The SFVERC ACS Data
The Census Bureau’s American Community Survey (ACS) is a rich source of information about our communities; it covers our residents’ general, social, economic, and housing characteristics. The 2006 ACS data elements in these categories have been released for the recently designated San Fernando Valley Sub-County Census District (CCD) and for the special San Fernando Valley “Regions”. These data are reported in the tables listed in the left column. These tables present general, social, economic, and housing characteristics of our Valley and the Valley’s Regions and compare them to those of LA City, LA County, California, and the nation.
The San Fernando Valley CCD recently was created by a request from the Los Angeles County Board of Supervisors to the Census Bureau. The Bureau conducted a special tabulation for the new Valley CCD along with its inaugural release of the 2005 ACS data for the nation, states, counties, places, and other census statistical areas. We have posted the newly available 2006 along with the previously released 2005 ACS data for the San Fernando Valley, the Valley Regions, and its comparison areas.
The ACS data for the San Fernando Valley “Regions” are a legacy of a Census Bureau pilot project in Los Angeles County involving a special tabulation of the 2005 ACS data for the County’s Special Planning Areas (SPA) and designated subareas of these SPAs. A group representing several stakeholders and data providers (including the SFV Economic Research Center) developed these subareas to maximize the resulting data’s usefulness to planners and other interested parties. Early this year, the Center, United Way of Los Angeles, and the Healthy City Project contracted with the Census Bureau to obtain the 2006 ACS data for the same sub-regions of Los Angeles County. Again, we have posted both the 2006 and the 2005 ACS data for those sub-regions in the San Fernando Valley on this website.
Valley CCD vs. Valley Regions Coverage: The Valley’s boundaries for the Sub-County Census District (CCD) are shown in the accompanying CCD Map while the ACS Valley Regions boundaries appear in the Valley Regions map. Note that there are some minor differences in aggregate coverage of these two areas resulting from the boundaries of LA County’s SPAs, which existed prior to the designation of the Valley CCD. First, 13 of the Valley’s 15 defined Regions lie fully within the Valley CCD while two contain some extra territory. On the east end of the Valley Regions, North Glendale is combined with La Crescenta and La Canada to form the North Glendale-La Canada region. La Crescenta and La Canada are not in the Valley CCD but are on the eastern edge of the local SPA (SPA 2) and had to be included in its subareas.
A second difference between the CCD and the Regions involves Agoura Hills which is not considered to be in the Valley. The Census Bureau’s minimum population rule and SPA 2 boundaries required that we combine Agoura Hills with Calabasas and Hidden Hills in the western part of the Valley. The Census Bureau requires a minimum population of 65,000 in an area to qualify for annual releases of ACS data*. Because Calabasas and Hidden Hills alone do not contain enough population to meet the Census Bureau’s minimum size for annual data releases, Agoura Hills had to be included. In fact, the Census Bureau did not release the 2005 ACS data for the Calabasas, Hidden Hills, and Agoura Hills region because the population they counted fell just short of 65,000. As a result, Calabasas, Hidden Hills, and Agoura Hills data are not included in the 2005 ACS data tables but appear in the 2006 ACS tables. Note that the Census Bureau’s population rule for annual data release required us to combine what we may think of as separate Valley communities into single Valley regions to meet the 65,000 minimum, but we combined communities in light of their characteristics so that the larger areas were similar in important respects.
ACS 2006 and 2005 Coverage Differences—Area and Population: As mentioned above the 2005 ACS data for the Valley Regions does not include data on residents of the Calabasas, Hidden Hills, and Agoura Hills Area because the population that the Census Bureau counted there in 2005 did not total at least 65,000. Consequently, these data are not in the 2005 table but are in the 2006 tables, since the counted 2006 population did exceed 65,000.
One reason for the different population counts between 2005 and 2006 is that the 2005 ACS did not cover the entire population while the 2006 ACS did. The population consists of people living in households and group quarters. The difference is that the 2005 ACS surveyed only the Household Population, while the 2006 ACS covered Total Population consisting of both the household population and the group quarter population. This means that the characteristics data in the 2005 ACS data pertain only to household population while 2006 ACS data pertain to the entire population including both household population and group quarter population. Any comparisons between the 2005 ACS data and either the Census 2000 data or the 2006 ACS data (both of which includes the population in group quarters) must be interpreted in light of that difference. Group quarters include detention/correctional facilities, halfway houses, shelters, nursing homes and other residential care facilities, hospitals, dormitories, religious group quarters, military barracks, campgrounds, and people living in hotels or motels. For more information, see ACS Group Quarters.
Percentages vs. Population: The reader should note that the overriding purpose of the ACS, like its predecessor (the long-form survey), is to profile the people in the various geographies covered—enumerating their characteristics and providing estimated proportions or percentages of the residents with the certain age, gender, employment, housing, or whatever characteristics. And, like its predecessor, the ACS is not designed to count the total populations of the various geographies. That job falls to the constitutionally mandated, once-a-decade Census of the Population and its intervening updates from the on-going Population Estimates Program, in which the Census Bureau uses a number of indicators to gauge the change in the population of larger areas from the most recent actual Census count (Census 2000 in this case).
The ACS “Annual” Advantage: A strong advantage of the ACS over the Census data source it replaces—the “long-form” data in the decennial Census released in the Summary 3 file—is that the ACS is available on an annual basis for areas in excess of 65,000. In contrast, the Summary 3 file data was collected and available only every 10 years. The ACS advantage comes at a price though—the annual ACS sample is smaller than the long-form decennial Census was which, in sampling terms, means that the ACS survey have a larger sampling error than the long-form surveys did. The greater “margin of error” terms need to be taken into account relative to the ACS characteristic estimates whereas the larger long-form samples meant that the decennial Census estimates had (largely irrelevant) small margins of error.
ACS “Margin of Error” or Confidence Intervals: Please note that the tables also provide the confidence range for each estimate, which shows a 90% confidence interval around each ACS estimate. These confidence intervals are important because the annual sample size is small relative to the sample sizes in the previous decennial censuses. With the large samples of the past, the confidence intervals around the census estimates were so small as to be negligible, and therefore were not published. The smaller sample sizes of the ACS means that the confidence intervals around the estimates are larger and must be considered when comparing estimates over time. The 90% confidence interval for the Valley’s household population is +\-2.6%, and confidence intervals range from that level to +/-5% for mid-range numerical estimates, to +/-10% and beyond for the numerically smaller estimates. For more information, see accuracy of the data.
*The ACS is a new data release for the Census Bureau that replaces the so-called long-form decennial census data. Instead of releasing population, social, economic, and housing data for a large sample of the population every 10 years, the Census Bureau is conducting a smaller sample of the population every year. Over a 5-year period, the new ACS survey data will provide accurate sample data down to the census tract level. Each year the sample size will provide relatively accurate data for the larger census geographies (those with a population over 65,000) and data will be released for those larger geographies every year. Medium-sized census geographies (25,000 to 65,000) will get data releases after 3 years of data accumulation (as the sample size grows sufficiently to provide reliable estimates for this population level), and finally, ACS estimates will be available for census tracts after 5 years of data accumulation, which will provide a sample size sufficient for these still smaller populations.