In any system with geographic districts, how the lines are drawn affects results. The techniques for manipulating this and for detecting the manipulation have both developed considerably.
The requirement
Districts generally must contain roughly equal populations, which requires redrawing as populations shift.
Which is why redistricting follows census cycles in most systems.
Additional criteria commonly include contiguity, compactness, respect for existing political boundaries and, in some jurisdictions, protection of minority representation.
The two techniques
Packing concentrates opposing voters into few districts, so they win those overwhelmingly and lose everywhere else.
Cracking splits opposing voters across many districts so they form a majority in none.
Which together allow a party to win a majority of seats with a minority of votes, and the arithmetic is straightforward.
Who draws them
Varies enormously.
Legislatures in some jurisdictions, which produces an obvious conflict of interest since they are drawing their own districts.
Independent commissions in others, with varying degrees of independence and different appointment mechanisms.
Courts, where other processes fail or are struck down.
Comparative research generally finds independent commissions produce less partisan bias, with the design of the commission mattering considerably.
Measuring bias
Where the field has advanced most.
Compactness measures assess whether districts have reasonable shapes, which is intuitive and easy to defeat.
Partisan symmetry measures ask whether the seat outcome would be the same if vote shares were reversed.
The efficiency gap counts wasted votes on each side, defined as votes for losing candidates plus votes beyond the winning threshold.
Simulation approaches generate many maps meeting the legal criteria and compare the enacted map against that distribution, which is now widely used in litigation.
The litigation position
Racial gerrymandering claims and partisan gerrymandering claims are treated differently in many legal systems.
Which means maps have been struck down on racial grounds where partisan claims were held to be non-justiciable.
State-level courts in some systems have found partisan gerrymandering to violate state constitutional provisions where federal courts declined to intervene.
Incumbency
Maps drawn to protect incumbents of both parties are common and produce a different problem — reduced competition without partisan skew.
Which reduces accountability, since safe seats mean the meaningful contest is the party selection rather than the general election.
The proportion of competitive seats has declined in several systems, driven by both boundaries and by geographic sorting of voters.
Geographic sorting
Voters have become more geographically clustered by political preference in several countries.
Which produces some of the same effects as deliberate manipulation without anyone drawing lines badly.
Simulation methods help distinguish the two, by establishing what a neutral map would produce given the actual distribution of voters.
Reform
Ballot initiatives, constitutional amendments and legislation establishing commissions have all been used.
Which face the obstacle that reform must generally be enacted by the people who benefit from the current arrangement.
Where reform has occurred it has frequently come through direct democracy mechanisms rather than through legislatures.
Multi-member districts
Where several representatives are elected from one district, which changes the calculation entirely.
Manipulation is harder, since a district electing several members will generally return a mix.
Which is why proportional systems using multi-member districts are less vulnerable to boundary manipulation.
The trade-off is a weaker link between a representative and a specific local area.
Census and enumeration
Boundaries depend on population data, which makes census accuracy consequential.
Undercounting of specific populations has been documented in several countries, with effects on both representation and funding allocation.
Which is why census methodology and questions have been politically contested, since the count determines the districts.
Prison populations
Where incarcerated people are counted affects district populations.
Counting them at the prison rather than their home address shifts population from one area to another, which has been challenged and changed in several jurisdictions.
Timing
Mid-decade redistricting, redrawing outside the normal cycle, has occurred and been litigated.
Which is permitted in some systems and prohibited in others, and it allows boundaries to be redrawn when political control changes.
Ballot design and administration
Separate from boundaries and equally consequential.
Ballot order effects, where earlier-listed candidates receive more votes, are documented and have been addressed by rotation in some jurisdictions.
Polling place allocation, hours and staffing affect waiting times, which affect turnout, and inequalities in these have been measured.
Independent administration
Who runs elections varies from partisan officials to independent commissions.
Which affects confidence, and comparative research generally favours independent administration on perceived legitimacy.
Where partisan officials administer elections in which they are candidates, the conflict is direct and has attracted attention.
Voting method
Postal, early and in-person voting have different administrative requirements and different accessibility.
Which affects who finds it easier to vote, and rules on these have been among the most contested electoral questions in several countries.
Evidence on turnout effects of specific rules is mixed and generally finds convenience measures produce modest increases.