Redistricting Reform Bill Does Not Go Far Enough

Pennsylvania's 7th Congressional District

Pennsylvania’s 7th Congressional District

Pennsylvania is one of the most gerrymandered states in the country. Under Republican control, the Pennsylvania legislature crammed as many Democrats as possible into a small number of districts in the redistricting following the 2010 census. Accordingly, in the following Congressional election, despite winning the support of a majority of voters statewide, the Democrats lost a seat leaving themselves with only 5 seats while the Republicans retained 13 seats.

Pennsylvania State Senator Lisa Boscola (D)

Pennsylvania State Senator Lisa Boscola (D)

State Senator Lisa Boscola and her eight cosponsors have reintroduced bipartisan legislation which they hope will take partisanship and politics out of how the state’s legislative maps are created. As Senator Boscola said “We need a reliable system that honors the concept that voters choose their leaders, not the other way around. Gerrymandered legislative districts rob citizens of competitive races and spur the kind of partisan polarization that is playing out in both Washington, D.C. and Harrisburg.”

Too Late?

“It is crucial that we fix this redistricting process before the next round of reapportionment in 2020,” Boscola said.

Committee Minority Chair and bill co-sponsor Anthony Williams (D)

Committee Minority Chair and bill co-sponsor Anthony Williams (D)

PA Senate State Government Committee Chairman MIke Folmer (R)

PA Senate State Government Committee Chairman Mike Folmer (R)

Her bill now under consideration by the Senate State Government Committee would amended the Pennsylvania Constitution to create an independent citizen’s commission charged with drawing up both legislative and Congressional redistricting plans. To take effect, the bill would have to:

  1. be approved by a majority of both chambers during this session of the Pennsylvania legislature,
  2. be reapproved during the following session (beginning 2017), and
  3. be approved by the majority of voters in a statewide ballot.

Boscola introduced similar legislation in previous sessions, and the bills were assigned to Senate State Government Committee, but the committee never voted on them or reported them out. (Pennsylvania committee chairmen effectively have the power to veto legislation by simply failing to ever schedule a committee vote.) If this bill fares better and is approved in a timely fashion, it could still take effect before the 2020 census triggers another round of redistricting.

The State of Voting in Our State

Ideally, Democrats and Republicans would both have 9 seats in a fairly balanced state like Pennsylvania with 2,793,538 votes for Democrats, 2,710,070 votes for Republicans and 52,722 votes for other candidates in the 2012 Congressional elections following the 2011 redistricting.

If the Democrats controlled the redistricting process and could create equally populated districts to their fancy, they could divide the state into 18 districts each with around 155,196 Democratic votes, 150,559 Republican votes and 2,929 votes for independents.  That 2% edge would ensure a Democratic sweep.

Similarly without any limitations on partisan gerrymandering the Republicans could lock 308,685 Democrats in a single district, and divide the remainder of the state into 17 nearly identical districts each with around 146,168 Democratic votes, 159,416 Republican votes and 3,101 votes for independents, thus ensuring a 17-1 Republican majority.
Such extreme results illustrate the power of map drawers to impose their will on the voters. While the 13 to 5 bias in our Congressional delegation is not as bad as it could mathematically be, that is of little consolation to the Democratic majority who have under one-third the representation in our state’s Congressional delegation.

Too Little?

Would the proposed legislation actually decrease the amount of bias imposed by redistricting in Pennsylvania?

Amending the State Constitution is a difficult process especially in Pennsylvania. By law, if an amendment fails, we must wait five years before proposing it again, so we must make sure that any change proposed to redistricting will actually mitigate the bias it is designed to address.

Bad reform might be worse than no reform at all, since bad reform would forestall other possible reforms.

Missing the Forest for the Trees

The goal of redistricting is not merely to produce a pretty map whose districts have aesthetic forms with familiar boundaries. Redistricting and reapportionment are the first steps in the process by which voters in our state are given the opportunity to select their representatives. The goal is thus to increase the likelihood that the state’s legislature and Congressional delegation will reflect the interests, beliefs and priorities of our citizens.

To accomplish this mission, map makers have to anticipate how their maps will be used and to do this they need to understand the political geography of the state.

However, the proposed legislation sets up a Legislative and Congressional Redistricting Commission which would be expressly forbidden from doing this:

In establishing districts, no use shall be made of any of the following data:
(i) Addresses of incumbent legislators or members of Congress.
(ii) Political affiliations of registered voters.
(iii) Previous election results.

Thought Crimes

These pieces of information are public knowledge. Either we are hoping that the commissioners are woefully ignorant of their own state, or we are turning redistricting into a “thought crime”.

