Profiles Methods
Sampling
Profiles employs random, systematic, equal-probability sampling strategies to produce representative samples of schools that serve students in grades 6 through 12 in each jurisdiction. In most jurisdictions, the sampling frame consists of all regular secondary public schools with one or more of grades 6 through 12.
In 2022, 9 states and 26 school districts modified this sampling procedure by conducting a census of schools. That is, they invited all secondary schools in their jurisdiction, rather than just a sample, to participate in the surveys (Table 1).
Data Collection
For the 2022 Profiles cycle, all 44 states and 28 districts included in Profiles Explorer began data collection in sampled schools during the 2022 spring semester, but several jurisdictions continued data collection into fall 2022.
Note: Puerto Rico and the Cherokee Nation tribe also collected 2022 Profiles data but did not give permission to have their data included in Profiles Explorer.
For each middle or high school that was sampled, the principal and the lead health education teacher (the person most knowledgeable about health education at the school) each completed a standard, self-administered questionnaire. State, local, tribal, and territorial agencies had the option of conducting their survey using paper-and-pencil questionnaires or by using a secure Web-based system.
Among the sites included in Profiles Explorer, 13 states and 10 school districts used a paper-and-pencil survey administration. In these sites, the principal and lead health education teacher questionnaire booklets were mailed by the state or local education or health agency to the principal, who then designated the school’s lead health education teacher to complete the teacher questionnaire. Participation in the survey was confidential and voluntary; follow-up telephone calls, emails, and written reminders were used to encourage participation. The principal and teacher recorded their responses in the computer-scannable questionnaire booklets and returned them directly to the state or local education or health agency.
In 2022, 31 states and 18 school districts conducted Profiles using Web-based systems that contained the same questions as the computer-scannable questionnaire booklets. In these sites, principals were notified by the agency conducting the survey about Profiles and were provided with directions about how to access the Web-based principal questionnaire. They also were asked to designate the school’s lead health education teacher to complete the Web-based teacher questionnaire. These teachers were then provided with directions about how to access the Web-based teacher questionnaire.
Respondents who had difficulty with the Web-based system or who did not want to use it were offered paper questionnaires. Responses to these paper questionnaires were then entered into the Web-based system by the agency.
Data collected via Web-based systems were processed using the same procedures as those used for the data collected via computer-scannable booklets.
Data from sites with response rates of 70% or greater (separately for the principal and teacher surveys) were weighted to reflect the likelihood of principals or teachers being selected and to adjust for differing patterns of nonresponse based on school size (large, medium, small) and school level (middle school, other). This weighting process produced data representative of all public secondary schools in each jurisdiction.
For sites with response rates below 70% for either the principal or teacher survey or both, CDC conducted nonresponse bias analyses. These analyses examined whether responding schools differed from nonresponding schools on the following variables available from the sampling frames: school enrollment size (small, medium, large), grade level (middle school, junior/senior high school, high school), and school type (if the site included schools besides regular public schools in their sampling frame [e.g., charter schools]).
To obtain auxiliary variables that could be used to compare responding and nonresponding schools, school addresses were geocoded to Census Tract. Geocoded coordinates were then used to determine the school’s locale type from the National Center for Education Statistics (city, suburb, town, rural).
Geocoding allowed demographic and socioeconomic variables from the 2017-2021 American Community Survey (ACS) 5-Year Estimates data file to be merged onto the sampling frame.
The following variables from ACS were examined for each site:
- Population density
- Percent of population in each race/ethnic group (white, Black/African American, American Indian/ Alaska Native, Asian, Native Hawaiian/Other Pacific Islander, Hispanic, some other race, two or more races)
- Percent of population below poverty level
- Percent of youth ages 5-17 below 100% poverty level
- Percent of population with minimum of high school education
- Percent of population with a college degree or more
- Percent of households without a computer, tablet, or smartphone
- Percent of households with Internet access
- Percent of households with broadband Internet access
- Average median household income
- Percent owner-occupied housing units
- Percent of households on public assistance
- Percent of population with Medicaid coverage
- Percent unemployment
If a site’s survey (principal or teacher) had few significant differences between responding and nonresponding schools for the frame and auxiliary variables, data from that survey were weighted to be representative of public secondary schools in the jurisdiction. Only sites with representative data are included in Profiles Explorer. SAS software was used to compute the weighted estimates and confidence intervals provided in Profiles Explorer.
To calculate these estimates, data from regular public secondary schools were combined across all participating states plus the District of Columbia, including those without representative data, and state-level final weights were aggregated.
To calculate the final state-level weights, the sampling base weights were adjusted for non-response and the non-response-adjusted weights were scaled to the state population size of all regular public secondary schools based on school size and school level.