The first step in the line drawing process is preparing a database. In a redistricting database, each data item reflects a single geographic unit. The geographic units, to be used as the basic building blocks for creating the districts, can reflect administrative areas such as counties or cities and towns, census geography such as census blocks, or election geography such as voting areas. The database should include as much of the following information as possible for each geographic unit:
- a unique name or identification number for each unit
- additional identifiers, such as town or city, county, and state or province for each unit
- the total population or the number of registered voters for each unit
Additional demographic data, such as total population and voting age population counts by race or ethnicity for each unit, may also be included if useful for redistricting.
The Use of Digitised Geographic Data
If geographic information systems (GIS) software is to be used to develop district plans, the redistricting database must also include information related to the location of the geographic unit (spatial co-ordinates). This information allows the computer to automatically generate computer screen maps that reflect each new assignment of a geographic unit to a district.
The Use of Political Data
If the redistricting database is to include political information, then voter registration counts by political party, if available, and votes for candidates from previous elections should be entered for each geographic unit. The number of previous elections to be included in the database depends on the availability of previous election results, the amount of time required to enter the results into the database, and the financial resources available for creating the database.
If a redistricting database is to include both political and census demographic data, the election geography and the population geography must correspond with one another. If they do not, they will have to be matched.
Matching Census and Election Geography
Matching census geography and election geography can be quite challenging. The easiest method for matching these units is to merge the smaller geographic units until they approximate the boundaries of the larger unit. If, for example, census data is available for units as small as city blocks, then these blocks can simply be aggregated together until they match an election precinct (voting area). On the other hand, if the units of election geography are smaller, then they can be merged to produce political data for the larger pieces of census geography (e.g., villages and towns). It is always easier to aggregate data up to the higher level than it is to try and break it down to smaller levels of geography.
In the United States, however, the courts often require states to draw districts with census geography rather than election precincts (voting areas). This requirement can be handled in one of two ways:
- election precincts and their corresponding political data can be merged until they approximate census tracts;
- alternatively, precinct level election results can be disaggregated down to the census block level.
If redistricters want the flexibility of being able to draw at the smaller census block level (and many states do, in fact, prefer this approach), then political data reported at the precinct level have to be disaggregated and distributed among the census blocks that fall within that election precinct. This process is difficult, involves a great deal of estimating, and is subject to a great deal of error. Because it is a very tedious and labour intensive task, it will absorb a substantial amount of the database construction time.