When a similar provision was proposed in 2010, I argued that Fair Districts Florida’s amendments were the wrong solution to the redistricting problem.

The first clause judges redistricting plans not on their effect but by their alleged motivation, turning redistricting into a “thought crime.” A redistricting plan requires input from scores of legislators, staff, and consultants familiar with statistics and demographic data. One could allege any of these participants of being motivated by political bias, and the redistricting plan would be rejected as a result. Inevitably, all discussion of the redistricting plan would have to be done off the record.

Florida voters approved of their amendments and the Florida Supreme Court accordingly rejected maps drawn by Florida’s Republican legislature after email records revealed that the map drawers had conspired to reduce Democratic representation. Nevertheless, Fair Districts Florida’s success may be short-lived. Republican legislature may have simply learned the lesson to keep their plots off-the-record and work their strategies out face-to-face instead of using email.

Conversely, actual honest legislators can easily be accused of violating the law and would be hard-pressed to prove that they did not take this “forbidden” information into account somehow in their map drawing.

Instead of forbidding map drawers from making use of this information, it should be provided to them and its use should be mandated. The resulting map should be scored on how likely it is to produce competitive elections resulting in a legislature or congressional delegation reflecting the will of the people.

Mandated Bias: An Unintended Consequence

One might think that all bias results from underhanded political gamesmanship. However, Boscola’s bill requires that “district boundaries shall coincide with the boundaries of political subdivisions” as much as possible subject only to population equality. This has the effect of locking up a huge number of Democratic votes in Philadelphia and Pittsburgh where they will win a single Congressional seat by an overwhelming margin and allowing the Republicans to win more seats state-wide.

The current system of redistricting can be and has been abused to create bias in favor of either party, but we can at least hope that future legislature might not abuse the system.

On the other hand, the proposed bill would mandate a requirement which would incidentally prevent fair maps from being drawn.

Ironically, instead of outlawing biased maps, the proposed bill would accidentally enshrine bias as a requirement in the Pennsylvania Constitution.

Hand it to the Courts

The bill provides for an 11-member Legislative and Congressional Redistricting Committee including four Democrats, four Republicans and three independents.

The final redistricting plan must be approved by at least seven affirmative votes, which must include at least one vote of a member registered from each of the two largest political parties in this Commonwealth based on registration and one vote from a member who is not registered with either of the two largest political parties.

If a final redistricting plan is not approved by a supermajority of the commission by October 16 of the year following the census, the Pennsylvania Supreme Court would then appoint “a special master to develop and complete a final redistricting plan.”   This would give a strong incentive to whichever party controls the Supreme Court to withhold its votes and let the Supreme Court decide in its place.

A Wolf in Sheep’s Clothing

The Democratic, Republican and independent commissioners are selected at random from pools of candidates. The candidate must have been active voters registered with that party for the last three years.

The lists are vetted by the Secretary of the Commonwealth but there is an appeal process in the legislation. Clearly, parties could act with guile and cross-register their supporters starting in 2018 so that they could flood all three applicant pools in 2021. (After candidates are selected they can re-register with their true party only to repeat the process in 2028.)

Worse yet, the bill supposes that everyone registered with a political party reflects the interests of that party state-wide. However, in Philadelphia, Pittsburgh and many other localities, the partisan balance is so skewed that the general election in November is a mere formality; the real contest is the party primary in April. Accordingly, some conservatives register as “Democrats” in order to continue to have some say over their local representation. Similarly, some liberals in rural areas register as “Republicans”.

Should any of these voters be selected randomly to serve on the commission, they could provide the crucial vote required for a map to be considered bipartisan.

Anti-Gerrymandering Standards

Given Boscola’s bill’s flaws, what alternative is there to the status quo where representatives choose their voters instead of voters choosing their representatives?

Imagine that multiple redistricting maps are proposed. We would calculate the partisan advantage for each district on each map as the average results in those districts from the three most recent top-of-ballot statewide elections.

If the partisan advantage is small, then either candidate is almost equally likely to win. If the partisan advantage is huge, then the “underdog” would be very unlikely to win. Such an upset might require a huge scandal involving the opposing candidate.

Historical data can be used to create a table by which we calculate the probability of an underdog party winning a given election despite a certain partisan disadvantage. For example, in Presidential elections 1828-2012, the underdog party would prevail in 26% of states where they had a 5% disadvantage in the previous elections.

By adding up the probability of the Democrats and Republicans winning each district, we can calculate the expected number of districts each party will win. The expected distribution of seats and the partisan advantage statewide is an indication of bias.

We should require that bias be reduced as much as possible. If anyone can produce a map with significantly less bias than the state’s map, then the state should be required to adopt that map.

