provided by the U.S. Census Bureau’s International Data Base . Rates for Puerto Rico are
based on population estimates from the 2010 census as of July 1, 2012, and are provided by the
U.S. Census Bureau .
Rates by state and territory shown in this report may differ from rates computed on the
basis of other population estimates; rates for smaller population subgroups such as those for teen
mothers may be particularly affected by differences in population estimates. Birth and fertility
rates by month shown in Internet Table I–2 are based on monthly population estimates also
based on the 2010 census estimates. Rates for unmarried women shown in Tables 15 and 16 are
based on distributions of the population by marital status averaged over a 3-year period for
2011–2013 as reported by the U.S. Census Bureau in the March Current Population Survey
(CPS) for each year [67-69], which have been adjusted to July 1, 2012 (2010 census), population
levels  by NCHS’ Division of Vital Statistics .
Population estimates for the specific Hispanic groups
Beginning in 2011, birth and fertility rates for the specific Hispanic population groups
(Mexican, Puerto Rican, Cuban, Central and South America, and Other Hispanic populations),
shown in Tables 5, 7-9, and 14 of the 2012 Final Report , are based on population estimates
derived from the 1-year American Community Survey (ACS)  and adjusted to the U.S.
resident population control totals by the U.S. Census Bureau. Rates for the specific Hispanic
population groups prior to 2010 shown in the 2012 Final Report were based on population
estimates derived from the Current Population Survey (CPS) and adjusted to the U.S. resident
population control totals by the U.S. Census Bureau and/or NCHS’ Division of Vital Statistics.
The change to the ACS-based rates was made because ACS estimates are more statistically
reliable and represent the entire United States population. ACS estimates are based on an
approximately 3 million annual sample of the U.S. population, including all households (civilian
and military) and the institutionalized population (persons living in group quarters) . CPS
estimates are based on an approximate 200,000 sample of only the civilian, non-institutionalized
U.S. population . The larger ACS sample makes it possible to show rates in the 2012 Final
Report in more detail than in previous years, especially for Cuban and Puerto Rican women. The
2012 population estimates are derived from the 2012 1-year ACS and are adjusted according to
the (2010-based) postcensal estimates for July 1, 2012. The 2010 birth and fertility rates for the
specific Hispanic population groups were also revised using 2010 ACS-based population
estimates which were adjusted to the (2010-based) postcensal estimates for July 1, 2010. The
2012 special population estimates for Hispanic groups are presented in Table 1.
Revised population estimates
Birth and fertility rates by race for 2001–2009 shown in this report have been revised
based on newly released revised intercensal population estimates based on the 2000 and 2010
censuses, as of July 1 of each year, to provide more accurate rates for the period . A full
series of revised rates by state, by live birth order, and by age for 2001–2009 based on these
revised population estimates is forthcoming. These revised rates may differ from the intercensal
rates published in Births: Final Data for 2010 and the original rates published in Births: Final
Data for 2009 and earlier reports which were based on 2000 postcensal population estimates [75,
76]. Differences in the revised intercensal rates compared with previous intercensal rates are
slight and may vary by age and race and Hispanic origin population group.
The populations by race used in this report were produced under a collaborative
arrangement with the U.S. Census Bureau and are based on the 2010 census counts. Reflecting
the new OMB guidelines issued in 1997, the 2010 census (and 2000 census) included an option
for persons to report more than one race as appropriate for themselves and household members
. In addition, the 1997 OMB guidelines called for the reporting of Asian persons separately
from NHOPI. In the earlier 1977 OMB guidelines, data for API persons were collected as a
single group . For the nonmultiple-race reporting areas (10 states, American Samoa, U.S.
Virgin Islands, and Puerto Rico), birth certificates currently report only one race for each parent
in the categories specified in the 1977 OMB guidelines (see ‘‘Hispanic origin and race’’). In
addition, birth certificate data for the states using the 1989 birth certificate revision do not report
Asians separately from NHOPI. Thus, birth certificate data by race (the numerators for birth and
fertility rates) currently are largely incompatible with the population data collected in the 2010
census (the denominators for the rates).
To produce birth and fertility rates for 1991 through 2012, the reported population data
for multiple-race persons were bridged to single-race categories. In addition, the 2010 (and 2000)
census counts were modified to be consistent with the 1977 OMB racial categories, that is, to
report the data for Asian and NHOPI persons as the combined category of API . The
procedures used to produce the bridged populations are described in a separate publication .
Revised intercensal population estimates for the specified Hispanic groups from 2001
through 2009 shown in the 2012 Final Report are not currently available from the U.S. Census
Bureau and rates have been recalculated using population estimates prepared by DVS. The
population estimates were produced by applying proportions derived from the 2000-based
population estimates (according to year, sex, and age for the specified Hispanic population
groups) to the 2010-based population of Hispanic females by age group, and adjusting the sum of
the population estimates to be consistent with the total population of Hispanics females by age
(2010 based). These populations are available upon request from NCHS (Births@cdc.gov).
The population data used to compile birth and fertility rates by race and ethnicity shown
in the 2012 Final Report are based on special estimation procedures and are not actual counts.
This is the case even for the 2000 and 2010 populations that are based on the 2000 and 2010
censuses. As a result, the estimation procedures used to develop these populations may contain
some errors. Smaller populations, for example, AIAN, are likely to be affected much more than
larger populations by potential measurement error . Although the nature and magnitude of
error is unknown, the potential for error should be kept in mind when evaluating trends and
differentials. As more accurate information becomes available, further revisions of the estimates
may be necessary.
