The following is quoted from a helpful user guide on the ACS (for more information on the how to use the ACS, please see the full document here: http://www.ofm.wa.gov/pop/acs/
Is ACS data comparable to Census 2000 long form data? What do I need to know?
1. The ACS is based on a smaller sample: The number of ACS households sampled each year is fairly small (about 1 in 40), compared to the much larger sample for the Census long form (about 1 in 6). The smaller sample size associated with the ACS means that there is increased error surrounding estimates produced by the ACS compared to the decennial Census.
2. Point of time vs. period estimates: Data collection for the decennial census targets a specific point in time whereas ACS data are collected continuously. While decennial census enumeration lasts from around mid-March to late August, most of the data is collected around the April 1st time period and, for all practical purposes, represents the characteristics of the population as of April 1st in the decennial census year.
ACS data are collected on a continual basis and are combined to represent the characteristics of the population over a period of one, three, or five years. For more information about period estimates, please see Appendix 1.
3. Residence rules: The Census 2000 defines place of residence as “usual place of residence” while the ACS defines it as the place a person has resided for two months or more. This can impact the count and the characteristics of the population in places such as college towns or other areas with highly seasonal populations. For more information, please see place of residence in Appendix 1.
4. Income: The Census 2000 captures income from the previous calendar year while the ACS captures income from the previous 12 months. For example, ACS 2007 income data reflect income received between January 2006 and November 2007. All income data are adjusted for inflation using the National CPI for the last year in the series. For more information on income and other dollar measures, please see income, other dollar measures, and gas and electric costs in Appendix 1.
5. Poverty: Because poverty data are based on income, the differences between the 2000 Census and the ACS are the same as those listed above under income.
6. Migration: The Census 2000 captured place of residence five years ago for the population aged five and older, while the ACS captures place of residence one year ago for those aged one year or more.
7. Group quarters (GQ): GQs are included in the ACS from 2006 onwards. The Census 2000 collected a complete count of persons in GQs. The ACS includes fewer population groups in the The ACS includes fewer population groups in the 3 GQ category than the Census 2000.
The ACS does not survey people in “domestic violence shelters, soup kitchens, regularly scheduled mobile food vans, targeted non-sheltered outdoor locations, crews of maritime vessels and living quarters for victims of natural disasters” (U.S. Census Bureau, 2009). In addition, some GQ categories no longer exist such as a “residential care facility providing ‘Protective Oversight’, hospital/wards for the chronically ill and hospitals/wards for drug/alcohol abuse” (U.S. Census Bureau, 2009).
In addition, the ACS is based on a sample of the GQ population controlled to the U.S. Census Bureau’s estimate of GQs at the state level while Census 2000 values were based on a complete count. GQ data contained in the ACS may not be reliable for sub-state geographies, particularly in areas where few GQs exist.
8. School Enrollment: While both the Census 2000 and the ACS ask essentially the same question used to determine school enrollment, the reference periods are different. Census 2000 asked whether each person had attended school or college since February 1st, while the ACS asks about the last three months.
For most Census 2000 households the reference period for the school enrollment question was around two months, but for some the reference period was as short as six weeks and as long as seven months. The ACS, in contrast, asks about school enrollment for the last three months, and data are collected from January to December.
Given typical school schedules, one might expect to find differences in enrollment patterns based on these two reference periods. For more information, please see the following Census Bureau presentation: Measuring Education: A Comparison Between the Decennial Census and the American Community Survey.