Skip Navigation
Expert Brief

Maryland’s Extreme Gerrymander

The drawing of the Sixth Congressional District’s lines merits scrutiny.

Published: March 7, 2019

In late March, the Supreme Court will hear a case where Repub­lican voters argue that Maryland’s Sixth Congres­sional District is an uncon­sti­tu­tional partisan gerry­mander.

Mary­land defends the map, saying that legit­im­ate consid­er­a­tions related to the histor­ical config­ur­a­tion of the district drove decisions about district bound­ar­ies. But there are good reas­ons to be suspi­cious of that claim, on its face. The biggest of these is simply the extremeness of the changes made to the exist­ing district.

The Sixth District was over­pop­u­lated by about 17,414 people as Mary­land star­ted the 2010 redis­trict­ing cycle.[1] But we estim­ate that Demo­cratic map draw­ers, rather than tweak the district at the edges to achieve the popu­la­tion parity that the Consti­tu­tion requires, moved a total of 711,162 people into or out of the district (that’s 353,088 people moved into the district and 358,074 people moved out).[2] That’s more than 40 times the number needed to meet popu­la­tion equal­ity require­ments.

To gauge how that popu­la­tion shift compares with changes made in other districts around the coun­try, the Bren­nan Center analyzed needed versus actual popu­la­tion shifts for every congres­sional district in the United States. The results were extraordin­ary.

An earlier study, by Antoine Yosh­inaka and Chad Murphy, found that members of the out-of-power party tended to have more drastic changes made to their districts as a way for the in-power party to insu­late its members.[3] But our analyses show that what happened in Maryland’s Sixth District was well outside the norm. Only 7 of 132 districts in compar­able states saw changes as extreme as that in the Mary­land Sixth. And five of those were in North Caro­lina, where an aggress­ive racial gerry­mander of the state’s congres­sional map was struck down by the Supreme Court in 2016.

By almost any meas­ure, the Mary­land Sixth is an outlier. We show where the Mary­land Sixth falls relat­ive to all the districts that were redrawn after the 2010 census, and then in rela­tion to slices Yosh­inaka and Murphy iden­ti­fied as signi­fic­ant in their study: among districts that are a similar distance from popu­la­tion equal­ity or among simil­arly popu­lated states.

Needed versus Actual Changes to Districts: The National View

One of the main legal require­ments of redis­trict­ing is to redraw elect­oral districts so that districts have the same number of constitu­ents.[4] For most districts, this is a small under­tak­ing. We estim­ate that, exclud­ing the seven at-large districts, an aver­age of 59,976 people needed to be moved in or out of a district in 2011 to achieve popu­la­tion equal­ity. The Mary­land Sixth District, need­ing just 17,414 people, was well below this aver­age.

As Figure 1 shows, in states that did not gain or lose a congres­sional district (with the excep­tion of Cali­for­nia and, to a lesser extent, North Caro­lina), the changes made to achieve popu­la­tion equal­ity tended to be roughly propor­tional to the number of people that needed to be moved. In other words, if a district needed to lose 10,000 people then some­where around 10,000 people were moved out of the district. In short, states were minim­al­ists, chan­ging districts as little as possible.

The main excep­tion among states where the number of districts stayed the same was Cali­for­nia, where a newly created inde­pend­ent commis­sion undid an infam­ous bipar­tisan gerry­mander that had used “mangled lines” to make every seat a safe seat regard­less of party.[5] The new map “radic­ally, and more logic­ally, rearranged the state’s 53 seats” and resul­ted in a higher-than-aver­age number of moves. The other note­worthy excep­tion was North Caro­lina, where Repub­lic­ans, after gain­ing control of the redis­trict­ing process for the first time since Recon­struc­tion, “painstak­ingly packed Demo­cratic voters into just three of the state’s 13 seats.”[6] Here, too, large numbers of people were moved in and out of districts despite the fact the size of North Caro­lin­a’s congres­sional deleg­a­tion remained the same.

Not surpris­ingly, changes also were more extens­ive if a state gained or lost congres­sional districts. When this happened, even districts that needed to add or lose only a small number of people some­times shif­ted consid­er­ably to accom­mod­ate the increased or decreased number of districts. The yellow dots in Figure 1 indic­ate that a change in the size of a state’s congres­sional deleg­a­tion has a pronounced effect on the popu­la­tion shifts within the state. The previ­ous study by Yosh­inaka and Murphy found this to be the case in the 2000 redis­trict­ing cycle as well.

