Gerrymandering and the 2018 midterm elections

What is gerrymandering?

The name “gerrymandering” reveals the term’s American roots, dating back to a districting plan enacted in Massachusetts in 1812, approved by the state’s governor, Elbridge Gerry. A map revealed one of the districts to have a serpentine shape, thought to resemble a salamander, immortalised in a cartoon in the Boston Gazette and referred to as a “Gerry-mander”.

Media reports of gerrymandering continue to dwell on the irregular shapes that gerrymandered districts sometimes exhibit; the 4th Congressional district of Illinois, shown in the map above, is a case in point. But irregular district shapes are neither necessary nor sufficient for a districting plan to be a gerrymander. For instance, North Carolina’s Congressional districts (displayed below) do not appear to have odd shapes, but as I show in the next section, do encode a sizeable pro-Republican advantage. The combination of (a) high rates of party loyalty among voters and (b) modern software for spatial data analysis and map-making means that today’s gerrymanders do not necessarily have districts with “odd shapes”.

Partisan control of elections in the United States

Elections for Australia’s national parliament are administered by the Australian Electoral Commission, a non-partisan, professional agency, charged with maintaining the Commonwealth electoral roll, the conduct of elections and conducting electoral redistributions, dividing states and territories into seats for Australia’s House of Representatives. Australian political parties make submissions to redistribution commissions, but at least in recent decades, the determination of electoral boundaries appears to be free of partisan interference and manipulation.

The situation is different in the United States, where state and local governments have responsibility for the administration of elections. Article I, Section 2 of the U.S. Constitution provides that every ten years, a census shall be conducted, in order to determine the apportionment of the House of Representatives seats (Congressional districts, or CDs) across the fifty states, in proportion to the population of those states.² Within any given state, the US Supreme Court insists on strict adherence to equal population in each state’s CDs, ruling out an older form of electoral manipulation known as malapportionment.³

Every U.S. state entitled to more than one CD engages in “redistricting” at the top of each decade, once the results from the decennial Census are provided. With the Census held in the “0” year (e.g., 2000, 2010, 2020 etc), the first set of elections held on the new districts are usually those in the “2” year (e.g., 2002, 2012, 2022 etc). Thus, in most US states, the district boundaries in place for the 2018 midterms are those drawn in 2011-2012, after the 2010 Census.

In at least 26 US states, redistricting is a political affair, an act of the state legislature, requiring majority approval from the houses of the state legislature and the governor’s assent. In 2011-12, Republicans controlled the redistricting process in 17 of these 26 states and Democrats in 5 states. Earlier this year the Pennsylvania State Supreme Court overturned a Republican-drawn plan and instated a court-drawn plan. Courts or independent redistricting commissions controlled redistricting in 14 states.

The following table shows the frequency of different types of redistricting plans in place for the 2018 midterms, and the number of CDs under each plan. Note that 141 CDs — about 1/3 of the Congress — will be elected under plans drawn by Republican state legislatures and governors. Plans enacted by courts or commissions span 16 states and 224 CDs, or about half the Congress. Democrats controlled the redistricting process in just 5 states spanning 42 CDs, or about 10% of the Congress.

Control	# States	#CDs	States
Commission or Court	16	224	AZ, CA, CO, CT, FL, IA, KS, MN, MS, NJ, NM, NV, NY, TX, VA, WA
Democratic	5	42	AR, IL, MA, MD, WV
Divided	2	11	KY, OR
Republican	15	141	AL, GA, IN, LA, MI, MO, NC, NE, OH, OK, PA, SC, TN, UT, WI
Small states (1 or 2 CDs)	12	17	AK, DE, HI, ID, ME, MT, ND, NH, RI, SD, VT, WY

Partisan gerrymandering: an example

Partisan gerrymandering is a direct consequence of partisan control of electoral redistricting. A partisan gerrymander is created when the controlling party “packs” their opponents’ voters into a relatively small number of districts and “cracks” their opponents’ remaining voters such that they form minorities in the remaining districts. Indeed, more extreme partisan gerrymanders can see the favoured party winning a majority of seats in a jurisdiction without a winning a majority of votes.

An example appears in the following table, showing the results of the 2016 election to the US House of Representatives across North Carolina’s 13 CDs.

