A few weeks ago, reports on the composition of the populace of the United States were disseminated widely throughout news media outlets. Their source for the data, the American Community Survey, was a noteworthy part of the story along with the actual numbers being reported.

For many people, this was their first introduction to the American Community Survey (ASC). Even though early trial forms of the ACS have been around since 1996, this is the first year that the ACS has been fully operational and has publicized a completed year's report. In the words of the United States Census Bureau, the department that operates the ACS, the ACS is

"a new nationwide survey designed to provide communities a fresh look at how they are changing. It is an ongoing survey which will allow the Census Bureau to provide the nation with demographic data on a yearly basis instead of every 10 years. The ACS is a critical element in the reengineered 2010 Decennial Program and will replace the census "long form" beginning in 2010. The Census Bureau will continue to use the census "short form" in the decennial census of population and housing to obtain the population count and other demographic information."

The statistic that this new program produces should be of great importance to all of us because these numbers are used to help influence many decisions made in government and the private sector. For example, the statistics reported might affect decisions that are made on school funding as well as decisions that your company may makes in their everyday planning. You can read about the ACS and the importance of the data they produce in a seminar given by the Brookings Institution.

The Census Bureau conducts many censuses and surveys. One of the other Census Bureau programs is the Population Estimates Program (PEP). This program has had year to year population demographics since at least the 1970s. PEP uses the decennial census (the census taken every tenth year since 1790) as a starting point for its figures and then derives numbers by adding and subtracting people based on births, deaths, and immigration records.

ACS derives it figures by conducting surveys in sample populations around the country. The sample size and selection is careful planned according to complex statistical methods. The ACS collects a wide variety of information that is similar to the data compiled with the long form surveys used in the decennial censuses. This can include general population data as well as data on income, occupation, housing type and cost, educational enrollment and attainment, and even commuter information. PEP only estimates population numbers organized by age, sex, and race/ethnicity.

Asian Pacific Islander Population Findings
Both the ACS and PEP show that the Asian Pacific Islander (API) American population continues to grow at a faster rate than the general population. Thus the API percentage of the total US population is getting more sizable each passing year. According to our calculations for the nation, the 2005 percentage of people that state they are all or part API is at 4.950sing the ACS numbers and 5.020sing PEP. PEP's figure for API and part API in 2000 is at 4.46%. The furthest back

figure that we could calculate from the ACS is 2002. That year's ACS API and part API figure is 4.68%.

A word about our reported API statistics - ACS and PEP do not combine Asian American and Pacific Islander data. People that list themselves as being part Asian American or part Pacific Islander may be members of both partial groups. In the calculations that we used to derive the above figures, we only counted the people that listed themselves as all or part Asian American and added that figure to the number of people that stated that they were all Pacific Islanders. Thus the all and part API figures listed above should actually be slightly higher. We will include links to all our statistical worksheets and the data that we collected from the Census Bureau website at the end of this article.

Continuing with the report, California's 2005 API and part API population is 13.8 0x0p+0ccording to the ACS and 13.89 0x0p+0ccording to PEP. The ACS 2002 figure is 13.35 0x0p+0nd the PEP 2000 figure is 12.9%.

For my home city of San Francisco, we were only able to derive API data from the ACS. The API and part API population in San Francisco for 2005 is 35.01 0x0p+0nd 34.83 0n 2002.

An interesting figure that you will see in the worksheets is a calculation for children in the 5-17 year age range. These numbers vary slightly up and down over the past five years, but as a percentage of total population, these figures are showing a slight downward trend because the overall United States and California populations are increasing. Overall, school age children's population is significantly higher than the highest figure during the baby boom era.

Accuracy
ACS numbers appear to be smaller than PEP's in most categories that we looked at. For example, the ACS 2005 total US population is 8,032,267 less than PEP's estimate. ACS's 2005 report also shows 616,349 fewer API and part API residents than PEP.

The most glaring difference between ACS and PEP for the API community is the Pacific Islander figure. For 2005 ACS reports only 397,030 Native Hawaiian and Pacific Islanders in the US, whereas PEP estimates their numbers at 516,612. Which figure is more accurate holds great importance to the Pacific Islander American community. Based on the ACS's general under reporting relative to the 2000 census and the PEP figures, it is our guess that the PEP figures are closer to the real Pacific Islander population size. Because of the serious consequences this number may have, we encourage you all to investigate the reasons for this discrepancy.

Accuracy of surveys like the ACS is only as good as the sample quality. If the sample is truly representative of the whole, then you will get a good view of the big picture. Nothing is as accurate as a complete count, but even the decennial census, which attempts to count everybody, misses a lot of people.

The ACS and PEP numbers reported in past years get revised each year, so what we tell you may not hold true as they collect more data and make revisions to their methodology. As the ACS sampling methods get better, it could potentially have superior accuracy to even the decennial census.

Data Extracted from the Census Bureau's ACS Website
• ACS data from 2000-2001 - 92K Excel File
• ACS data from 2002-2003 - 121K Excel File
• ACS data from 2004-2005 - 119K Excel File

Data Extracted from the Census Bureau's PEP Website
• PEP data for the US 2000-2005 - 82K Excel File
• PEP data for CA 2000-2005 - 43K Excel File