Welcome to the Congressional Representation Project!

What happened

After the 1990 Census, North Carolina’s population growth earned the state a twelfth U.S. House seat. Under the Voting Rights Act rules then in force, and after the Justice Department pressed the state to create a second district where Black voters could elect a candidate of their choice, the General Assembly drew a new map that included the now-famous 12th district: a long, narrow corridor that followed Interstate 85 and linked Black communities from the Charlotte area through Greensboro and Durham.

The first election under that map was 1992. Mel Watt won the open 12th, and Eva Clayton won the 1st. Together they became the first Black members of Congress from North Carolina since Reconstruction. Almost immediately, the 12th’s unusual shape became a national flashpoint. White voters sued, arguing that race had driven the line-drawing. In Shaw v. Reno (1993), the Supreme Court held that race could not be the predominant factor in drawing districts, even when the stated goal was minority representation. The ruling did not erase the 1992 results, but it opened years of litigation and forced repeated map revisions through the mid-1990s and beyond.

In short: a Census-driven seat gain, VRA-era mapmaking pressure, a highly visible district geometry, a historic election outcome, and a Supreme Court doctrine that reshaped how race could be used in redistricting. The metrics below capture the statewide structural shift that arrived with that first election under the new plan.

What the metrics show

Because the controversy was about how seats and votes lined up under a new map, not about missing opponents, the key scores are fairness and the efficiency gap. Contestation stayed perfect, so raw vote totals are trustworthy here.

• Raw fairness fell from 52.2 (1990) to 38.3 (1992). That 14-point drop lands in the first election under the new twelve-district plan. Fairness measures how closely seat outcomes track statewide vote share; the timing matches the map change, not a sudden collapse in ballot competition.
• Efficiency gap score fell from 62.7 to 19.5. A lower score means wasted votes became more one-sided statewide. That fits a plan that concentrates some voters into safe seats (including the packed 12th) while rearranging the rest of the map.
• Contestation stayed at 100% and competitiveness stayed high (93.1 to 92.3). Every district still had two-party vote totals. The fairness slide is therefore about geography and seat allocation, not about uncontested races or missing data.
• State EAI eased from 82.5 to 78.0. Accountability dipped mainly because fairness dropped; competition and contestation were still strong, which is why EAI did not crash the way it does in Massachusetts-style uncontested cases.
• In the 12th itself (1992), Watt won 72.0% in a contested race (176,664 votes; district competitiveness 55.9). The district was a clear Democratic hold with real opposition on the ballot. The legal fight was over shape and racial predominance, not over whether voters got a choice that year.

What happened

After the 2000 Census, Texas could not agree on a congressional map in the legislature, so a federal court drew the lines used in 2002. That election produced a nearly even House delegation: 17 Democrats and 15 Republicans. The same year, Republicans also won control of both chambers of the Texas Legislature for the first time in more than a century.

In 2003, U.S. House Majority Leader Tom DeLay and allied state leaders pushed an unusual second redraw mid-decade, before the next Census. Democratic legislators briefly fled to Oklahoma to break quorum and stall the bill, but the plan eventually passed. The new map was in place for the 2004 election. Lawsuits followed, culminating in League of United Latin American Citizens v. Perry (2006), where the Supreme Court allowed most mid-decade redistricting but struck down one district on Voting Rights Act grounds and required further fixes.

What makes Texas distinctive is the double redraw: a court map in 2002, then a partisan legislative map in 2004, in a presidential year when George W. Bush also carried the state by a wide margin. Seat outcomes moved much faster than the statewide House vote, which is exactly what the fairness and partisan-advantage scores pick up.

What the metrics show

The IRL story is a rare mid-decade map swap that flipped seat control. Compare 2002 (court map) to 2004 (DeLay map) on fairness, seat-versus-vote balance, and partisan advantage. EAI is useful mainly as a contrast: races stayed contested even as fairness fell.

• Delegation: 17D/15R (2002) to 11D/21R (2004). Republicans gained six seats under the new lines. That matches the political purpose of the mid-decade redraw and is the clearest on-the-ground outcome of the map change.
• Dem House vote share: 45.1% to 40.3%; Dem seat share: 53% to 34%. Votes moved only about five points, but seats moved far more. That gap is why fairness falls: the map translated a modest vote shift into a large seat shift.
• Raw fairness: 56.9 to 45.9; adjusted fairness: 79.4 to 40.9. Both pipelines show a fairness decline after the new map, and the adjusted drop is even steeper. The timing lines up with the DeLay plan taking effect, not with a routine Census cycle.
• Partisan advantage index: −0.44 to +0.49. The signed score flips from a Democratic structural lean to a Republican one between the same two elections. That is the metric version of “who benefited from the new lines.”
• Raw EAI: 63.8 to 65.3, with contestation 71.9% to 78.1% and competitiveness 62.6 to 70.0. Accountability did not collapse because voters still saw more two-party races and closer contests. Fairness and EAI diverge here: Texas became less proportional even while ballot competition improved.