District	Dem Votes	Rep Votes	Dem %	Winner	Clinton %
1	240,661	101,567	70.3	D	67.5
4	279,380	130,161	68.2	D	68.2
12	234,115	115,185	67.0	D	68.4
13	156,049	199,443	43.9	R	44.0
2	169,082	221,485	43.3	R	43.6
9	139,041	193,452	41.8	R	42.8
5	147,887	207,625	41.6	R	39.8
8	133,182	189,863	41.2	R	41.1
6	143,167	207,983	40.8	R	41.4
7	135,905	211,801	39.1	R	39.9
10	128,919	220,825	36.9	R	36.4
11	129,103	230,405	35.9	R	34.0
3	106,170	217,531	32.8	R	36.9
State-wide	2,142,661	2,447,326	46.7	3D/10R	48.1

These data reveal several characteristics of partisan gerrymandering. First, although Democrats win 46.7% of the House vote state-wide, they win only 3 out of 13 seats; conversely, Republicans win 10 out of 13 seats with 53.3% of the vote.⁴

Second, it is evident that the Democratic vote has been “packed” into three districts (1, 4 and 12), where Democrats win by margins of no less than 67-33. The remaining Democrat votes in North Carolina are dispersed throughout the 10 seats won by Republicans, with no Republican winning by less than 56-44 margins.

Third, the apparent “packing” and “cracking” of Democratic partisans is no accident nor merely a peculiarity of the candidates and issues in North Carolina’s Congressional races in 2016, but is also closely mirrored in votes for Hillary Clinton in the presidential election of 2016. That is, the districts have been designed with the partisanship of the voters in mind, as the Republicans who drafted this set of boundaries brazenly conceded.⁵

The North Carolina case reveals that Democrats would require a 6.1 percentage point swing to pick up a seat in that state. With a 6.1 point state-wide swing, the Democratic state-wide share of the vote across the state would rise to 52.8%, yet Democrats would still only have four (31%) of North Carolina’s 13 CDs. In order to win seven (a majority) of North Carolina’s 13 CDs, the state-wide swing needed is 8.4%, at which point Democrats have 55.1% of the state-wide, two-party vote for Congress. That is, Democrats require far in excess of 50% of the vote in order to win 50% (or more) of North Carolina’s CDs. Conversely, Republicans can retain 7 out of North Carolina’s 13 seats with as little as 44.9% of the vote.

Partisan gerrymandering is widespread

North Carolina is not the only case of a pro-Republican gerrymander stifling the ability of Democrats to convert votes into seats. Similar phenomena play out in many other US states.

Consider a state like Florida, with a Republican-drawn map for its 27 House seats. Republican House candidates there won 53.1% of the two-party vote for Congress, and 16 out 27 seats (59%). Democrats could pick up an additional seat there with a 4.9% swing, but require a 8.6% swing — or 55.5% of the state-wide, two-party vote to win a majority of Florida’s CDs. Conversely, Republicans can retain a majority of Florida’s CDs with just far less than a majority of the state-wide Congressional vote (44.6%).

A similar story holds in Michigan, another state with a Republican-drawn map. Republicans won 8 of Michigan’s 14 seats in 2016 with 50.6% of the state-wide, two-party vote for Congress. Democrats require a 6.9% swing to pick up an additional seat, which would take their state-wide vote to 56.3% of the two-party vote. That is, Republicans can retain a majority of Michigan’s CDs with as a little as 43.7% of the state-wide vote.

Ohio has 16 CDs, which split 12 Republican to 4 Democrats in 2016. A 9.2% swing is needed to unseat the most marginal Republican; the most marginal Democratic seat is on 17 point margin, indicating the extent of “packing” in the Republican-drawn boundaries. Relative to 2016, Democrats require a swing of 16.2% to win a majority of Ohio’s House seats, at which point their state-wide share of the two-party vote for Congress would be 58.0%. Conversely, Republicans can retain 8 of Ohio’s CDs with just 42% of the state-wide vote.

Maryland is perhaps the clearest case of a pro-Democratic gerrymander at present. Republicans hold just one of Maryland’s 8 CDs, winning it 70-30, consistent with Republican voters being packed. With a 8.3% swing Republicans could flip the most marginal Democratic seat, but require a 15.2% swing before Maryland’s eight CDs split 4D-4R. Such a swing would imply Republicans would need to win 52.2% of the state-wide, two-party Congressional vote in order to win 50% of the state’s seats, which is a mild level of partisan bias relative to the pro-Republican biases in Ohio, Michigan, Florida and North Carolina.

