Last lesson separated the question of who approves a district map from the rules that constrain it. This lesson turns to a different source of partisan advantage: where voters live. You already know that packing and cracking can be deliberate strategies. Now you will see why a similar concentration of votes can arise when districts are drawn without using party information.
Allow about 40 minutes, including the short video, reading, and pauses to examine the map.
What the supplied map shows

Start with the precinct-level panel at upper left. Its fine-grained patches are not evenly sprinkled across the state: some similarly shaded places lie next to one another. Compare it with the other three panels, where broader light and dark areas are easier to see. That spatial relationship—whether voters with similar preferences tend to live near one another—is the feature that matters here.
The legend describes partisan probabilities, not numbers of voters. A large dark area need not contain more voters than a small light area. Nor do these panels provide district vote totals, seat counts, or evidence of why any boundary was chosen. In particular, the image alone does not tell you which light areas are cities. Treat it as a picture of geographic pattern, not a verdict about a mapmaker.
Gerrymandering: Crash Course Government and Politics #37
CrashCourse’s “Gerrymandering: Crash Course Government and Politics #37” briefly introduces residential clustering as an alternative explanation for some lopsided districts.
Watch the geography explanation, immediately after the district-drawing examples. Focus on the proposed mechanism—supporters living close together—rather than treating its state example as proof that a particular enacted map was neutral.
How a cluster becomes a seat disadvantage
A district awards one seat to the candidate with the most votes, regardless of the size of the winning margin. This makes the location of votes consequential as well as their statewide total.
Suppose one party’s strongest supporters live in a large, dense cluster. A mapmaker trying to create contiguous, reasonably compact, equal-population districts may form several districts largely within that cluster. The party wins those districts by wide margins, but votes beyond what was needed to win do not earn additional seats. Outside the cluster, its supporters may be present in many districts without forming a majority in any of them. Its opponent can then win more districts by smaller margins.
This resembles the result of packing, but not necessarily the act of deliberate packing. In the earlier packing lesson, the boundary was chosen to concentrate opposing voters. Here, a concentration already exists in the electorate; even a mapmaker who does not consult party data might produce heavily one-sided districts while working with that geography.
Chen and Rodden develop this distinction and explain why proximity matters when neighboring small areas are assembled into districts.
[PDF] Unintentional Gerrymandering: Political Geography and Electoral ...
Read the opening argument and the later district-building illustration in Chen and Rodden’s “Unintentional Gerrymandering.” Together, they distinguish intentional line-drawing from an advantage produced by the locations of voters.
In the introduction (approximately pp. 240–242), begin at “Why does this type of electoral bias emerge?” Read the opening argument through the discussion of neighboring precincts. Focus on the difference between a party’s statewide support and the proximity of its supporters. Then find Section 1, “Political Geography and the Roots of Electoral Bias in the United States,” around p. 247. Begin with the paragraph starting “The process of building electoral districts involves someone.” Read the district illustration through its concluding explanation of how compact, contiguous districts can yield fewer seats than a party’s vote share might suggest. Concentrate on the three geographic situations described, rather than the article’s later statistical methods.
The reading describes two ways votes can fail to produce many seats: supporters in a large cluster may win a few districts overwhelmingly, while smaller, separated pockets of supporters may be included in districts won by the other party. Neither effect requires a mapmaker to seek those outcomes deliberately. Both still depend on the particular geography, district size, and rules used to draw the lines.
Work through the vote-to-seat arithmetic
Consider a hypothetical five-district plan. Each district has 100 voters, and each is assumed to meet the applicable population and geographic rules. Two districts cover neighboring parts of a concentrated party-D area; three cover surrounding areas with more mixed support. The table is an illustration of what such a plan could produce, not a calculation from the North Carolina image.
| Area represented by district | Party D votes | Party R votes | Winning party |
|---|---|---|---|
| Cluster district 1 | 90 | 10 | D |
| Cluster district 2 | 90 | 10 | D |
| Surrounding district 1 | 40 | 60 | R |
| Surrounding district 2 | 40 | 60 | R |
| Surrounding district 3 | 40 | 60 | R |
Party D receives of the votes——but wins only two of five seats, or . Party R receives of the votes and wins three seats. The electorate’s preferences have not changed between counting votes and awarding seats; the difference comes from applying five separate plurality contests to voters distributed unevenly across space.
Notice the contrast in margins. Party D wins each cluster district with 90 votes when 51 would suffice against 10; it also receives 40 votes in each district it loses. Party R wins its three districts with 60 votes apiece. A statewide vote total hides this distribution of comfortable wins and narrower losses.
The example does not establish that geography forces this exact result. Another permissible set of boundaries might combine voters differently. It establishes the narrower, important point: a seat advantage is possible without selecting lines for partisan gain when residential patterns and ordinary districting constraints put one party’s supporters together more strongly than the other’s.
Separate a mechanism from evidence of intent
Return to the North Carolina image. Its adjacent light and dark areas help you visualize why nearby voters are not interchangeable with distant voters: a district must be built from places, not from an arbitrary statewide list of votes. The image cannot tell you how many equal-population districts fit within a light patch, how voters actually cast ballots in each proposed district, or whether the person approving a map sought a partisan advantage.
This distinction also prevents an overly broad claim about cities. Chen and Rodden study settings in which Democratic support is strongly clustered in urban areas, while Republican support is more dispersed. That is a finding about particular electoral geographies, not a rule that every state favors the same party. Different residential patterns, district sizes, turnout, and boundaries can yield different outcomes. A party concentrated in a few places may even benefit under some configurations.
There are therefore two separate judgments:
- Outcome: Did one party win more seats than its statewide vote share might lead you to expect?
- Explanation: How much of that result reflects where voters live, the applicable map-drawing rules, and choices made among the available maps?
A lopsided outcome answers the first question, not the second. Equally, the fact that geography can produce a lopsided outcome does not establish that an actual map was drawn without partisan intent.
Key takeaways
Residential clustering can give a party many votes in districts it already wins while leaving it short of a majority elsewhere. Under single-member elections, that can produce fewer seats than its statewide vote share suggests, even without deliberate partisan manipulation. Use the supplied map to notice spatial concentration, but do not read population totals, seat results, or intent directly from its shading.
The next lesson will compare an enacted map with alternative maps drawn under stated constraints. That comparison can help assess whether its outcome is unusual relative to the available choices—while still leaving intent as a separate question.