The foreign born, especially recent immigrants, are believed to be a hard-to-count group which increases the likelihood of coverage error for this population. In fact, research has shown that English language ability, literacy skills, understanding of the census, residential attachment, and legal status are all factors that contribute to coverage error in censuses and surveys (Fein and West 1988; Iversen, Furstenberg and Belzer 1999; Martin 2007; Massey and Capoferro 2004). Because of data limitations, there have been no studies that empirically measure the coverage of the foreign-born population in the ACS. http://www.census.gov/content/dam/Census/library/working-papers/2015/demo/POP-twps0103.pdf
For complete information on the limitations of ACS data, go to http://www.census.gov/programs-surveys/acs/methodology.html.
Some considerations when using and interpreting the ACS data include:
ACS estimates represent the conditions that might have been present at any time within the estimate time period. ACS estimates should only be compared with like estimates. For example, 1-year data can only be compared with other 1-year data and cannot be compared with 3- or 5-year data.
If using the ACS for longitudinal analysis (comparisons over time), multi-year estimates should not overlap.