Added seats | Final seats |
| Ohio | 6.344 | 6 | 0.344 | 6 | |
| Pennsylvania | 22.261 | 22 | 0.261 | 22 | |
| Rhode Island | 2.114 | 2 | 0.114 | 2 | |
| South Carolina | 9.252 | 9 | 0.252 | 9 | |
| Tennessee | 6.705 | 6 | 0.705 | 1 | 7 |
| Vermonth | 5.99 | 5 | 0.99 | 1 | 6 |
| Virginia | 22.476 | 22 | 0.476 | 1 | 23 |
| Total | 181 | 173 | 8 | 181 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (10 of 39)
Example: A city has four districts―North, South, East, and West. There are 10 city councilors allocated according to population with at least one from each district. Given the populations below, calculate the quota for each district and use Hamilton’s solution to determine the number of councilors to represent each district.
| District | Population |
| North | 3900 |
| East | 4800 |
| West | 18,000 |
| South | 3300 |
| Total | 30,000 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (11 of 39)
Solution: The 10 seats to fill and the population of 30,000 are key.
| District | Quota | Initial seats | Quota remainder | Added seats | Final seats |
| North | 1.3 | 1 | 0.3 | 1 | |
| East | 1.6 | 1 | 0.6 | 1 | 2 |
| West | 6.0 | 6 | 0.0 | 6 | |
| South | 1.1 | 1 | 0.1 | 1 | |
| Total | 10 | 9 | 10 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (12 of 39)
Jefferson’s solution: Adjusting the divisor
Step 1: Start using the ideal district size as the divisor.
Step 2: Calculate the quota for each state by dividing the population by the divisor.
Step 3: Round down each quota to the nearest whole number, but not less than 1, and sum the rounded quotas.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (13 of 39)
Step 4: If the sum from step 3 is larger than the size of the House, increase the divisor and repeat steps 2 and 3. If the sum is too small, decrease the divisor and repeat steps 2 and 3.
This process is repeated until a divisor is found for which the sum of the rounded quotas is equal to the House size.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (14 of 39)
Example:
| State | Population | Quota using divisor 36,377.10 | First try: Quota rounded down | Second try: Decrease divisor to 34,000, round down | Third try: Increase divisor to 35,000, round down |
| Connecticut | 261,818 | 7.197 | 7 | 7 | 7 |
| Delaware | 71,004 | 1.952 | 1 | 2 | 2 |
| Georgia | 210,346 | 5.782 | 5 | 6 | 6 |
| Kentucky | 374,287 | 10.289 | 10 | 11 | 10 |
| Maryland | 335,946 | 9.235 | 9 | 9 | 9 |
| Massachusetts | 700,745 | 19.263 | 19 | 20 | 20 |
| New Hampshire | 214,460 | 5.895 | 5 | 6 | 6 |
| New Jersey | 241,222 | 6.631 | 6 | 7 | 6 |
| New York | 953,043 | 26.199 | 26 | 28 | 27 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (15 of 39)
| State | Population | Quota using divisor 36,377.10 | First try: Quota rounded down | Second try: Decrease divisor to 34,000, round down | Third try: Increase divisor to 35,000, round down |
| North Carolina | 487,971 | 13.414 | 13 | 14 | 13 |
| Ohio | 230,760 | 6.344 | 6 | 6 | 6 |
| Pennsylvania | 809,773 | 22.261 | 22 | 23 | 23 |
| Rhode Island | 76,888 | 2.114 | 2 | 2 | 2 |
| South Carolina | 336,569 | 9.252 | 9 | 9 | 9 |
| Tennessee | 243,913 | 6.705 | 6 | 7 | 6 |
| Vermonth | 217,895 | 5.99 | 5 | 6 | 6 |
| Virginia | 817,615 | 22.476 | 22 | 24 | 23 |
| Total | 6,584,255 | 181 | 179 | 187 | 181 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (16 of 39)
Example: Given the populations below, from the previous example, use Jefferson’s solution to apportion the 10 councilors to the different districts.
| District | Population |
| North | 3900 |
| East | 4800 |
| West | 18,000 |
| South | 3300 |
| Total | 30,000 |
Solution: Ten councilors and a population of 30,000 gives an ideal district size of 3000. Use this as the first divisor.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (17 of 39)
The following table represents Jefferson’s solution. At each attempt of apportionment, the quota is rounded down.
| District | Population | Quota using divisor 3000 | First try: Quota round down | Second try: Decrease divisor to 2,500, round down |
| North | 3900 | 1.3 | 1 | 1 |
| East | 4800 | 1.6 | 1 | 1 |
| West | 18,000 | 6.0 | 6 | 7 |
| South | 3300 | 1.1 | 1 | 1 |
| Total | 30,000 | 10 | 9 | 10 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (18 of 39)
A desirable trait of any apportionment method is to stay within quota, meaning that the final apportionment for each state would be within one of the quota.
