TITLE
Recommendation to review the population data by Council District and direct City Manager to work with the appropriate Departments to conduct the next population review for redistricting after the 2020 decennial census. (Citywide)
DISCUSSION
The City Council is authorized by the City Charter to review the City population by Council District and enter into a redistricting process if there is an uneven distribution of residents. Typically, cities redistrict by Ordinance every decade, one year after the decennial census data is released. The City Charter states the City Council may choose to redistrict every five years or whenever it is determined to be necessary. Mid-decade redistricting is possible, though it requires additional levels of population estimation to extrapolate from the available data. The last redistricting process occurred in 2011 and was done based on 2010 decennial census data.
Section 103 of the City Charter reads in part: The City shall be divided, for electoral purposes, into nine (9) Councilmanic Districts approximately equal in population. Commencing the second quarter of 1981 and at intervals of five (5) years, or at any other time the City Council may direct, the Planning Commission shall ascertain the number of inhabitants in each Councilmanic District and report its findings to the City Council. If the report shows that the Councilmanic Districts are not approximately equal in number of inhabitants, the City Council shall, by Ordinance, redistrict the City into nine (9) Councilmanic Districts, each having approximately an equal number of inhabitants.
On March 2, 2017, the Planning Commission reviewed population estimates by Council District and unanimously voted to transmit the population estimates to the City Council. The City Council should use this data to determine if there is a significant imbalance between Council District populations that would prompt the need to redistrict. The Council District ideal population (citywide population divided by nine) and the percentage difference from the ideal population are key factors in making this determination. The limited data available mid-decade from the U.S. Census Bureau limits the accuracy and usefulness for redistricting. The 2011-2015 American Community Survey (ACS) five-year estimates data were used to estimate the population for each Council District. Table 1 depicts the population by Council District in 2010, estimates for 2015, and the percent difference from the ideal population. A negative percentage indicates that the district has a smaller population than the ideal distribution; a positive percentage indicates that the district’s population is larger than the ideal.
Table 1: Population by Council District
|
Council District |
2010 Total Population |
Percent Difference from 2010 Ideal Population |
2015 Estimate Total Population |
Percent Difference from 2015 Ideal Population |
|
1 |
49,117 |
-4.37 |
49,051 |
-6.59 |
|
2 |
51,218 |
-0.28 |
51,106 |
-2.68 |
|
3 |
52,371 |
1.96 |
53,298 |
1.50 |
|
4 |
51,405 |
0.08 |
52,249 |
-0.50 |
|
5 |
49,852 |
-2.94 |
50,732 |
-3.39 |
|
6 |
49,444 |
-3.73 |
49,310 |
-6.10 |
|
7 |
52,013 |
1.27 |
56,075 |
6.78 |
|
8 |
53,009 |
3.21 |
56,090 |
6.81 |
|
9 |
53,828 |
4.80 |
54,702 |
4.17 |
|
Total |
462,257 |
- |
472,613 |
- |
|
Ideal Distribution |
51,362 |
- |
52,513 |
- |
|
+/- 5% range |
48,794-53,930 |
- |
49,888-55,138 |
- |
While not defined within the City Charter or any statute, as a general rule, Council Districts should be within 5 percent above or below the ideal population (citywide population divided by nine). According to the ACS data, Long Beach's population increased by 2.19 percent between April 2010 and December 2015. Population change varied by Council District, with six districts increasing and three districts decreasing. The percent difference from the ideal population varies between 0.5 and 6.81 percent among the nine districts. The estimates show that five of nine Council Districts are within the 5 percent margin. The estimated population changes are shown in the attached map (Exhibit A - Population Change Map by Block Group). The divergence in population distribution is within the statistical margin of error of the data itself. There is no guarantee that a redistricting process would result in an actual population that is closer to the ideal population or whether the various sampling and estimating errors could in fact counterbalance any attempt to bring the district boundaries toward the ideal population.
Every ten years, the U. S. Census Bureau conducts a “point in time” count of every person in the United States. When this decennial census data is first released, it is the most accurate population data available. However, the data continually ages and becomes less accurate year after year. In the intervening years, the U. S. Census Bureau completes rolling estimates of population known as the ACS. The ACS data are “time period” estimates, not a complete population count. The U.S. Census Bureau mails a survey to a random sample of addresses to collect population and other demographic data. The survey is sent to approximately 1 in 38 households per year, which accounts for between 0.5 and 10 percent of households in a given census tract. Non-responding households are contacted by phone, but unlike the decennial census, there are no serious consequences for a resident who chooses not to respond.