Residential population base --Birth rates for the United States and individual states are
based on the total resident populations of the respective areas (Table 2). These populations
exclude the Armed Forces abroad but include the Armed Forces stationed in each area. The
residential population of the birth- and death-registration states for 1900–1932 and for the United
States for 1900–2012 is shown in Table 3. In addition, the population including Armed Forces
abroad is shown for the United States. Table E in these Notes shows the sources for these
populations. A detailed discussion of historical population bases is presented elsewhere .
Small populations as denominators -- An asterisk (*) is shown in place of any derived
rate in the following situations: 1) the rate is based on fewer than 20 births in the numerator 2)
for the Hispanic subgroups, a relative standard error of 23 percent or more for the ACS-based
rates of 2010-2012 or 3) there were fewer than 50 women for census years and 75,000 women
for noncensus years in the denominator for the CPS-based rates for 1989-2009. Rates based on
populations below these minimum levels lack sufficient reliability for analytic purposes. These
guidelines follow the suggestions of the U.S. Census Bureau [77,78].
Net census undercounts and overcounts -- Studies conducted by the U.S. Census Bureau
indicate that some age, race, and sex groups are more completely enumerated than others.
Census miscounts can have consequences for vital statistics measures. For example, an
adjustment to increase the population denominator would result in a smaller rate compared to the
unadjusted population. A more detailed discussion of census undercounts and overcounts can be
found in the “1999 Technical Appendix” . Adjusted rates for 2012 can be computed by
multiplying the reported rates by ratios from the 2012 census-level population adjusted for the
estimated age-specific census over- and undercounts, which are shown in Table F of these
Cohort fertility tables
Various fertility measures for cohorts of women are computed from births adjusted for
underregistration and population estimates corrected for under enumeration and misstatement of
age. Cohort fertility tables are available through 2009 and have recently been revised and
updated to incorporate new rates for black women [79-82]. A detailed description of the
methods used in deriving these measures is available in an earlier publication as well as detailed
data for earlier years .
Total fertility rates
The total fertility rate is the sum of the birth rates by age of mother (in 5–year age
groups) multiplied by 5. It is an age–adjusted rate because it is based on the assumption that
there is the same number of women in each age group. The rate of 1,880.5 in 2012, for example,
means that if a hypothetical group of 1,000 women were to have the same birth rates in each age
group that were observed in the actual childbearing population in 2012, they would have a total
of 1,880.5 children by the time they reached the end of the reproductive period (taken here to be
age 50 years), assuming that all of the women survived to that age.
Seasonal adjustment of rates
The seasonally adjusted birth and fertility rates are computed from the X–11 variant of
Census Method II . This method, used since 1964, differs slightly from the U.S. Bureau of
Labor Statistics (BLS) Seasonal Factor Method, which was used for Vital Statistics of the United
States, 1964. The fundamental technique is the same in that it is an adaptation of the
ratio-to-moving-average method. Before 1964, the method of seasonal adjustment was based on
the X–9 variant and other variants of Census Method II. A comparison of the Census Method II
with the BLS Seasonal Factor Method shows the differences in the seasonal patterns of births to
Computation of percentages, percentage distributions, and means
Births for which a particular characteristic is unknown were subtracted from the figures
for total births that were used as denominators before percentages, percentage distributions, and
means were computed. The percentage of records with missing information for each item is
shown by state in Table B. The mean age of mother is the arithmetic average of the age of
mothers at the time of birth, computed directly from the frequency of births by age of mother.
An asterisk is shown in place of any derived statistic based on fewer than 20 births in the
numerator or denominator.
Computation of Measures of Variability
Random variation and significance testing for natality data
For information and discussion on random variation and significance testing for natality
data, with the exception of specified Hispanic groups (see below), see the 2010 User Guide .
Specified Hispanic population groups
Birth and fertility rates for Mexican, Puerto Rican, Cuban, Central and Other Hispanic
populations are based on population estimates derived from the ACS [71,72] for 2012 and
adjusted to resident population control totals. As a result, the rates are subject to sampling
variability in the denominator as well as random variability in the numerator .
The standard error for birth and fertility rates (SE(R)) is calculated as:
where R denotes the fertility or birth rate, B the number of births, SE(P) the standard errors of the
ACS population estimates, P. The standard errors of ACS population estimates used in the 2012
Final Report are presented in Table 1 with the population estimates by Mexican, Puerto Rican,
Cuban, and Other Hispanic.
When the number of births is large, a normal approximation may be used in calculating
confidence intervals and statistical tests. In general, for birth and fertility rates, the normal
approximation performs well when the number of births is 100 or greater. Formula 2 is used to
calculate 95 percent confidence limits for the birth or fertility rate when the normal
approximation is appropriate:
1.96(SE(R)) and U (R)
Or, substituting the relative standard error terms in formula 2,
and U (R)
where L(R) and U(R) are the lower and upper limits of the confidence interval, respectively. The
resulting 95 percent confidence interval can be interpreted to mean that the chances are 95 in 100
that the ‘‘true’’ birth or fertility rate falls between L(R) and U(R).
For example, suppose that the general fertility rate for Mexican women is 76.8 per 1,000 women
age 15-44 (based on 587,429 births and a population of 7,648,815 women). The standard error of
the ACS population estimate is 293,528. Lower and upper 95 percent confidence limits using
Formula 3 are calculated as:
Documents you may be interested
Documents you may be interested