So the final apportionment should be the quota rounded up one or down one. Apportionment methods that do not stay within the quota are said to violate quota.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (19 of 39)
The Alabama Paradox: a problem arising when calculating the quota for certain states.
For a House size of 299, Alabama would be allocated eight representatives.
For a House size of 300, Alabama receives only seven representatives.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (20 of 39)
Example: Explore the Alabama Paradox:
For a House size of 299, determine the ideal district size and verify the quotas for Alabama and Illinois.
How do the fractional parts of the quotas for the two states compare for 299 members.
How do the fractional parts of the quotas for the two states compare for 300 members?
Use parts 2 and 3 to find plausible explanations for the paradox.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (21 of 39)
Table 8.2 Alabama Paradox: 1880 Census, Hamilton’s Method
| State | Population | House size of 299: Quota | House size of 299: Initial seats | House size of 299: Added seats | House size of 299: Total seats | House size of 300: Quota | House size of 300: Initial seats | House size of 300: Added seats | House size of 300: Total seats |
| Alabama | 1,262,505 | 7.646 | 7 | 1 | 8 | 7.671 | 7 | 7 | |
| Arizona | 802,525 | 4.86 | 4 | 1 | 5 | 4.876 | 4 | 1 | 5 |
| California | 864,694 | 5.237 | 5 | 5 | 5.254 | 5 | 5 | ||
| Colorado | 194,327 | 1.177 | 1 | 1 | 1.181 | 1 | 1 | ||
| Connecticut | 622,700 | 3.771 | 3 | 1 | 4 | 3.784 | 3 | 4 | |
| Delaware | 146,608 | 0.888 | 1 | 1 | 0.891 | 1 | 1 | ||
| Florida | 269,493 | 1.632 | 1 | 1 | 1.638 | 1 | 1 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (22 of 39)
Table 8.2 Alabama Paradox: 1880 Census, Hamilton’s Method
| State | Population | House size of 299: Quota | House size of 299: Initial seats | House size of 299: Added seats | House size of 299: Total seats | House size of 300: Quota | House size of 300: Initial seats | House size of 300: Added seats | House size of 300: Total seats |
| Georgia | 1,542,180 | 9.34 | 9 | 9 | 9.371 | 9 | 9 | ||
| Illinois | 3,077,871 | 18.64 | 18 | 18 | 18.702 | 18 | 1 | 19 | |
| Indiana | 1,978,301 | 11.98 | 11 | 1 | 12 | 12.021 | 12 | 12 | |
| Iowa | 1,624,615 | 9.839 | 9 | 1 | 10 | 9.872 | 9 | 1 | 10 |
| Kansas | 996,096 | 6.033 | 6 | 6 | 6.053 | 6 | 6 | ||
| Kentucky | 1,648,690 | 9.985 | 9 | 1 | 10 | 10.018 | 10 | 10 | |
| Louisiana | 939,946 | 5.692 | 5 | 1 | 6 | 5.711 | 5 | 1 | 6 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (23 of 39)
Table 8.2 Alabama Paradox: 1880 Census, Hamilton’s Method
| State | Population | House size of 299: Quota | House size of 299: Initial seats | House size of 299: Added seats | House size of 299: Total seats | House size of 300: Quota | House size of 300: Initial seats | House size of 300: Added seats | House size of 300: Total seats |
| Maine | 648,936 | 3.93 | 3 | 1 | 4 | 3.943 | 3 | 1 | 4 |
| Maryland | 934,943 | 5.662 | 5 | 1 | 6 | 5.681 | 5 | 1 | 6 |
| Massachusetts | 1,783,085 | 10.8 | 10 | 1 | 11 | 10.835 | 10 | 1 | II |
| Michigan | 1,636,937 | 9.914 | 9 | 1 | 10 | 9.947 | 9 | 1 | 10 |
| Minnesota | 780,773 | 4.728 | 4 | 1 | 5 | 4.744 | 4 | 1 | 5 |
| Mississippi | 1,131,597 | 6.853 | 6 | 1 | 7 | 6.876 | 6 | 1 | 7 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (24 of 39)
Solution:
The table shows the total population is 49,371,340.