Even viewed at the national level, Maryland’s Sixth District stands out. Although it is in the middle of the pack among all districts, almost all districts where a greater number of people were moved were either in states that gained or lost a congres­sional seat or in Cali­for­nia, where a gerry­mander was being unwound.

Drilling Down: Changes to Districts in Compar­able States

When we omit districts in the states that gained or lost a district and in Cali­for­nia (Figure 2), and focus on states where the number of districts stayed the same, the aggress­ive­ness of the changes to the Mary­land Sixth becomes even clearer.

Needed vs Actual Population Changes

In this set of 132 compar­able districts, the Mary­land Sixth is almost two stand­ard devi­ations above the mean in terms of the popu­la­tion that was moved into the district during the 2010 redis­trict­ing cycle. In this set of cases, there are only seven districts that moved more people—on an abso­lute basis—than the Mary­land Sixth. Put another way, the Mary­land Sixth moved more people into the district than 94 percent of the compar­able districts in the coun­try. Of the districts that moved more people, five of them are in North Caro­lin­a’s racially gerry­mandered map.

Indeed, the only districts not in North Caro­lina where more people were moved were the Color­ado Second District, which was drawn by a court after legis­lat­ive dead­lock, and the Tennessee Fourth District, which was drawn by a Repub­lican legis­lat­ive trifecta.

Drilling Down Further: Changes to Districts Need­ing Similar Levels of Popu­la­tion Adjust­ment

We confirm our obser­va­tions about the Mary­land Sixth by look­ing at districts in compar­able states that had to move between 50 percent and 150 percent of the number that needed to be added to the Mary­land Sixth. As Figure 3 shows, out of 44 districts in this category, only three districts, once we set aside the Cali­for­nia and North Caro­lina cases, had a higher number of moves than the Mary­land Sixth. Even count­ing Cali­for­nia districts, the Mary­land Sixth still ranks in the top third. We depart from Yosh­inaka and Murphy’s research design by includ­ing districts from Cali­for­nia here.[7] Our inten­tion is to demon­strate the extent to which the inde­pend­ent redis­trict­ing commis­sion worked to remove the bipar­tisan gerry­mander in that state.

Population Shifts in Congressional Districts

Yosh­inaka and Murphy noted that states with a larger U.S. House deleg­a­tion tend to see more disrup­tion to districts, and we wanted to reduce the possib­il­ity that this factor would skew our analyses of the Mary­land Sixth District. There­fore, Figure 4 shows popu­la­tion move­ment only for states with seven to nine congres­sional seats, compar­able to Maryland’s eight. It is appar­ent from this figure that the changes to the Mary­land Sixth were not only unusual among states but unusual within Mary­land. Most of Mary­land saw minimal popu­la­tion move­ment between districts in the 2011 round of redis­trict­ing. By contrast, the shifts between the Sixth and Eighth Districts stand out. They are far larger than any other shifts in Mary­land, and also excep­tional among similar-size states. In total we report 276 obser­va­tions in this slice of data. The two indic­ated obser­va­tions from Mary­land are the 16th and 18th in order of the popu­la­tion shift, or more extreme than 94 percent of the other obser­va­tions in the figure.

In sum, the Mary­land Sixth is an example of a district that deserves close exam­in­a­tion to ensure that district lines were not redrawn purely on the basis of partisan motives.
 


Appendix: Meth­od­o­logy

We used a vari­ety of data from the 2010 U.S. Census in the analyses we conduc­ted for this report. We began with the total popu­la­tion estim­ates for each block group in the United States. Census data are aggreg­ated at differ­ent geographic levels, from small units like neigh­bor­hoods to the coun­try as a whole. The most gran­u­lar unit of geography in the U.S. Census is the census block, a unit of area defined as bounded by visible features (e.g., a street) and nonvis­ible bound­ar­ies, like prop­erty lines and city limits.[8] There are more than 11 million census blocks in the 50 states and the District of Columbia. Census blocks are nested within block groups, the second-most gran­u­lar layer of aggreg­ated Census data and the basis of our analyses here. A block group is a contigu­ous stat­ist­ical divi­sion of a census tract and gener­ally contains between 600 and 3,000 people.[9] There are 217,740 block groups in the United States, exclud­ing Puerto Rico and the Island Areas.[10]

