Count plurality votes without inventing a majority
The largest vote total can be well below half, and district wins do not add up like national vote shares.
By the end: Calculate plurality outcomes and distinguish local victories from aggregate proportionality.
The largest total wins
In a single-member plurality election, the candidate with the most votes wins the seat. A majority means more than half of the relevant votes. With totals A 40, B 35, C 25, A has a plurality but not a majority. The remaining sixty votes do not establish a unified preference for either B or C.
A ballot marking one candidate reveals that choice, not a complete ranking or intensity of support. Later experiments stipulate full rankings so that we can compare additional rules. Do not infer those rankings from the first-choice totals alone.
Find the leader without inventing later preferences
Plurality compares the candidate totals and awards the single available seat to the largest. With A forty, B thirty-five, and C twenty-five, compare A with each rival separately. Forty exceeds thirty-five by five and exceeds twenty-five by fifteen. That is enough for A to win under this rule. The majority calculation asks something else. One hundred votes have a halfway point of fifty, and more than half requires at least fifty-one whole votes here. A falls eleven votes short of that requirement, despite winning the seat.
The sixty votes not marked A cannot be added to one opponent without an extra assumption. Thirty-five are marked B and twenty-five are marked C. Their voters may disagree about which alternatives are acceptable. A ballot with one mark does not record what either group would choose in a two-candidate contest. It also does not show how strongly they support their marked candidate. 'Sixty did not choose A on this ballot' describes the reports. 'Sixty prefer B to A' invents information that the reports do not contain.
To see the missing information, construct two complete rankings with the same first choices. In one, forty voters rank A above C above B, thirty-five rank B above C above A, and twenty-five rank C above B above A. B defeats A head-to-head by sixty to forty. In another, keep the first two groups unchanged but let the twenty-five C voters rank C above A above B. A now defeats B by sixty-five to thirty-five. Plurality still elects A on the identical first-choice totals in both profiles.
Identical first choices, opposite A-versus-B results
Voters
Complete profile P
Complete profile Q
40
A > C > B
A > C > B
35
B > C > A
B > C > A
25
C > B > A
C > A > B
First choices A / B / C
40 / 35 / 25
40 / 35 / 25
A versus B
A 40; B 60
A 65; B 35
Plurality winner
A
A
P and Q are stipulated complete profiles. One-mark totals alone cannot select between them. Only the 25 C-first voters change their A/B comparison.
These complete profiles are added hypotheses, not recovered facts about a one-mark election. Their contrast proves that the aggregate forty, thirty-five, twenty-five totals cannot identify the A-versus-B result. The same distinction applies to claims about a candidate's withdrawal. Removing C requires assumptions about where C supporters go, whether they participate, and whether other voters respond. A table of first choices alone cannot settle any of those questions. Later ranked-ballot models provide their own explicit information rather than borrowing it from a plurality count.
An exact tie also needs an instruction that comparing totals cannot supply. With A forty, B forty, and C twenty, there are two equal leaders. If this invented contest uses alphabetical priority, A receives the seat. Another declared policy could produce another result without changing the totals. Do not apply a rounding rule to manufacture a unique leader. In the displayed plurality presets, A is already the unique leader, so no tie decision is used. The tie example identifies a specification requirement rather than a hidden extra step in those presets.
Practice by covering the winner column in the comparisons below. Find the unique maximum where one exists, then calculate whether it exceeds half of the stated total. Keep the two answers separate even when the same candidate satisfies both. Finally inspect the paired complete profiles and identify exactly which group's later preference changes the head-to-head outcome. That group contributes twenty-five to either side of the A-versus-B comparison while contributing the same twenty-five first choices to C. Its two roles explain why the measures can disagree.
The maximum and the halfway test are separate
Votes A / B / C
Plurality outcome
Majority of these votes?
40 / 35 / 25, total 100
A, unique maximum 40
No; 40 does not exceed 50
51 / 29 / 20, total 100
A, unique maximum 51
Yes; 51 exceeds 50
40 / 40 / 20, total 100
A under declared alphabetical priority
No; tie policy selects a winner, not a majority
All ballots here contribute one whole vote. The exact-tie row stipulates alphabetical priority for this invented contest. Neither displayed plurality preset requires a tie decision.
District victories discard margins
A party that wins a district by one vote receives the same single seat as a party that wins it by a thousand. National seat totals therefore depend on where support is distributed. A national vote lead does not guarantee a seat majority under separate district contests.