For a house size of 299, the ideal district size is total population divided by 299, yielding about 165,121.539.
To verify the quotas divide the state population by the ideal district size.
For Alabama, the quota is .
For Illinois, the quota is .
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (25 of 39)
With 299 members, Alabama’s fractional part is 0.646 and Illinois’ is 0.640. Alabama has the larger fractional part.
The table shows that with 300 members, the fractional part for Alabama is 0.671 and for Illinois is 0.702. Illinois has the larger fractional part.
From parts 2 and 3, it is apparent that increasing the number of seats changes the fractional parts of the quotas for the two states. Because Hamilton’s method allocates leftover seats according to higher fractional parts, it is reasonable that Alabama could lose a seat and Illinois could gain a seat with the change.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (26 of 39)
Other problems arising from Hamilton’s procedure are the:
Population paradox: Where the numbers in question are being affected by different state populations growing at different rates, and
New states paradox: Where problems arise because of new states joining the United States.
Methods were devised to address some of the problems that arose with the procedures, for example the three adjusted divisor methods for apportionment.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (27 of 39)
Three Adjusted Divisor Methods for Apportionment:
Calculate the ideal district size by dividing the total population by the size of the House.
Calculate the quota for each state by dividing its population by the divisor.
Round each quota to a whole number as follows:
Jefferson’s Method: Round down, but not less than 1.
Adam’s Method: Round up.
Webster’s Method: Round to the nearest whole number―up if the fractional part is 0.5 or greater and down otherwise, but not less than 1.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (28 of 39)
Then sum the resulting rounded quotas.
If the sum from step 3 is larger than the size of the House, increase the divisor and repeat steps 2 and 3. If the sum is too small, decrease the divisor and repeat steps 2 and 3.
Continue this process until a divisor is found for which the sum of the rounded quota is equal to the number of House members.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (29 of 39)
Example: According to the 2010 census, the population was:
| State | Population |
| Montana | 994,416 |
| Alaska | 721,523 |
| Florida | 18,900,773 |
| West Virginia | 1,859,815 |
| Total | 22,476,527 |
Congress allocated 32 seats in total to these states.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (30 of 39)
Calculate the ideal district size and each state’s quota. Keep three places beyond the decimal point.
Calculate the apportionment of the 32 seats according to Hamilton’s method.
Calculate the apportionment of the 32 seats according to Jefferson’s method.
Calculate the apportionment of the 32 seats according to Adam’s method.
Calculate the apportionment of the 32 seats according to Webster’s method.
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (31 of 39)
Solution:
.
The quota for each state is calculated by dividing the state’s population by this divisor.
| State | Population | Quota |
| Montana | 994,416 | 1.416 |
| Alaska | 721,523 | 1.027 |
| Florida | 18,900,773 | 26.909 |
| West Virginia | 1,859,815 | 2.648 |
| Total | 22,476,527 | 32 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (32 of 39)
For Hamilton’s method, round the quotas down. This fills 30 seats. The remaining two are given to the largest fractional parts of the states’ quotas, Florida and West Virginia.
| State | Population | Quota | Quota round down | Fractional part of quota | Added seats | Final seats |
| Montana | 994,416 | 1.416 | 1 | 0.416 | 0 | 1 |
| Alaska | 721,523 | 1.027 | 1 | 0.027 | 0 | 1 |
| Florida | 18,900,773 | 26.909 | 26 | 0.909 | 1 | 27 |
| W.Virginia | 1,859,815 | 2.648 | 2 | 0.648 | 1 | 3 |
| Total | 22,476,527 | 32 | 30 | 2 | 32 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (33 of 39)
For Jefferson’s method, the quota is again rounded down; the total is less than 32 so the divisor is adjusted to a smaller number of 700,000. On this attempt it is still too few, so a third divisor of 675,000 is used.