2018 implications

With polls suggesting Democrats will out poll Republicans by about 8 percentage points, one might reasonably conclude that Democrats will win a comfortable majority in the House. They may. But gerrymandering systematically reduces the number of marginal Republican districts, meaning that Democrats must reach quite high “up the tree” to win the 23 or so seats they need to form a majority.

The following table lists the 30 most marginal Republican seats by their marginality based on 2016 results. Just nine seats are on margins of less than four percent, with Democrats needing swings of six percentage points or more in order to win the seats needed for them to win a majority of the House.

Recent large swings in House elections are rarely this large. In 2010, when Democrats lost the House, they suffered a swing of 8.1 percentage points; in 2006, Democrats took the House with 5.4 point swing; in the 1994 midterms, Republicans took the House with 5.4 percentage point swing. A uniform swing of five to six points will probably be insufficient to deliver a majority to the Democrats. Much depends on where the swing occurs.

	State	District	Incumbent	Open Seat	Swing needed	Trump margin
1	CA	49	Darrell Issa	Yes	0.3	-3.7
2	NE	2	Don Bacon	No	0.6	1.1
3	TX	23	Will Hurd	No	0.7	-1.7
4	MN	2	Jason Lewis	No	0.9	0.6
5	CA	10	Jeff Denham	No	1.7	-1.5
6	NY	22	Claudia Tenney	No	2.7	7.7
7	VA	10	Barbara Comstock	No	2.9	-5.0
8	CA	25	Steve Knight	No	3.1	-3.3
9	IA	1	Rod Blum	No	3.8	1.8
10	AL	2	Martha Roby	No	4.1	16.0
11	CO	6	Mike Coffman	No	4.2	-4.5
12	NY	19	John Faso	No	4.3	3.4
13	PA	8	Brian Fitzpatrick	No	4.5	0.1
14	ME	2	Bruce Poliquin	No	4.8	5.1
15	FL	27	Ileana Ros-Lehtinen	Yes	4.9	-9.8
16	FL	18	Brian Mast	No	5.3	4.6
17	KS	3	Kevin Yoder	No	5.4	-0.6
18	PA	16	Lloyd Smucker	No	5.4	3.4
19	NJ	7	Leonard Lance	No	5.5	-0.6
20	FL	26	Carlos Curbelo	No	5.9	-8.1
21	NC	13	Ted Budd	No	6.1	4.7
22	TX	7	John Culberson	No	6.2	-0.7
23	UT	4	Mia Love	No	6.2	3.4
24	MI	11	Dave Trott	Yes	6.4	2.2
25	NC	2	George Holding	No	6.7	4.8
26	CA	21	David Valadao	No	6.7	-7.8
27	IN	9	Trey Hollingsworth	No	6.8	13.4
28	MN	3	Erik Paulsen	No	6.8	-4.7
29	IA	3	David Young	No	6.9	1.8
30	AZ	2	Martha McSally	Yes	7.0	-2.5

By design, partisan gerrymandering also suppresses the responsiveness of elections, stifling the translation of more votes for one party into additional legislative seats for that party.

The following graph shows how the magnitude of the swing required — and the size of the Democratic majority with respect to votes — in order to overcome pro-Republican gerrymandering. The effect of the pro-Republican gerrymanders is to shift that “steep” or “responsive” component of the seats-votes curves (below) to the right of the graph.

Nationally, Democrats could require as much as 55% of the two-party vote for Congress in order to win 50% of the seats in the Congress. Democratic seat share grows only slowly as Democratic vote share grows from 49% into the low 50% range. More than a few Republican incumbents enjoy comfortable margins of electoral safety, in the neighbourhood of 5 percent or greater. On the other hand, if a Democratic wave is sufficiently large - and rises up into the high single digits, sufficient to overwhelm gerrymanders protecting Republican incumbents - relatively more Republican seats will fall.

Partisan gerrymandering is a corruption of American democracy

Under a fair system, each party should have roughly the same chance of forming a majority if they win roughly half of the votes. That is manifestly not the case in contemporary American elections.

The systematic, deliberate manipulation of district lines — with the express intent of suppressing the responsiveness of legislatures, laws and policy to public opinion — is a corruption of American democracy, and alas, will likely play a key role in the 2018 midterm elections.

Endnotes