Imagine three equal-turnout districts with A shares 51%, 51%, and 0%. A wins two of three seats with 34% of the combined two-party vote. This arithmetic requires no fraud or unequal district populations. It follows from counting district winners instead of pooling votes for proportional seats.
Aggregate district seats separately from district votes
In a legislature elected through separate single-member plurality districts, count each local winner before adding party seats. With only A and B competing, a district containing fifty-one A votes and forty-nine B votes returns one A representative. A district containing ninety-nine A votes and one B vote also returns one A representative. The second margin is much larger, but it creates no second seat. These are whole-seat local decisions. Pooling the two districts' votes first and allocating proportional seats would answer a different institutional question.
The three equal-turnout districts in the original example contain one hundred votes each. A receives fifty-one, fifty-one, and zero; B receives forty-nine, forty-nine, and one hundred. A wins the first two districts and loses the third. Add A's votes to get one hundred two, then divide by three hundred for thirty-four percent. Separately add its district victories to get two, then divide by three for about 66.67 percent of seats. The large seat bonus follows without unequal voter counts or an error in either local tally.
Now redistribute A's same one hundred two votes across the three districts as thirty-four in each. Give B sixty-six in each, preserving B's one hundred ninety-eight votes. Every district still has one hundred voters. A now loses every seat despite keeping thirty-four percent of the pooled vote. This counterfactual changes the geographic assignment of support, not the national totals or plurality formula. It demonstrates why national shares cannot uniquely determine district seat totals. It does not describe voters moving in an observed election or a real redistricting proposal.
The same pooled votes can return two A seats or none
District
Distribution P, A / B
Distribution Q, A / B
1
51 / 49: A seat
34 / 66: B seat
2
51 / 49: A seat
34 / 66: B seat
3
0 / 100: B seat
34 / 66: B seat
Pooled votes A / B
102 / 198
102 / 198
A vote share
102/300 = 34%
102/300 = 34%
A seats and seat share
2/3 ≈ 66.67%
0/3 = 0%
Each district has 100 votes and returns one member under two-candidate plurality. P and Q change the distribution of support, not pooled votes or the formula.
Equal turnout guarantees that the simple average of district percentages equals the pooled share. With unequal turnout, use vote counts or turnout weights; equality is not generally guaranteed. Suppose A has fifty-one of one hundred votes in one district and nine of ten in another. Averaging fifty-one percent and ninety percent gives 70.5 percent. The correct pooled share is sixty of one hundred ten, about 54.55 percent. Weight each district by its vote count, or simply add counts before dividing. The two districts still elect one member each, so A's seat share is two of two in this example.
For an audit, write two separate rows of totals. The vote row adds votes, retaining each district's turnout weight. The seat row adds indicators equal to one for a local victory and zero otherwise. Large majorities and narrow victories both add one to that second row. This explains the arithmetic behind the phrase that district victories discard margins. It does not imply that votes beyond a winning margin are meaningless to voters, parties, campaign decisions, or other political activities. That broader meaning is not part of this seat formula.
Use the alternative distribution table to make an intermediate comparison before looking at its seat totals. Mark every district above fifty A votes, then count the marks. Compare the identical national numerators afterward. For the unequal-turnout case, compute both the unweighted average and the pooled fraction, and identify which operation gives each district equal influence. Equal influence per district is appropriate for counting its one seat here. Equal influence per district percentage is not appropriate for estimating the fraction of all votes cast for A.
Pooling votes is a turnout-weighted calculation
District or operation
A votes / total votes
A share
District 1
51 / 100
51%
District 2
9 / 10
90%
Pool the counts
(51 + 9) / (100 + 10)
60/110 ≈ 54.55%
Average district percentages equally
(51% + 90%) / 2
70.5%, not the pooled vote share
Count district victories
2 / 2 seats
100% of seats
The smaller district gets the same weight as the larger one in the unweighted average, despite casting only one tenth as many votes. Each elects one representative in this stipulated example.
Work through the mechanism
The largest first-choice total
A wins the single seat with 40 of 100 votes. No candidate has more than 50 votes.
Changing the preset replaces the selected vote totals and leaves the reference visible. Bars use the same 0–100-vote scale. These first-choice totals alone do not reveal later preferences.
District
A votes out of 100
Seat winner
1
51
A
2
51
A
3
0
B
Combined
102 out of 300 = 34%
A wins 2 of 3 seats
Equal turnout is assumed. The example separates national votes from district victories.