| State | Population | Quota using divisor 702,391.469 | First try: Quota rounded down | Second try: Decrease divisor to 700,000, round down | Another try: Decrease divisor to 675,000, round down |
| Montana | 994,416 | 1.416 | 1 | 1 | 1 |
| Alaska | 721,523 | 1.027 | 1 | 1 | 1 |
| Florida | 18,900,773 | 26.909 | 26 | 27 | 28 |
| West Virginia | 1,859,815 | 2.648 | 2 | 2 | 2 |
| Total | 22,476,527 | 32 | 30 | 31 | 32 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (34 of 39)
For Adam’s method, the initial allocation is made by rounding each quota up. The total is more than 32 so the divisor is adjusted higher to 750,000. The 32 desired seats are obtained.
| State | Population | Quota using divisor 702,391.469 | First try: Quota rounded down | Second try: Decrease divisor to 700,000, round down | Another try: Decrease divisor to 675,000, round down |
| Montana | 994,416 | 1.416 | 1 | 1 | 1 |
| Alaska | 721,523 | 1.027 | 1 | 1 | 1 |
| Florida | 18,900,773 | 26.909 | 26 | 27 | 28 |
| West Virginia | 1,859,815 | 2.648 | 2 | 2 | 2 |
| Total | 22,476,527 | 32 | 30 | 31 | 32 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (35 of 39)
For Webster’s Method, round the initial quotas to the nearest whole number, round down if the decimal is less than 0.5 and up otherwise. Because the total from the initial allocations is 32, no adjustments need to be made.
| State | Population | Quota using divisor 702,391.469 | First try: Quota rounded down |
| Montana | 994,416 | 1.416 | 1 |
| Alaska | 721,523 | 1.027 | 1 |
| Florida | 18,900,773 | 26.909 | 27 |
| West Virginia | 1,859,815 | 2.648 | 2 |
| Total | 22,476,527 | 32 | 32 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (36 of 39)
The Huntington-Hill method: is an adjusted divisor method and it is the one that is used today. It follows the steps of the other methods, but the divisor is found using a geometric mean: if n is the whole number part of the quotient, to find the geometric mean use the formula:
The quota for each state is rounded up if it is at least as large as the geometric mean or down otherwise.
A table of Geometric means follows:
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (37 of 39)
TABLE 8.3 Geometric Means
| Whole number | Geometric mean | Whole number | Geometric mean | Whole number | Geometric mean |
| 1 | 1.414 | 11 | 11.489 | 21 | 21.494 |
| 2 | 2.449 | 12 | 12.490 | 22 | 22.494 |
| 3 | 3.464 | 13 | 13.491 | 23 | 23.495 |
| 4 | 4.472 | 14 | 14.491 | 24 | 24.495 |
| 5 | 5.477 | 15 | 15.492 | 25 | 25.495 |
| 6 | 6.481 | 16 | 16.492 | 26 | 26.495 |
| 7 | 7.483 | 17 | 17.493 | 27 | 27.495 |
| 8 | 8.485 | 18 | 18.493 | 28 | 28.496 |
| 9 | 9.487 | 19 | 19.494 | 29 | 29.496 |
| 10 | 10.488 | 20 | 20.494 | 30 | 30.496 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (38 of 39)
Example: Use the Huntington-Hill method to allocate 32 seats to the four states from previous example.
Solution:
| State | Population | Quota divisor of 702,391.469 | Geometric mean | First try: Round Quota | Second try: Divisor 640,000 | Rounded quota |
| Montana | 994,416 | 1.416 | 1.414 | 2 | 1.401 | 1 |
| Alaska | 721,523 | 1.027 | 1.414 | 1 | 1.016 | 1 |
| Florida | 18,900,773 | 26.909 | 26.909 | 27 | 26.621 | 27 |
| W. Virginia | 1,859,815 | 2.648 | 2.449 | 3 | 2.619 | 3 |
| Total | 22,476,527 | 32 | 33 | 32 |
© 2018 W. H. Freeman and Company
8.4 Apportionment: Am I represented? (39 of 39)
© 2018 W. H. Freeman and Company
Chapter Summary (1 of 2)
Voting power: does your vote count?
coalitions, quotas, swing voters
Banzhaf power index
Stanley-Shubik power index
Voting systems: choosing a winner
plurality
top-two runoff
elimination runoff
Borda count
Condorcet winner
© 2018 W. H. Freeman and Company
Chapter Summary (2 of 2)
Fair division:
Methods:
divide and choose
adjusted winner procedure
Apportionment:
Hamilton’s method
Jefferson’s method
Adjusted divisor methods
Huntington-Hill method
© 2018 W. H. Freeman and Company