Once we had counts of the total popu­la­tion in each block group, we joined this data with a map of block groups in each state. The U.S. Census has produced geograph­ical inform­a­tion systems (GIS) data since the 1990 Census as part of its TIGER data­base.[11] These GIS files allow research­ers to assign a given data point to a loca­tion in space to then relate it to other points in space. For example, the total popu­la­tion of the first block group within the first census tract of Alleg­any County in west­ern Mary­land is 826. We can add this unit of geography together with other block groups to divide Mary­land into congres­sional districts.[12] The Census Bureau’s TIGER data­base includes the neces­sary data files to define each congres­sional district in the United States and then to project those shapes on a computer-based map. For our purposes, the congres­sional districts in the 111th Congress (which sat from Janu­ary 2009 to Janu­ary 2011, follow­ing the 2008 elec­tions) and 113th Congress (in session from Janu­ary 2013 to Janu­ary 2015, follow­ing the 2012 elec­tions) allowed us to estim­ate the pre- and post-redis­trict­ing popu­la­tion in each congres­sional district.

We accoun­ted for the popu­la­tions moved around in the course of the redis­trict­ing process using the districts as a sort of bookend. We over­laid a map of the congres­sional districts as they were configured in the 2010 elec­tions on top of the map of block groups in each state to assign a block group to one district (or more than one, if a block group is split).[13] We then repeated the same process using the map of districts as they were configured in the 2012 elec­tions to assign each block group to a district (or districts). By compar­ing the 2010 district number with the 2012 district number, we could separ­ate the popu­la­tions that stayed in the same district from those that moved from one district to another. These popu­la­tion estim­ates were then directly incor­por­ated into our analyses of differ­ent slices of the data from each congres­sional district in the graph­ics we show and discuss.

This proced­ure allowed us to assess the accur­acy of our popu­la­tion estim­ates. The U.S. Census provides offi­cial pre- and post-redis­trict­ing counts of the total popu­la­tion of each congres­sional district.[14] In the table below, we compare our post-redis­trict­ing popu­la­tion estim­ate for states with more than one congres­sional district to the numbers provided by the U.S. Census. We report the minimum, aver­age, and maximum percent­age differ­ence between our district popu­la­tion estim­ate and the offi­cial post-redis­trict­ing popu­la­tion enumer­a­tion. Our estim­ates are, on aver­age, within about one-tenth of one percent­age point of the offi­cial count in each state. At the district level, our estim­ates are no more than about 4 percent­age points below the offi­cially repor­ted quant­ity and no more than 2.5 percent­age points above the offi­cial report. Our estim­ates are some­what less accur­ate as the number of congres­sional districts increases, but they do not rise to the point of markedly devi­at­ing from the offi­cial reports. The analyses we present in this report follow from these data, a combin­a­tion of offi­cial Census figures and our supple­ments to these popu­la­tion counts.

 

 


[1] This differ­ence in popu­la­tion was calcu­lated on the basis of pre- and post-redis­trict­ing total popu­la­tions repor­ted for the district. The U.S. Census repor­ted that the pre-redis­trict­ing total popu­la­tion in the Sixth District was 738,943. The General Assembly of Mary­land repor­ted that the post-redis­trict­ing popu­la­tion, after adjust­ing the numer­a­tion to account for insti­tu­tion­al­ized persons, of the Sixth District was 721,529. General Assembly of Mary­land, County Popu­la­tion Totals by District, Octo­ber 2011, http://mlis.state.md.us/Other/Redis­trict­ing/grac­sb1h­b1­county­pop.pdf.

[2] We describe the meth­od­o­logy lead­ing to this estim­ate in an appendix that follows our analyses. We estim­ate much of the popu­la­tion that moved out of the Sixth District after 2010 ended up in the Eighth District, and smal­ler portions moved to the First, Seventh, and Second Districts. Popu­la­tions that moved into the Sixth District after 2010 origin­ated in the Eighth and Fourth Districts.