Read the presets and trace a fixed electorate
Start the plurality view with the compromise preset. Read the selected first-choice bars against the fixed reference. Both show forty A votes, thirty-five B votes, and twenty-five C votes on the same zero-to-one-hundred scale. The bars do not represent seats or rankings. One seat goes to A because its bar is longest, not because it crosses fifty. There is no transfer stage to play in this view. The preset buttons immediately replace the selected profile while leaving the reference visible for comparison.
Before choosing the cycle preset, predict which information will change. Its first choices are thirty-five A, thirty-three B, and thirty-two C. A remains the leader, now two votes ahead of B rather than five. Its vote share falls from forty percent to thirty-five percent, but its seat award remains one of one. Check those predictions against the displayed bars and result. This is a comparison between two stipulated profiles. Pressing a preset does not move five named people or simulate a campaign that persuaded them to switch.
Return to the compromise preset and record three quantities for each scenario on paper. Use the leading count, the gap between the largest and second-largest counts, and the winner's seat share. You should obtain forty and thirty-five for the leading counts, five and two for the gaps, and one hundred percent for both single-seat shares. The unchanged seat share is not evidence that the electorate is unchanged. It shows how a discrete one-seat output can remain constant when the underlying vote distribution changes within the same winning region.
What the two plurality presets actually change
Quantity
Compromise preset
Cycle preset
First choices A / B / C
40 / 35 / 25
35 / 33 / 32
All first-choice votes
100
100
Leader minus runner-up
40 − 35 = 5 votes
35 − 33 = 2 votes
A vote share
40%
35%
A seat award
1 of 1
1 of 1
A majority of first choices?
No
No
The interactive view replaces the selected profile immediately and keeps the 40/35/25 reference visible. This comparison traces no movement of identified voters and has no transfer or playback stage.
The fixed twenty-five-voter map below extends that reasoning to a delegation. Each labeled cell represents one voter. There are twelve A voters and thirteen B voters in both panels. Rows and columns each partition the identical grid into five connected districts containing five voters apiece. In row order, A's counts are three, three, two, two, two. In column order, they are three, three, three, three, zero. A therefore wins two row seats but four column seats, without any voter changing the marked choice.
Count the third row and third column yourself before accepting the delegation totals. The third row contains two A voters and three B voters, so B wins it. The third column contains three A voters and two B voters, so A wins it. Then inspect the fifth column, which contains all B voters. Concentrating five B votes there leaves enough A majorities elsewhere for four A victories. These are exact cell counts, not area measurements. The drawing's equal square sizes do not represent measured population density or geographic distance.
This grid comparison changes boundaries, unlike the plurality presets, which replace profiles. Keep those interventions distinct when explaining their outcomes. The grid has equal numbers of participating voters in every district; it contains no separate resident counts or legal boundary criteria. It establishes sensitivity to grouping, not malapportionment or partisan intent. To investigate a real map, identify its population basis and permissible alternatives. The result here is a controlled arithmetic example showing that a fixed national vote total and a fixed counting rule can still permit different legislative compositions.
Count actual voters inside each district boundary
Each labeled square is one voter, not a unit of area or resident population. The same fixed 12 A and 13 B voters form five connected equal-voter districts in each panel. Only the partition changes.
Explain incentives without assuming behavior
Plurality can give voters and parties reasons to coordinate around viable contenders, because splitting support among similar candidates can change which candidate leads. But the strength of this incentive depends on expectations, geography, party organization, and the office at stake. A counting example does not prove an inevitable two-party system.
The counterfactual is also delicate. If C withdraws, C’s supporters may switch, abstain, or respond to changed campaigns. To claim a spoiler effect in a particular election, you need evidence about those alternatives. A changed winner in a stipulated preference profile demonstrates possibility, not that every third candidate caused the same effect.
Test a coordination claim with an explicit counterfactual
A possible coordination incentive appears when several candidates divide support that could otherwise combine behind one. To make that claim precise, specify the alternative action and resulting ballots. Use the complete compromise profile, not just its first-choice summary. If C withdraws and every voter chooses the highest-ranked remaining candidate without abstaining or revising preferences, C's twenty-five supporters choose B. A stays at forty, B rises from thirty-five to sixty, and B wins. Under those assumptions, including C changes the winner of this stipulated plurality contest.
Now change only C supporters' later preference to C above A above B. Withdrawal under the same reporting assumption sends twenty-five to A, producing A sixty-five and B thirty-five. A wins with or without C, so the same first-choice totals no longer show a changed winner. A third hypothetical lets all twenty-five C supporters abstain. A then wins forty to thirty-five among seventy-five votes cast. Its share is about 53.33 percent of those votes, but it still has forty votes and forty percent of the original participants.