Moreover, we can legislate other requirements. For example, you could construct an unbiased map in Pennsylvania by creating 9 districts dominated by Democrats and 9 district dominated by Republicans. This often happens through “bipartisan gerrymandering” whereby neighboring legislators from opposing parties trade their constituents in order to increase the ideological purity of their districts and mutually ensure their own reelection.

To avoid this we could look at the probability of underdogs winning on the proposed maps. This would be nearly 50% in a district which is almost balanced, nearly 0% in a heavily skewed district and 26% as mentioned above in a district with 5% bias. This probability is a measure of how competitive each district is. The total of these probabilities across all of the districts is a measure of competitiveness statewide.

Among maps with similar bias, the state should be required to choose the map with the greatest competitiveness. Other factors such as compactness can be required as desired.

Other standards have been proposed by various authors such as Prof. John Nagle of Carnegie Mellon University and Prof. Samuel Wang of Princeton University.

The question is not whether a better standard exists but whether our politicians have the political will to approve such a standard and if not, can a standard be imposed by the courts?

Prof. Wang has submitted an amicus brief applying his standard to the U.S. Supreme Court case Harris v. Arizona Independent Redistricting Commission. Oral arguments will be heard on December 8 and the court’s decision next year may affect redistricting reform efforts not only in Arizona, but across the United States.



  1. says

    The author is right that competitiveness is what the public wants – but not necessarily the legislators or the courts. The tests proposed for deciding if election districts are “competitive,” meaning that either major party has a fair chance of winning a hypothetical election, are complicated. Only a mathematician could love any of the ones I have seen. For a practicing lawyer, getting a judge up to speed on any of those tests is a problem, to say the least. Moreover, the tests do not bring us to a single unique district map, but rather, a continuum of more or less competitive (and more or less compact and contiguous) district maps. And historically, the courts have made much of this fact that there is no single “correct” answer. So they label redistricting issues as non-justiciable political matters. Second, someone might convince a legislator to support a proposed amendment to the state constitution that will guarantee him or her a compact, contiguous district that closely follows township and county lines. But telling the legislator that the district is guaranteed to be competitive and require a vigorous campaign … how do you sell that? Ohio just adopted a citizen commission to set election districts without considering political makeup at all. Makes sense!

    • pennsylvanian8808 says

      As mentioned above, this bill would also create a citizen commission.

      “…independent citizen’s commission charged with drawing up both legislative and Congressional redistricting plans.’

  2. burrowsx says

    Dan and I have been arguing about this for some time. I would first pass the sections of the Voting Rights Act that were recently gutted by the Federal Supreme Court as a state law, removing the objectionable language which limits their application to polities deemed with a history of segregation, to which the Court objected. This would give citizens the right to sue the Commonwealth and to file class actions, if they felt that the political maps (generated by whatever process) denied them fair representation for state or federal legislative districts. With the threat of legal costs, and possible civil rights convictions, legislators might worry about the people’s hammer dropping when they gerrymander their district maps.

    Another area where we differ is in defining what constitutes a gerrymandered shape. The criterion I would posit for a gerrymandered shape is twofold: (1) the district should have a border which contains at most 12 edges; (2) the square of the perimeter should not exceed 25 times the area of the enclosed district. To make this work, we need to define an edge within a border: first, any legislative district boundary which coincides with a larger fixed political district (city or county) will be considered a single edge; similarly, a legislative district boundary which coincides the edge to a body of water, or major transportation thoroughfare (e.g. rail or highway), will also be considered a single edge, regardless of sharp changes in direction. Edges to a legislative district will otherwise be counted whenever there is a discontinuity in the linear direction of the border. This definition encourages the use of natural boundaries, and the coincidence of legislative boundaries with county and urban polities. The ratio of perimeters and areas provides that the area enclosed by a legislative district border efficiently encloses its area. There have been cases where the Federal district court has imposed what ordinarily would appear to be a gerrymandered district. The law should posit that consideration of (gerrymandered) shape should be prioritized behind the equality of population, and the obligation of the legislature to assure that minority populations receive representation in proportion to their presence in the population.

    Unlike consideration of F-distribution or Chi-squared distributions of expected boundary map partitions (suggested in the article), the calculation of these gerrymandering requirements are both restrictive and relatively easy to compute. If enough states pass similar legislation at the local level, the re-passage of the Civil Rights provisions can become a national issue. It is important that we establish maps not by procedural guarantees, which can always be corrupted, but by empowering people to gain standing when their rights are violated. The current legislation does not provide sufficient detail to limit the partisan corruption of a supposedly non-partisan citizens committee. We know from the recent state constitutional convention that such “neutral bodies” can be hijacked, to do the work of political evildoers.

Leave a Reply