Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Arrow’sImpossibilityTheorem
1) Nospecialtreatmentofparticularvotersorcandidates
2) Transitivity• A>BandB>CimpliesA>C• Nocycles
3) Monotonicity• Avoterchangingtheirballotinawayfavoringcannotcausethatcandidate’soverallrankingtogodown.
4) Independenceofirrelevantalternatives• Overallrelativerankingoftwocandidatesdependsononlytheirrelativerankingonvoterballots
A
BC
Whyindependenceofirrelevantalternativesmatters:1995FigureSkatingWorldChampionship
• RankingspriortoMichelleKwanskating:
• 1st place:ChenLu(China)
• 2nd place:NicoleBobek (USA)
• 3rd place:SuryaBonaly (France)
• RankingsafterjudgingofMichelleKwan:
• 1st place:ChenLu(China)
• 2nd place:SuryaBonaly (France)
• 3rd place:NicoleBobek (USA)
• 4th place:MichelleKwan(USA)
Plurality:whoevergetsthemostvoteswinsStrengths• Simpleballottofillout• Transparentresults• Easytounderstand• Monotonic
Weaknesses• Votesplitting• Spoilers• Tacticalvoting• Negativecampaigning
• 1860USPresidentialElection• AbrahamLincoln• StephenDouglas• JohnBreckinridge• JohnBell
Borda count
PointsystemforfieldofNcandidates,e.g.,• N-1pointsfor1st place• N-2pointsfor2nd place• …• 0pointsforlastplace
(orotherpointscheme,forinstance,weighting1stplacemoreheavily)
Strengths• Takesintoaccountfullsetofpreferences• Canpromotecompromisecandidates• Monotonic
Weaknesses• Vulnerabletostrategicvoting,suchasburyingfavorite’smainrivals
Borda count:1999baseballMVPelections
28voters14pointsfor1st place9pointsfor2nd place8pointsfor3rd place7pointsfor4th place…
http://www.baseball-reference.com/awards/awards_1999.shtml
Approvalvoting
• Voteforallcandidatesyoufindacceptable• Mayreducevotesplittingandsupportthirdparties• Notasexpressiveasrankedmethods
Saari’s example:• 9,999votersstronglysupportA,findBmarginallyacceptable,andstronglyopposeC• 1voterstronglysupportsC,findsBmarginallyacceptable,andstronglyopposesA
Pairwisecomparison/Condorcetmethod
• Winnerbasedonhead-to-headmatchesofallpossiblepairingsofcandidates
• Beatpath/CSSDtakesintoaccountmarginsofvictoryusingaweighteddirectedgraphcalculation
• Condorcetwinner:candidatewhowinsallhead-to-headmatches
• Condorcetwinnercriterion:whenaCondorcetwinnerexists,thatcandidateshouldwintheelection.
Instantrunoffvoting(IRV)/rankedchoice
• Eliminatecandidatewithleast1st placevotes• Moveupcandidatesandrepeatuntilsinglewinnerleft
• Burlington,VT2009mayoralraceusedIRV• IRVwinnerwasKiss,followedbyWrightthenMontroll• Montroll wasCondorcetwinner• IfKisshadwonmore1st placevotes,hewouldhavelost
IRVisnotmonotonicIRVdoesn’tsatisfyCondorcetwinnercriterion
Gibbard-SatterthwaiteTheorem
• Tacticalvoting:dishonestvotingtoimproverankingofyourpreferredcandidate.• Allrankedvotingsystemswithnospecialtreatmentofparticularvotersorcandidatesaresusceptibletotacticalvoting.
Gerrymandering
• Incumbent(sweetheart)• RuledOKbycourt
• Racial• VotingRightsActof1965
• Partisan• Noclearmeasure
• Packingandcracking
http://www.redistrictingthenation.com
BakervsCarr,1962SupremeCourtcase• “Oneperson,onevote”• Eachindividualisweightedequallyinapportionment(doesn’tmatterwhetherlegallyabletovoteornot)
• Establishedrightoffederalcourtstoreviewredistrictingmaps(redrawnevery10yearsaftercensus)• FoundTennesseedistrictmapunconstitutional
• Districtsdidnotreflectmovementofpopulationtocities• 2/3ofrepresentativeselectedby1/3ofthestatepopulation
github.com/JeffreyBLewis/congressional-district-boundaries
CoopervsHarris:NorthCarolinadistrictmap• SupremeCourtruled5-3earlierthisweekthatDistricts1and12exhibitunconstitutionalracialgerrymandering• District12electedAfrican-American-favoredcandidateswith64-72%ofvote• NewmapincreasedpackingofAfrican-Americanvoters
2003-13map:7Demto6Repseatsin2011 2013-16map:10Repto3Demseatsin2015
QuantifyingpartisangerrymanderingEfficiencygap
• StephanopoulosandMcGhee• Assesses“wastedvotes”in2-partyelection• Ifapartylosestheelection,allofthatparty’svotesarewasted.
• Ifapartywinstheelection,thevotespast50%arewasted.
• Sumwastedvotesforeachpartyacrossthedistrictsinthatstate• Finddifferenceintotalwastedvotesbetweenthe2parties,dividedbytotal#ofvotes
6Red:4Blue1vs4wasted
1 Red:9Blue1vs4wasted
Efficiencygapof30%
QuantifyingpartisangerrymanderingEfficiencygapofzerodoesn’timplyproportionalrepresentationDistrict Red Blue Winner Wastedvotes
1 6 4 Red 1vs4
2 6 4 Red 1vs4
3 6 4 Red 1vs4
4 4 6 Blue 4vs1
5 3 7 Blue 3vs2
6 3 7 Blue 3vs2
7 3 7 Blue 3vs2
8 3 7 Blue 3vs2
9 3 7 Blue 3vs2
10 3 7 Blue 3vs2
Total 40 60 25vs25
• 100votersin10districts• 40totalRedvoters• 60totalBluevoters
• Redwins3districts• Bluewins7districts
• Efficiencygap=0• BiasedtowardBlue
FelonydisenfranchisementintheUS• Dependsonstatelaws• OverallinUS,7.7%ofblackadultsdisenfranchised,comparedto1.8%ofnon-blackadults.• Largeprisonpopulationsalsousedasformofgerrymandering(countaspopulationbutcan’tvote)• Stateswithmostseverelaws:
• Florida(21%ofAfrican-Americansdisenfranchised)
• Kentucky(26%)• Virginia(22%)• Upto40%ofblackmendisenfranchisedinthesestates http://politicalmaps.org/6-million-lost-voters-state-
level-estimates-of-felony-disenfranchisement-2016/