A withdrawal can alter more than the departing candidate's supporters. Other voters can change turnout or choose differently after endorsements or campaigns. The fixed-ranking assumption excludes those changes. To estimate the actual counterfactual, seek evidence about substitution and participation, and explain what it cannot identify. A survey question about a second choice may help, but it is not an observation of an election that never happened. The example establishes a mechanism and its dependence on assumptions, not a verdict about any real candidate.
Specify where C supporters go before claiming a changed winner
Counterfactual after C withdraws
A / B votes
Winner and denominator
25 C>B>A voters choose their next remaining candidate
40 / 60
B, 60/100 votes
25 C>A>B voters choose their next remaining candidate
65 / 35
A, 65/100 votes
All 25 C supporters abstain
40 / 35
A, 40/75 ≈ 53.33% of votes cast
Original contest, C remains
40 / 35 / 25
A, 40/100 votes
The 40 A-first and 35 B-first voters keep their reports in every row. The abstention row is a withdrawal scenario, not a ranked count with exhausted ballots. The identical fraction 40/75 does not make those mechanisms identical.
Coordination before voting can involve candidate nomination or agreements between parties. Coordination by voters can involve choosing a less-preferred candidate they believe has a better chance of defeating another. Such a choice depends on expectations about viability and on which outcome the voter prefers. The plurality formula supplies an incentive structure, not those expectations. A locally viable smaller party can face different incentives from a nationally similar party with dispersed support. National party counts therefore cannot be read straight from the instruction to select the largest local total.
Distinguish an opportunity to change the winner from an opportunity to obtain a preferred outcome. In the compromise profile, a C-first voter ranks B above A, so the stipulated withdrawal outcome favors that voter's ranking over the original A victory. In the alternative profile, that same first-choice label accompanies a preference for A over B. The label C-first alone cannot establish which alternative outcome would be an improvement. Nor does a possible improvement show that any individual voter could cause it alone or reliably identify it in advance.
For a final guided check, choose one row of the withdrawal comparison and state every assumption needed to reproduce it. Name who stays in the electorate, which later preference is supplied, and whether other reports remain fixed. Then state the mechanical conclusion as a conditional sentence rather than a prediction. Evaluate any recommendation separately against a declared objective, such as local candidate choice or party proportionality. The selected handbook discussion supports attention to coordination and context. Its historical cases do not establish an inevitable party system or current rules for any jurisdiction.
A coordination argument needs more than a first-choice label
Question
What is specified here
What would require further evidence
Can withdrawal change the winner?
Yes in profile P under next-choice voting
Actual substitution and participation after withdrawal
Would C supporters prefer B winning to A?
Yes in C>B>A; no in C>A>B
Their later preferences in a real one-mark election
Can one voter cause the change?
Not established by a 25-person group switch
Pivotality and others’ ballots
Will plurality produce a particular party system?
Not determined by the largest-total formula
Geographic support, entry, organization, and expectations
A mechanism can be possible without being inevitable, frequent, or individually actionable. These invented profiles do not estimate behavioral responses.
Exercises
A plurality is not half
With 40 votes for A, 35 for B, and 25 for C, who wins under plurality? What can you say about majority support?
Hint
Use the rule, then examine what information the ballots contain.
A wins because forty is the largest total. A has 40% of votes, not a majority. The one-choice totals do not say whether A would beat B or C head-to-head, or what C voters would do if C withdrew.
National vote share versus seats
Calculate A’s vote share and seat share across three equal-turnout districts with A shares 51%, 51%, and 0%.
Hint
Average the percentages only because turnout is equal.
Worked answer
Measure
Calculation
Result
Vote share
102 / 300
34%
Seat share
2 / 3
66.7%
A large seat bonus can arise from the aggregation rule.
A receives 102 of 300 votes, or 34%. It wins two districts, giving 2/3 of seats, about 66.7%. With unequal turnout, use vote counts rather than a simple average of district percentages. Neither calculation by itself establishes why the support is geographically distributed that way.
Sources and further reading
The explanations, counts, and worked answers are complete without these optional links. Reference annotations identify the selected material reviewed. Historical country examples are not statements of current law, and the bounded models are not certified election-counting software. Mechanical results, behavioral evidence, and judgments about desirable representation remain separate.
Selected explanations of system families, district magnitude, thresholds, proportional formulas, mixed systems, and design objectives. Country examples and legal thresholds are historical, not statements of current law.