by Jeffrey S. Passel, Senior Demographer and D’Vera Cohn, Senior Writer
The number of Hispanics counted in the 2010 Census has been larger than expected in most states for which the Census Bureau has released detailed population totals so far, according to an analysis by the Pew Hispanic Center, a project of the Pew Research Center. The gap between the Census 2010 count and Census Bureau population estimates has been widest in states with relatively small Hispanic populations.
The Census Bureau has released 2010 Census counts of Hispanics in 33 states; Hispanics accounted for the majority — 58% — of population growth over the decade in those states. Those states’ combined Census 2010 total of 38.7 million Hispanics was higher by 590,000 people1 (or 1.5%) than the bureau’s own estimates for those states. Census counts for Hispanics in the remaining states will be released by the end of March.
In 23 of the 33 states, census counts were higher than the latest census estimates of Hispanics by at least 2%. In three states, the census count was at least 2% lower than the census estimate. In the remaining seven states, the difference was less than 2% in either direction. By comparison, the difference between census estimates and the 2010 Census count for the total population in these 33 states was well under 1% (.2%).
Unlike the decennial Census, designed to be a 100% count of the U.S. population, the Census Bureau’s population estimates are annual updates of counts from the previous census based largely on birth certificates, death certificates, immigration data and other government records.2 The most recent published state population estimates for Hispanics were as of July 1, 2009. For this analysis, the Hispanic estimates were updated to Census Day, April 1, 2010, by extrapolating the 2009 estimates based on each state’s Hispanic population growth rate from 2008 to 2009.
The Pew Hispanic Center analysis indicates that states with large percentage differences between their Hispanic census counts and census estimates also were likely to have large percentage differences between census counts and census estimates for their total populations. This reflects the large role that Hispanics play in overall population growth. Hispanics have accounted for most of the discrepancy between 2010 Census counts and census estimates of states’ total populations.
In addition, states with small Hispanic populations and states with rapidly growing Hispanic populations were most likely to have a mismatch between their census counts and census estimates of Hispanics, according to the Pew Hispanic Center analysis. As a group, the 27 states that have Hispanic populations under a million people (including many where Hispanic counts grew sharply) had a larger percentage gap between their census counts and census estimates than did the six states with larger, long-duration Hispanic communities.
The six traditional Hispanic states for which Hispanic counts from the 2010 Census have been released include Arizona, California, Colorado, Illinois, New Jersey and Texas. Each has more than a million Hispanic residents and collectively, 31% of their population is Hispanic. As a group, those states house 30 million Hispanics, according to the 2010 Census, yet their aggregate census count was only 88,000 (or .3%) larger than their aggregate census estimate.
In the other 27 states, Hispanics make up 7% of the total population. These states as a group house 8.7 million Hispanics, and their combined 2010 Census count was 501,000 people (or 6.1%) higher than their combined census estimate. Among them are Alabama, where the Hispanic census count of 186,000 people was 16% higher than its census estimate, the largest gap among states seen so far. At the other extreme, the census count of 22,000 Hispanics in South Dakota was 10% below the most recent census estimate. (Smaller populations by nature tend to be more volatile than large ones, so even a small numerical change could result in a large percentage change.)
In the six states with large Hispanic populations, two had notable gaps between census estimates and census counts. In New Jersey, the census estimate was 4.6% short of the census count of 1.555 million for Hispanics. In Arizona, the census count of 1.895 million Hispanics was 8.7% lower than the estimate; it also was lower than the Census Bureau’s estimates for 2008 and 2009. The gap in Arizona was almost entirely due to a lower-than-expected Census count in Maricopa County, which includes Phoenix.
As the accompanying table shows, there were differences between census counts and census estimates for Hispanics in most parts of the country. A notable exception is the Mountain State group of Colorado, Nevada and Utah, none of which had census counts more than 1% different from their estimates.
Accuracy of Estimates
The accuracy of these census population estimates is important not only because they are the major source of basic demographic data in the years between census counts, but also because they are the basis for distributing billions of dollars in federal funds during those years. They are relied on for sample design and weighting in widely used federal surveys, including the bureau’s own American Community Survey and the Current Population Survey from which federal unemployment and poverty rates are calculated. The estimates also are used to calculate birth and death rates for the total population and for sub-populations such as race and ethnic groups.
The Census Bureau has invested study and effort over the past decade to improve its population estimates after the publication of 2000 Census counts pointed to a shortfall in census estimates published in the 1990s.
In 2000, the population estimate for April 1, 2000 of 274.5 million was about 7 million people short of the census count for that day of 281.4 million people, or 2.5%. Later analysis attributed much of the gap to a low census estimate of Hispanics, the nation’s largest minority group. The 2000 Census count of Hispanics of 35.3 million was nearly 10% larger than the official estimate for April 1, 2000 of 32.2 million.
Much of the problem, the bureau concluded, was that the estimates failed to account for growth in the number of unauthorized immigrants. Analysts also concluded the 1990 Census count had been too low, so the estimates began from a base that was too small.
At the state level, the gap between 2000 Census counts and census estimates of Hispanics was even wider (for this analysis the 1999 estimates were extrapolated to Census Day 2000). In eight states, the count was 50% or more above the estimate, higher than any variation found in the 2010 state census counts released so far. In only three states was the census count within 2% of the census estimate.
The bureau made several changes to its population estimates methodology over the past decade. Most notably, it began including state-level data obtained from the American Community Survey, which collects information on characteristics of the U.S. population, including immigrants. The bureau also devoted additional effort to outreach in the 2010 Census to groups that have been hard to count in the past, such as immigrants.
2. The Census Bureau also analyzes a sample of federal tax returns for people who moved from one state to another (linked to other data on age, sex, race and ethnicity of the tax filers) to calculate the number and characteristics of in-migrants and out-migrants for each state. For group quarters such as prisons and college dormitories, the bureau mainly relies on counts supplied by states and localities.