1
Statistical Primer No. 17 Division of Public Health
N.C. Department of Health and Human Services State Center for Health Statistics
STATISTICAL
PRIMER
State Center for Health Statistics • 1908 Mail Service Center • Raleigh, NC 27699-1908 • 919/733-4728
www.schs.state.nc.us/SCHS/
No. 17 May 2007
Understanding Different Population Data Sources
and Their Impact on Vital Statistics Rates
by
Kathleen A. Jones-Vessey
NORTH CAROLINA DEPARTMENT OF HEALTH AND HUMAN SERVICES
The State Center for Health Statistics (SCHS)
routinely publishes reports that rely on population
data for calculating rates. Some people falsely
assume that annual population figures are precise,
fixed counts that are not subject to change over
time. In reality, population figures are essentially
educated estimates of the population based on
extrapolating figures from a variety of sources such
as decennial Census data, births, deaths, school
enrollment, and migration patterns.
Traditionally, the SCHS has relied on population
figures generated by the State Demographer’s
Office. Most of our annual reports, including North
Carolina Vital Statistics Volume I, Volume II, the
Pregnancy report, and other SCHS publications
continue to report rates based on population
denominators derived from the annual July 1
Certified County Population Estimates produced by
the State Demographer’s Office. The State
Demographer produces population figures for each
county by single-year age groups and by race and
sex. Using the State Demographer’s estimates,
racial population estimates are only available for
whites and minorities and do not include figures for
Hispanic populations or for specific races other
than white. To produce their estimates and
projections, the State Demographer’s Office
develops and enhances complex mathematical
computer models, and collects and reviews a variety
of data from federal, state, and local government
sources.1 The certified population estimates
produced by the State Demographer are periodically
updated and revised after the initial release.
Additionally, the State Demographer’s files are
typically “smoothed” after the next U.S. Census to
adjust for population changes deduced from the
decennial Census. For the latest North Carolina
population data from the State Demographer’s
Office, go to demog.state.nc.us.
In recent years, alternative population data sources
have become available. The Population Estimates
Program of the U.S. Census Bureau in collaboration
with the National Center for Health Statistics
(NCHS) has begun to generate annual county-level
resident post-census “bridged population estimates.”
NCHS bridged population files take 2000 Census
population data which included 31 different race
categories (and which allowed individuals to choose
more than one race) and bridge it back to four
single-race categories (White, Black/African
American, American Indian/Alaska Native, and
Asian/Pacific Islander). The file also includes
population estimates for ethnicity (Hispanic/Latino
and non-Hispanic/Latino) by race. The post-census
2
Statistical Primer No. 17 Division of Public Health
N.C. Department of Health and Human Services State Center for Health Statistics
estimates are updated annually as additional data
become available. The file was originally produced so
that the new Census population data, which adhere to
the 1997 federal Office of Management and Budget
(OMB) standards for the collection of multiple race
data, could be used by researchers analyzing vital
records which still included only a single race.2
Given the changing population of North Carolina and
the growing need for more information measuring the
extent of health disparities, SCHS statisticians
routinely receive requests for rates based on more
detailed racial and ethnic breakdowns. In the past,
these requests could not be answered. However, with
the availability of NCHS bridged population data, race
and ethnicity-specific rates are now produced. Several
of our periodic reports, including the Racial and Ethnic
Disparities Report Card and the Hispanic Health Fact
Sheet, now use these files to generate rates for specific
races and for Hispanics/Latinos.
Hispanic origin is not considered a racial category, but
rather an ethnicity. As such, unless noted otherwise,
rates presented for specific races include Hispanics/
Latinos in one of the racial categories. Typically,
according to NCHS coding specifications, Hispanic/
Latinos are coded to “white” in vital statistics data.3
Using the NCHS bridged population file, the SCHS
now often presents data which combine race and
ethnicity categories: for example, white, non-Hispanic;
African-American/Black, non-Hispanic; Other races/
non-Hispanic; and Hispanic/Latino.
Differences in Federal and State
Generated Population Estimates
The SCHS continues to use the July 1 Certified
Estimates of the State Demographer’s Office for our
annual reports, while also using the NCHS bridged
population files for special studies or ad hoc requests.
One negative effect of this is that some of the rates
reported in our annual publications using the State
Demographer’s population data in the denominator do
not match rates presented in special or periodic reports
which use the NCHS bridged population data. Overall,
there is not a large difference between the two files.
Table 1 presents population estimates from 2005 using
the NCHS bridged population files and the North
Carolina State Demographer’s population files. In
total, there was only a 1,176-person difference in the
population estimates from the two files – with the State
Demographer’s file being lower than the NCHS file
overall for 2005.
The two files also show some variability with regard to
population estimates by age group. Figure 1 presents
2005 North Carolina population estimates from the
two files by age group. While the overall total
percentage difference was 0.0%, the percentage
difference varies for specific age groups. The young
adult age groups show the greatest differences between
the two files with the 20-24 and 25-29 age groups
reflecting differences of more than eight percent.