What happened

The 2010 midterms handed Wisconsin Republicans the governorship (Scott Walker) and both legislative chambers just as the new Census required fresh maps. In 2011, that trifecta passed a sweeping redistricting package often discussed under Act 43. The law rewrote state legislative lines and accompanied a new congressional map that would first be used in 2012.

Critics argued the maps packed Democratic voters into a small number of districts (especially around Madison and Milwaukee) while spreading Republican voters efficiently across the rest of the state. Democrats later challenged related maps in federal court; Gill v. Whitford became the national test case for partisan gerrymandering claims from this era, even though that suit focused on state Assembly districts drawn in the same cycle. The congressional results still illustrate the packing pattern in House elections.

In 2012, Barack Obama carried Wisconsin and Democratic House candidates won a statewide popular-vote plurality, yet Republicans kept a 5-3 House majority, the same seat split as in the Republican wave year of 2010. That mismatch between votes and seats is the core of this case study, and it persisted for the rest of the decade under the same basic map.

What the metrics show

This is a textbook packing story, so watch fairness, efficiency gap, and partisan advantage. Contestation stayed at 100%, which matters: the extreme scores are not an artifact of missing opponents.

• Seats stayed 3D/5R from 2010 to 2012 while Dem vote share rose from 44.6% to 50.8%. Under the old map in 2010, a Republican-leaning statewide vote produced a 5-3 GOP delegation. Under the new map in 2012, a Democratic popular-vote win still produced 5-3 GOP. That is the real-world packing result the metrics are built to detect.
• Raw fairness: 80.5 to 15.7 (−64.8 points). One of the sharpest state-level fairness collapses in the dataset, timed exactly to the first election under the new Census map. Fairness here is answering: did seats follow votes? In 2012, they clearly did not.
• Efficiency gap score: 73.2 to 0.0. The score bottoms out when wasted votes become extremely one-sided. That matches packing Democrats into a few landslide districts while Republicans win more seats by narrower margins.
• Partisan advantage index: +0.20 to +0.84. The signed tilt toward Republicans strengthens sharply once the new lines are in place, even in a year Democrats won more House votes statewide.
• Contestation remained 100%; EAI fell from 89.9 to 71.1 mainly via fairness. Voters still had major-party choices in every district. The accountability drop is structural (vote-seat mismatch), not a measurement failure from uncontested races. By 2018, the map still yielded 3D/5R with Democrats at 53.8% of House votes and fairness at 11.6.

What happened

After the 2010 Census, Pennsylvania’s Republican legislature drew a congressional map that produced durable GOP majorities through the decade. In 2016, under that map, Republicans held 13 of 18 House seats while Democratic candidates won about 46% of the statewide House vote. Critics labeled districts such as the old 7th among the most contorted in the country.

Voters and the League of Women Voters sued in state court. In early 2018, the Pennsylvania Supreme Court struck the map down as an unconstitutional partisan gerrymander under the state constitution (League of Women Voters of Pennsylvania). When the legislature and governor failed to agree on a replacement in time, the court imposed its own remedial map for the 2018 midterms. Federal challenges failed to stop the new lines before Election Day.

Under the remedial map, Democrats and Republicans each won 9 seats in 2018. In 2020, with the same basic plan still in place and a nearly even statewide House vote, the delegation stayed 9-9. That sequence (gerrymandered map, court invalidation, remedial election, then a second election under the fixed lines) is one of the cleanest modern before-and-after tests in the dataset.

What the metrics show

Because the IRL event is a court replacing the map mid-cycle, compare 2016 (old map), 2018 (first remedial election), and 2020 (second election under the same plan). Fairness, EAI, and partisan advantage should move if the court’s geometry actually changed vote-to-seat translation.

• 2016 old map: 5D/13R, Dem vote share 45.9%, fairness 11.3, EAI 57.2, partisan advantage +0.89. Seats were far more Republican than the statewide vote. High positive PAI and rock-bottom fairness are the metric fingerprint of that decade’s map.
• 2018 remedial map: 9D/9R, Dem vote share 55.1%, fairness 48.9, EAI 76.2. In the first election after the court order, seats equalized and fairness/EAI jumped. That aligns with the remedial map taking effect, though 2018 was also a Democratic midterm wave, so both the lines and the national environment helped Democrats.
• 2020 same plan: still 9D/9R, Dem vote share 49.4%, fairness 92.0, EAI 92.7, partisan advantage −0.06. With statewide House votes nearly even, the even seat split held and fairness peaked. That second year is the stronger evidence that the remedial geometry, not only the 2018 wave, produced proportional outcomes.