Though the ACS data is the best available data source for mid-decade population estimates, this data can be problematic for redistricting. In the ACS data, survey response rates, sampling, weighting, and statistical adjustments vary by census tract. The ACS data is survey data, not a complete population count. Survey data inherently includes a range of uncertainty due to sampling error and the margin of error is high for the population estimate provided for this redistricting review. For ACS data, the smallest level of geography that the data is available for is the block group level. The margin of error for the block groups range from 6 to 2,607. For this data there is a 90 percent confidence level that the population estimate is within the margin of error. For example, if the population estimate for a block group is 1,313 persons and the margin of error is 559, then there is a 90 percent confidence that the population for that block group is between 754 and 1,872 persons (1,313 +/- 559). This is a wide range when discerning population difference from the ideal population of 52,513. The margin of error for the block group population estimates make it uncertain that redistricting would result in populations in each Council District that are closer to the ideal population when compared with the ACS population estimates in Table 1.
Exhibit B shows the Coefficient of Variation (CV) for each block group, which is the measure of relative error in the estimate (the amount of sampling error in the estimate relative to the size of the estimate itself). The CV is determined by the ratio of the standard of error to the value being estimated, expressed as a percentage. The lower the number, the higher the relative reliability of the estimate. Most of the block groups have a medium reliability, between 12 and 40 percent, which means the estimates should be used with caution. The reliability is increased if the data is reviewed at the census tract level instead of by block group, but that level of geography is not useful for redistricting purposes.
The margin of error varies drastically throughout the City based on the population and survey responses within each block group. Response rate to the surveys also vary significantly for different populations throughout the City. For example, people who live on the western and northern areas of the City may be less likely to return the surveys than people who live on the eastern or southern areas. Additionally, this issue is compounded because the data is reported at the census block group level, which does not align with Council District boundaries. Because of this, using ACS data poses challenges for assigning population estimates for block groups that are located within two Council Districts and does not provide a reliable means to calculate the population per Council District. In order to best estimate population in these scenarios, block group population estimates were split and assigned to Council Districts based on area within each district. While staff made appropriate assumptions to assign those block group population estimates to the correct Council District, the results are not statistically credible. The ACS data, therefore, does not have a standard margin of error the way the decennial census does. For analysis over a broad area, such as measuring the population of the state of California, the overall error remains low. However, for the task of local redistricting, which requires the use of census blocks to accurately determine Council District population, the error balloons to high levels in some block groups.
The redistricting process is meant to ensure that all Council Districts have nearly equal population. The City must comply with the California Election Code and the federal Voting Rights Act of 1965 for the redistricting process. In establishing the boundaries of the Council Districts, the City Council may give consideration to the following factors: (1) topography, (2) geography, (3) cohesiveness, contiguity, integrity, and compactness of territory, and; (4) community interest of the Council Districts.
Due to the issues with using ACS population estimates instead of the decennial census population counts, staff recommends conducting the next population review and redistricting process in 2021. However, the City Council has the option to begin the redistricting process now with the less accurate ACS data. During the last redistricting process in 2011, the City Council adopted criteria prior to the start of the process to guide staff. Should the City Council wish to continue with the redistricting process mid-decade, staff recommends the City Council adopt similar criteria to guide staff through the redistricting process. These criteria are important to ensure a smooth process, provide transparency in the process, allow for meaningful public input and ensure a legally defensible outcome. The previously City Council-adopted criteria would serve as a basis for developing an updated set of criteria (Exhibit C - 2011 Redistricting Process Criteria).
The California Environmental Quality Act (CEQA) requires analysis of discretionary projects carried out or approved by public agencies. Reviewing population data and providing direction to staff for redistricting is not defined as a “project” pursuant to Section 21065 of the CEQA guidelines.
This matter was reviewed by Assistant City Attorney Michael J. Mais on May 1, 2017 and by Budget Management Officer Rhutu Amin Gharib on May 4, 2017.
TIMING CONSIDERATIONS
City Council action is requested on May 23, 2017, to allow enough time to prepare for the redistricting process should the City Council choose to redistrict mid-decade.
FISCAL IMPACT
There is no fiscal or local job impact associated with the recommendation to conduct the next population review for redistricting after the 2020 decennial census. The preliminary estimated cost to conduct a mid-term redistricting process is between $150,000 and $200,000. This estimate is based on a citizens’ redistricting committee supported by two full-time staff members, including a GIS expert, technical staff to ensure that online maps are frequently updated for public viewing, an assigned City Clerk staff member to coordinate meeting minutes and agendas, and a project manager to shepherd the process. If the City Council decides to begin redistricting now, staff will return to the City Council with a more detailed estimate of costs.
SUGGESTED ACTION
Approve recommendation.
Respectfully Submitted,
AMY J. BODEK, AICP
DIRECTOR OF DEVELOPMENT SERVICES
APPROVED:
PATRICK H. WEST
CITY MANAGER