Table 1: 2005 North Carolina Resident Population Using the NCHS
Bridged Population File and the State Demographer’s Certified July 1
Population Estimates Maintained by SCHS
State
Demographer NCHS
Certified Bridged Numeric Percentage
Estimate Estimate Difference Difference
Gender:
Male 4,277,092 4,271,569 5,523 0.1%
Female 4,404,974 4,411,673 6,699 -0.2%
Race:
White 6,479,702 6,478,615 1,087 0.0%
Minority 2,202,364 2,204,627 2,263 -0.1%
TOTAL 8,682,066 8,683,242 1,176 0.0%
3
Statistical Primer No. 17 Division of Public Health
N.C. Department of Health and Human Services State Center for Health Statistics
0
100,000
200,000
300,000
400,000
500,000
600,000
700,000
800,000
00-04 05-09 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85+
Age Group
NCHS Bridged File State Demographer's File
Differences between the two population files by race
are harder to measure since the State Demographer’s
file does not contain specific race estimates other than
white and minority. Overall, the white population
estimates for 2005 generated by the two files differ by
1,087 people, with the State Demographer’s file
estimating higher figures than the bridged file. With
regard to the minority population, again there is
virtually no percentage difference between the files,
though the estimated minority populations differed by
more than 2,200 people.
Population estimates by gender are also available in
both of the population files. Comparing the bridged
population file estimates for 2005 and the 2005 State
Demographer’s population file, we do see some
differences in the estimates by gender. The 2005 State
Demographer’s file has 5,523 more male residents of
North Carolina than the NCHS bridged estimates. In
contrast, the NCHS bridged population file shows
6,699 more females for 2005 than the State
Demographer’s July 1 certified estimate. However,
given North Carolina’s large population, the
percentage differences with regard to gender in the two
files is extremely small.
Both the NCHS bridged population file and the State
Demographer’s certified July 1 file provide resident
population estimates for North Carolina counties.
Differences in county population estimates between the
two files range from a low of only five (Gates County)
to a high of 6,219 residents (Wake County – a
difference of less than one percent). More than half of
the counties (53) had higher 2005 population totals in
the State Demographer’s population file, while the
NCHS bridged population file revealed higher
population estimates for 47 counties.
Figure 1: 2005 North Carolina Resident Population Estimates
Population Estimate
4
Statistical Primer No. 17 Division of Public Health
N.C. Department of Health and Human Services State Center for Health Statistics
Differences in Population Estimates and
the Impact on Vital Statistics Rates
The State Center for Health Statistics frequently
presents vital statistics rates based on population
figures. Differences in population denominators
generated from the two different population files might
have an impact on some of our vital statistics rates. As
shown in Table 2, overall, the difference between rates
based on the State Demographer’s population file and
rates based on the NCHS bridged population file is
negligible with the exception of the 2005 teen
pregnancy rate and breast and prostate cancer death
rates. The teen pregnancy rate difference of three
percent between the two files is likely a result of the
differences we see in the population estimates for
females as well as for the younger age groups as
outlined in Figure 1. Caution should be taken when
generating age-specific rates for these younger age
groups using the two different population files. With
regard to age-adjusted breast and prostate cancer
mortality, differences in the rates are likely a reflection
of the differences in the population estimates by
gender in the two files, as well as the differences in the
younger age groups.
Differences in the two files with regard to county
population estimates have a small impact on county-specific
vital statistics rates as well. Unadjusted
mortality rates vary in some cases depending on which
population denominators are used. Six counties,
specifically Lee, Chatham, Orange, Swain, Hyde, and
Onslow counties, have rates that differ by more than
three percent depending on which 2005 population
estimates are used to calculate their resident mortality
rates. However, for the vast majority of counties, the
difference is negligible.
An additional issue to keep in mind with the use of the
NCHS bridged population file is that the file is updated
every year based on the latest available information.
Therefore, estimated North Carolina population figures
for 2003, for example, will vary as they are regularly
updated in later years. Variations in North Carolina
resident population estimates from the bridged
population files from 2002 to 2005 are presented in
Table 3. North Carolina’s 2001 population estimate
declined by approximately 7,800 from the estimates
produced in 2002 to the latest estimates produced in
2005. In contrast, the 2003 estimate increased by more
than 15,000 from 2003 to 2005.