[3] Antoine Yosh­inaka and Chad Murphy, “Partisan Gerry­man­der­ing and Popu­la­tion Instabil­ity: Complet­ing the Redis­trict­ing Puzzle,” Polit­ical Geography 28 (2009): 451–462, https://doi.org/10.1016/j.polgeo.2009.10.011.

[4] The initial claim in the 1962 Supreme Court case Baker v. Carr, from which much of the redis­trict­ing juris­pru­dence is descen­ded, was that the Tennessee legis­lature had not reap­por­tioned itself in the 61 previ­ous years, to the point that Justice Bren­nan, writ­ing for the major­ity, observed that “the State Senator from Tenness­ee’s most popu­lous senat­orial district repres­ents five and two-tenths times the number of voters repres­en­ted by the Senator from the least popu­lous district, while the corres­pond­ing ratio for most and least popu­lous House districts is more than eight­een to one.” The Court depar­ted from previ­ous prac­tice and signaled that the judi­ciary could step in to curb excesses in redis­trict­ing.

[5] Michael Barone and Chuck McCutcheon, The Almanac of Amer­ican Polit­ics 2014 (Chicago: Univer­sity of Chicago Press, 2013), 129. The legis­lature-drawn plan in Cali­for­nia is an example of a bipar­tisan gerry­mander designed to insu­late incum­bents. In 2002, for example, “[t]he smal­lest margin of victory for any Cali­for­nia congres­sional incum­bent was 18 percent­age points, and the aver­age incum­bent received a 68 percent vote share.” See Richard Forgette and Glenn Platt, “Redis­trict­ing Prin­ciples and Incum­bency Protec­tion in the U.S. Congress,” Polit­ical Geography 24 (2005): 934–951, https://doi.org/10.1016/j.polgeo.2005.05.002. The map in Cali­for­nia was also durable. In the elec­tions between 2002 and 2010, only one U.S. House seat changed parties.

[6] Barone and McCutcheon, The Almanac of Amer­ican Polit­ics 2014, 1233.

[7] Yosh­inaka and Murphy observe, “If large inter-district popu­la­tion shifts produced by redis­trict­ing can alter an incum­bent’s career and reelec­tion prospects, it is only natural to expect partisan mapmakers to use this tool stra­tegic­ally, and foster more instabil­ity for some incum­bents than for others” (see Yosh­inaka and Murphy, “Partisan Gerry­man­der­ing and Popu­la­tion Instabil­ity,” 452). The use of this tool by nonpar­tisan insti­tu­tions like courts or inde­pend­ent commis­sions is less well known in the academic liter­at­ure.

[8] See https://www.census.gov/geo/refer­ence/gtc/gtc_block.html for the defin­i­tion of a census block.

[9] See https://www.census.gov/geo/refer­ence/gtc/gtc_bg.html for the defin­i­tion of a block group.

[10] See https://www.census.gov/geo/maps-data/data/tallies/tract­b­lock.html for a tally of census tracts, block groups, and blocks for each state and territ­ory within the United States.

[11] The TIGER data­base is avail­able at https://www.census.gov/geo/maps-data/data/tiger.html.

[12] ESRI, Inc., the major developer and distrib­utor of GIS data and soft­ware, has created a train­ing module to famil­i­ar­ize users with its GIS soft­ware by draw­ing congres­sional districts in Mary­land using Census and geographic data: https://learn.arcgis.com/en/projects/redraw-polit­ical-bound­ar­ies-with-public-parti­cip­a­tion.

[13] We made an assump­tion that the popu­la­tion of block groups is uniformly distrib­uted within the block group. In cases where a block group is split, we estim­ated the popu­la­tion of each segment of the block group by multiply­ing the total popu­la­tion of the block group by the ratio of the area of the segment of the block group to the area of the whole block group. For example, if a block group covers 100 square miles, has 100 people in it, and is equally split between two districts, we assigned a popu­la­tion of 50 to each district.

[14] Pre-redis­trict­ing total popu­la­tion estim­ates by congres­sional district are avail­able at https://fact­finder.census.gov/bkmk/table/1.0/en/DEC/10_SF1/P1/0100000US.50000.111. Post-redis­trict­ing total popu­la­tion estim­ates by congres­sional district are avail­able at https://fact­finder.census.gov/bkmk/table/1.0/en/DEC/10_113/P1/0100000US.50000.113.