What happened

Michigan’s 2010-cycle congressional map, drawn by a Republican legislature, produced years of weak vote-to-seat balance. In 2012, Democratic House candidates won about 53% of the statewide vote but only 5 of 14 seats. That pattern helped fuel a reform campaign led by Voters Not Politicians.

In 2018, Michigan voters approved a ballot initiative creating an independent citizens redistricting commission and removing map-drawing power from the legislature. The new Michigan Independent Citizens Redistricting Commission held public hearings, drafted maps after the delayed 2020 Census data arrived, and adopted congressional lines for use in 2022, the first House election under the commission system.

Under that map, Michigan’s House delegation came out 7 Democrats and 6 Republicans on a statewide Democratic House vote share of about 51%. Compared with the 2010-cycle years, the commission era is a reform story rather than a packing story: voters changed who draws the lines, then the first election under those lines looked far more proportional in both seats and metrics.

What the metrics show

Compare the late 2010-cycle status quo (2018) and the first commission election (2022), with 2012 as the low-water mark that motivated reform. Fairness and partisan advantage are the headline scores; competitiveness stayed high, so this is not a contestation artifact.

• 2012 context: fairness 10.6, with Democrats at 52.7% of House votes but only 5 of 14 seats. That mismatch is the real-world problem the commission was sold as fixing. Fairness near the floor means seats were not tracking votes.
• 2018 (still the old-cycle map): 7D/7R, Dem vote share 54.0%, fairness 34.2, EAI 73.7, partisan advantage +0.63. Even when seats tied, Democrats still needed a clear popular-vote edge to get there, and PAI remained strongly Republican. Fairness had improved from 2012 but was still weak.
• 2022 (commission map): 7D/6R, Dem vote share 51.2%, fairness 92.2, EAI 94.2, partisan advantage −0.07. Seats nearly matched the statewide vote, fairness and EAI jumped into the top tier nationally, and PAI moved near neutral. Those shifts land in the first election after voters changed the mapmaking process, which is the IRL event this case study is about.

What happened

Massachusetts is a deep-blue state in presidential and statewide elections, and its nine House seats have been entirely Democratic for years. That dominance shapes candidate recruitment: in many cycles, Republicans struggle to field serious challengers, and some districts go without a major-party opponent at all.

In 2024, all nine incumbents won reelection. Republicans contested only two of nine districts (the 8th and 9th). In the other seven, Democratic candidates faced no Republican on the ballot, so raw district totals show essentially 100% Democratic major-party vote share. That is a ballot-structure and recruitment story more than a single mid-decade redraw, but it still reshapes every competition-based metric the project publishes.

The practical result for voters in those seven districts was no major-party choice in the House race, even as the state remained competitive enough at the top of the ticket to produce a clear Democratic presidential win (Harris carried Massachusetts by a wide margin). For the dataset, missing opponents force heavy use of adjusted estimates, which is why this case study is the clearest illustration of the raw-versus-adjusted split.

What the metrics show

Unlike the map-fight cases above, the IRL event here is missing Republican candidates. Lead with contestation, competitiveness, confidence, and the raw vs. adjusted gap. Fairness moves too, but it is harder to interpret when most races lack an opponent.

• Contestation fell to 22.2% (only 2 of 9 districts contested). That is the direct metric of the recruitment failure: seven districts had no Republican on the ballot. Every other score below is downstream of this fact.
• Raw competitiveness 20.2 vs. adjusted 57.6; raw EAI 22.9 vs. adjusted 37.0. Raw scores treat uncontested races as non-competitive and drag Massachusetts to the bottom of the national EAI ranking. Adjusted scores impute plausible two-party splits and partially recover the picture. The gap is the dataset saying: “headline numbers are punishing missing opponents, not just map geometry.”
• Raw fairness 18.2 vs. adjusted 27.8. Fairness still looks weak either way, but the jump under adjustment shows how sensitive seat-vote math is when seven districts have no recorded Republican votes. Use fairness cautiously here compared with Wisconsin or Texas.
• Metric confidence 61.1; distortion index 95.2. Confidence falls when uncontested races dominate. Distortion near the ceiling means raw and adjusted statewide scores disagree sharply, which matches a year when most districts required imputation. That is why the site publishes both pipelines.