Table 2: North Carolina Resident Rates Using State Demographer’s and NCHS Bridged Population Files
for 2005
Rate Based Rate Based
on State on NCHS Percentage
Demographer’s Bridged Difference
Population Population in the
Data Data Rates
2005 Teen (ages 15-19) pregnancy rate1 61.7 63.6 -3.1
2005 Birth rate2 14.2 14.2 0.0
2005 Age-adjusted mortality rate, all causes3 884.2 873.1 1.3
2005 Age-adjusted mortality rate, heart disease3 211.5 208.6 1.4
2005 Age-adjusted mortality rate, cancer3 194.2 192.2 1.0
2005 Age-adjusted mortality rate, prostate cancer4 25.9 25.1 3.1
2005 Age-adjusted mortality rate, breast cancer5 25.8 25.4 1.6
2005 Age-adjusted mortality rate, unintentional
motor vehicle injuries3 18.7 18.8 -0.5
2005 Age-adjusted mortality rate, suicide3 11.4 11.4 0.0
1 Rate per 1,000 women ages 15-19
2 Rate per 1,000 population
3 Rate per 100,000 population
4 Rate per 100,000 male population
5 Rate per 100,000 female population
5
Statistical Primer No. 17 Division of Public Health
N.C. Department of Health and Human Services State Center for Health Statistics
Conclusion
It is not known how long NCHS will continue to
produce the bridged population data. The NCHS
bridged files were originally created to bridge the gap
between the old vital statistics race/ethnicity standards,
where a single race was reported on vital records, and
the 1997 federal OMB standards used by the 2000
Census. Once national vital statistics data are
converted to the new OMB race/ethnicity standards,
which involve reporting multiple races, the creation of
these files could be deemed unnecessary and NCHS
would no longer generate these files. Even if the files
are created in the future, the single-race population
estimates would not be consistent with the multiple-race
data collected on vital records.
While the State Demographer’s Office does update and
revise their annual population estimates periodically,
the State Center for Health Statistics maintains static
or “closed” population files based on the July 1
certified estimates. These files are updated by the State
Center for Health Statistics only when smoothed files
are generated after every Census. We do not update
our files with multiple revisions so that our calculated
rates are as consistent as possible with data published
in our annual Vital Statistics reports. However, this
also means that our closed population files will not
always contain the most up-to-date population data.
As stated early in this report, population figures are
produced by estimation and projection methods. This is
not an exact science, and published population figures
are, of necessity, simply educated estimates of an ever-changing
population landscape. There may also be some
degree of error in the numerators of rates calculated by
the State Center for Health Statistics, even when there is
a complete count. Misclassification of race on vital
records, small numbers of events for some geographic
areas and population groups, and other factors may
result in errors in counting health events. So rates and
other health measures should be considered to have a
margin of error rather than being precise calculations.4
References
1. North Carolina State Demographics Unit. Certified
July 1 North Carolina Resident Population
Estimates 2001-2005. http://demog.state.nc.us/
frame_start.html
2. National Center for Health Statistics. Estimates of
the July 1, 2000-July 1, 2005, United States resident
population from the Vintage 2005 postcensal series
by year, county, age, sex, race, and Hispanic origin,
prepared under a collaborative arrangement with the
U.S. Census Bureau. http://www.cdc.gov/nchs/
about/major/dvs/popbridge/popbridge.htm. August
16, 2006.
3. Buescher PA, Gizlice Z, Jones-Vessey KA.
Discrepancies between published data on racial
classification and self-reported race: evidence from
2002 North Carolina live birth records. Public
Health Reports 2005; 120: 393-398.
4. Buescher PA. Problems with rates based on small
numbers. Statistical Primer, No. 13. State Center
for Health Statistics, April 1997. http://
www.schs.state.nc.us/SCHS/pdf/primer12.pdf.
Table 3: 2002-2005 Estimates of the 2001-2005 N.C. Resident Population from the NCHS
Bridged Population Files
2001 2002 2003 2004 2005
Population Population Population Population Population
Estimate Estimate Estimate Estimate Estimate
2002 Bridged Population File 8,206,105 8,320,146 n/a n/a n/a
2003 Bridged Population File 8,195,249 8,305,820 8,407,248 n/a n/a
2004 Bridged Population File 8,198,256 8,311,899 8,421,190 8,541,221 n/a
2005 Bridged Population File 8,198,279 8,312,755 8,422,375 8,540,468 8,683,242
6
Statistical Primer No. 17 Division of Public Health
N.C. Department of Health and Human Services State Center for Health Statistics
Department of Health and Human Services
State Center for Health Statistics
1908 Mail Service Center
Raleigh, N.C. 27699-1908
350 copies of this public document were printed at a cost of $130.09 or 4¢ per copy. 5/07
State of North Carolina
Michael F. Easley, Governor
Department of Health and Human Services
Carmen Hooker Odom, Secretary
www.ncdhhs.gov
Division of Public Health
Leah Devlin, D.D.S., M.P.H., State Health Director
Chronic Disease and Injury Section
Marcus Plescia, M.D., M.P.H., Section Chief
State Center for Health Statistics
Paul A. Buescher, Ph.D., Director
www.schs.state.nc.us/SCHS/
The N.C. Department of Health and Human Services does not discriminate on the basis of race, color,
national origin, sex, religion, age or disability in employment or the provision of services.