Population data is essential to the redistricting process. Redistricting operates by combining discrete geographically-based units of population in a series of assignments designed to produce districts of relatively equal population. The population data can be either total population counts based on a census enumeration or voter registration data; the geographic population units will therefore reflect either census geography or election geography (these units of geography are often, in fact, the same).
The Use of Census Enumeration Data
In the United States, the reapportionment of the House of Representatives of the Congress and the redrawing of congressional districts have always been based on a census count of the total population. The number of seats to which each state is entitled in the U.S. Congress is dependent on the relative size of the state's population according to the census enumeration. The U.S. Constitution requires the Census Bureau to conduct an enumeration of the entire population every ten years.
Therefore, a national census is conducted as of April 1 in the first year of each decade (for example, 1980, 1990, and 2000). The Bureau must report the results of the census to the president by December 31 of the census year. Since 1975, the Census Bureau has also been required to send population data specifically for redistricting purposes to each state governor within a year of conducting the census.
The redistricting data produced by the Census Bureau is referred to as PL 94-171 data (for Public Law 94-171). It includes population counts for several different levels of census geography, i.e., units as large as whole counties and units as small as census blocks (usually the equivalent of one city block) are reported in PL 94-171.
Although the Census Bureau collects data on a large number of population characteristics, only a few items are included in the data files sent to the states for redistricting. These items are total population, voting age population, and subtotals of the population for persons of Hispanic origin and for five major racial groups--white; black; Asian and Pacific Islander; American Indian, Eskimo, and Aleut; and "other" races. Population data by race and Hispanic ethnicity are needed to ensure that communities in which these minority groups predominate are not divided by electoral district boundaries. (An example of PL94-171 population data reported at the city level can be found in Use of Population Data in the Delimitation Process.)
Because children and non-citizens are counted in the U.S. census, the population base for redistricting in the United States includes many persons not eligible to vote. In the United Kingdom and many other countries, especially those with compulsory voter registration, the redistricting base is the number of qualified voters rather than the total population. The use of qualified voters as the redistricting base automatically eliminates children and non-citizens from the calculation. Whether this practice makes sense from the perspective of democratic representative theory is arguable.
The use of census population data as the redistricting base presents some problems, such as:
- census data may be inaccurate;
- census data can become obsolete;
- population equality does not necessarily produce an equal electorate.
Inaccurate Census Data
The U.S. Census Bureau estimates that in 1990 it undercounted the population by 1.6 percent. If certain racial or ethnic groups are undercounted at a higher rate than others groups, both the allocation of seats in a legislative body and the redrawing of legislative districts can be affected. Blacks in the United States, for example, were undercounted in 1990 at a much higher rate (approximately 5 percent) than the population in general. If a state is disproportionately undercounted because it has a large black minority population, the state could conceivably lose a congressional seat when the Congress is reapportioned.
In 2011, Canada adopted the use of Census population estimates for the initial assignment of seats to a province. The Minister responsible for introducing the legislation stated : “In determining the actual populations of the provinces, we’re using the population estimates. These are the same numbers, the same population estimates, that are used for the federal-provincial equalization program, the same numbers that are used for the Canada health transfer, the same numbers used for the Canada social transfer. So this is the best data available for the population figures for the provinces themselves.” The estimates correct for net undercoverage in the census and provide a more accurate representation of total provincial population.
Obsolete Census Data
Even if it is accurate at the time of enumeration, census data can rapidly become obsolete because of growth rates. The use of census data for redistricting may discriminate against localities experiencing high population growth, such as, for example, suburban areas in the United States. Over time, people in these suburban districts will become underrepresented compared to those in rural or urban districts.
Census Data May Produce Unequal Electorates
Even if all districts are drawn such that they have the exact same population, the number of electors per district would not necessarily be equal because the proportion of nonvoters—that is, noncitizens, those too young to vote, and even registered voters who choose not to cast a ballot – is not uniform across the population. In the United States, for example, a congressional district with a large Hispanic population is likely to have more children and more non-citizens than a district with few Hispanics. For this reason, the number of voters on Election Day is likely to be lower in the heavily Hispanic district than in other districts.
The Use of Voter Registration Data
These same problems affect the use of voter registration data as the redistricting base, although not to the same degree. Unless the registration system is very accurate, the registration count could be incorrect. For example, the registration rolls could be inflated due to failure to remove from the list deceased persons or persons who had moved from the district; alternatively, the rolls could be missing some potential voters if the lists are not updated often enough. And unless voting is compulsory, some registered voters may neglect to cast ballots on Election Day, making the number of voters unequal across districts. After some period of time, population shifts will render districts of unequal size, regardless of whether population or registration counts are used.
Australia has adopted a unique solution to the problem of population shifts over time. Federal electoral districts are redrawn in Australia using registration, or enrolment, figures--but not current enrolment figures. Instead, districts are redrawn based on projections of what the enrolment of each locality is likely to be in three and a half years, the midpoint of the redistribution period. (See the case study, "Federal Redistribution in Australia", for more information on the use of population projections.)
Conclusion
The choice of whether to use census data or voter registration data may be guided by either practical or theoretical concerns. For instance, reliable registration data may simply not be available or may not be adequate for redistricting purposes. In the United States, for example, racial data is needed to ensure that minority communities are not divided between districts, and racial data are usually not available from the registration rolls. On the other hand, registration rolls may provide the better measure of the current population when a general enumeration of the population is unavailable, outdated, or inaccurate.
Redistricting based on registration data is likely to produce districts that are more equal with regard to the number of voters contained in them. But are voters the only persons deserving of representation? From the point of view of democratic theory, an argument could be made that all persons, and not simply voters, should be represented. If a broader definition of "representation" is adopted--one that views representatives as acting on behalf of all his or her constituents, non-voters as well as voters--then the justification for the use of total population is that it produces equal constituent representation in a more expansive sense[1].
Notes:
[1] This discussion draws heavily from an article written by Carlton Henry entitled "The Impact of New Technology and New Census Data on Redistricting in the 1990s," in Redistricting in the 1990s: A Guide for Minority Groups, ed. William O'Hare (Population Reference Bureau, Inc., 1989)
