Health (NSCH) was fielded using the State and Local Area Integrated Telephone Survey (SLAITS) mechanism. SLAITS is conducted by the National Center for Health Statistics (NCHS), Centers for Disease Control and Prevention (CDC). It uses the same large-scale random-digit-dial sampling frame as the CDC’s National Immunization Survey.1
Approximately 2.8 million telephone numbers were randomly generated for inclusion in the NSCH. After eliminating numbers that were determined to be nonresidential or nonworking, the remaining numbers were called to identify households with children less than 18 years of age. From each household with children, one child was randomly selected to be the focus of the interview.
The respondent was the parent or guardian in the household who was knowledgeable about the health and health care of the randomly selected child. For 73.5 percent of the children, the respondent was the mother. Respondents for the remaining children were fathers (20.5 percent), grandparents (4.2 percent), or other relatives or guardians (1.8 percent).
Surveys were conducted in English, Spanish, Mandarin, Cantonese, Vietnamese, and Korean. Overall, 5.3 percent of the interviews were completed in Spanish, and 0.2 percent of the interviews were conducted in one of the four Asian languages.
Data collection began on April 5, 2007 and ended on July 27, 2008, with interviews conducted from telephone centers in Chicago, Illinois and Las Vegas, Nevada. A computer-assisted telephone interviewing system was used to collect the data. A total of 91,642 interviews were fully or partially completed for the NSCH, with 79 percent of the interviews completed in 2007. The number of completed interviews varied by state, ranging from 1,725 in Vermont to 1,932 in Illinois.
The interview completion rate, which is the proportion of interviews completed after a household was determined to include a child under age 18, was 66.0 percent. The overall response rate, which is the product of the resolution rate (the proportion of telephone numbers identified as residential or nonresidential), the screener completion rate (the proportion of households successfully screened for children), and the interview completion rate, was 51.2 percent. This rate is based on the assumption that telephone numbers that were busy or rang with no answer on all attempts were nonresidential.
Overall response rates ranged from 39.4 percent in New Jersey to 61.9 percent in North Dakota. Several efforts were made to increase response rates, including sending letters to households in advance to introduce the survey, toll-free numbers left on potential respondents’ answering machines to allow them to call back, and small monetary incentives for those households with children who initially declined to participate.
For producing the population-based estimates in this report, the data records for each interview were assigned a sampling weight. These weights are based on the probability of selection of each household telephone number within each State, with adjustments that compensate for households that have multiple telephone numbers, for households without telephones, and for nonresponse. With data from the U.S. Bureau of the Census, the weights were also adjusted by age, sex, race, ethnicity, household size, and educational attainment of the most educated household member to provide a dataset that was more representative of each State’s population of noninstitutionalized children less than 18 years of age. Analyses were conducted using statistical software that accounts for the weights and the complex survey design. Responses of “don’t know” and “refuse to answer” were considered to be missing data. Records with missing data on the variables of interest were excluded from all analyses, with one exception. For households with missing data for income or household size, the household income relative to the federal poverty level was multiply imputed.
Children’s areas of residence were classified according to the Rural- Urban Commuting Areas (RUCAs).2 The RUCA codes were developed by the U.S. Department of Agriculture's Economic Research Service and the University of Washington's Rural Health Research Center through funding provided by the Federal Office of Rural Health Policy. The 10 RUCA codes were grouped into three categories. “Urban-focused areas” (RUCA codes 1.0, 1.1, 2.0, 2.1, 3.0, 4.1, 5.1, 7.1, 8.1, and 10.1) include metropolitan areas and surrounding towns from which commuters flow to an urban area; large rural areas (RUCA codes, 4.0, 4.2, 5.0, 5.2, 6.0, and 6.1) include large towns (“micropolitan” areas) with populations of 10,000 to 49,999 and their surrounding areas; and small or isolated rural areas (all remaining codes) include small towns with populations of 2,500 to 9,999 and their surrounding areas. Children were classified by race and ethnicity in seven categories: non-Hispanic White, non-Hispanic Black, non-Hispanic American Indian/Alaska Native, other single races, other combined races, Hispanic (English speaking) and Hispanic (Spanish speaking). Racial and ethnic groups are mutually exclusive; that is, data reported for White, Black, American Indian/Alaska Native, multiracial, and children of other races do not include Hispanics, who may be of any race. These categories differ from the racial aggregation method recommended by the Office of Management and Budget, which keeps intact the five single-race categories and includes the four double-race categories that are most frequently reported. This analysis did not employ these nine groups because sample sizes did not support it. However, a separate category was included for American Indian/Alaska Natives, as well as those of other races, because their health risks may vary by locality.
The data from the NSCH are subject to the usual variability associated with sample surveys. Small differences between survey estimates may be due to random survey error and not to true differences among children or across States.
The precision of the survey estimates is based on the sample size and the measure of interest. Estimates at the national level will be more precise than estimates at the urban/rural level, and those for all children will be more precise than estimates for subgroups of children (for example, children in small rural areas or children of the same race). For national estimates of the health and health care of all children, the maximum margin of error is 0.8 percentage points.3 For estimates reported by area of residence for all children, the maximum margin of error is 3.8 percentage points.
Except for data suppressed to protect the confidentiality of the survey subjects, all data collected in the NSCH are available to the public on the NCHS and MCHB websites. Data documentation and additional details on the methodology4 are available from the NCHS .
Interactive data queries are possible through the Data Resource Center for the NSCH . The Data Resource Center provides immediate access to the survey data, as well as resources and assistance for interpreting and reporting findings.
1 Zell ER, Ezzati-Rice TM, Battaglia MP, Wright RA. National immunization survey: The methodology of a vaccination surveillance system. Public Health Reports 115:65-77. 2000.
2 USDA Economic Research Service and the WWAMI Rural Health Research Center. Rural-Urban Commuting Area Codes. http://www.ers.usda.gov/Data/RuralUrbanCommutingAreaCodes
3 U.S. Department of Health and Human Services, Health Resources and Services Administration, Maternal and Child Health Bureau. The Health and Well-Being of Children: A Portrait of States and the Nation 2007. Rockville, Maryland: U.S. Department of Health and Human Services 2009.
4 Blumberg SJ, Foster EB, Frasier AM, et al. Design and Operation of the National Survey of Children’s Health, 2007. National Center for Health Statistics. Vital Health Stat 1. 2009.