KFF COVID-19 Vaccine Monitor- Rural America

The KFF COVID-19 Vaccine Monitor – Rural America was designed and analyzed by public opinion researchers at the Kaiser Family Foundation (KFF). The survey was conducted March 15-29, 2021 via telephone and online among a nationally representative sample of 1,001 adults residing in rural counties (including interviews from 159 Hispanic adults and 170 non-Hispanic Black adults). For the telephone components, respondents were reached through randomly generated telephone numbers from cell phone and landline sampling frames associated with rural counties, with an overlapping frame design, and disproportionate stratification. Stratification was based on incidence of the race/ethnicity subgroups within each frame. Specifically, the cell phone frame was stratified as: (1) High Hispanic: Cell phone numbers associated with rate centers from counties where at least 35% of the population is Hispanic; (2) High Black: Cell phone numbers associated with remaining rate centers from counties where at least 35% of the population is non-Hispanic Black; (3) Else: numbers from all remaining rate centers. The landline frame was stratified as: (1) High Black: landline exchanges associated with Census block groups where at least 35% of the population is Black; (2) Else: all remaining landline exchanges. Rate centers and exchanges were considered likely rural if they were in a county that was not part of a metropolitan statistical area. Respondents’ rural residency was established by self-reported zip code or county of residence.

A total of 206 rural respondents were interviewed as part of the March KFF Vaccine Monitor (March 15-March 22), and 795 were part of a mixed-mode rural supplement from March 23 to March 29. The supplement used the same stratification plan, including only numbers in areas identified as micropolitan or noncore based on the CDC’s Urban-Rural Classification Scheme for Counties.  To reach a total minimum of 1,000 rural respondents, of whom 150 were Hispanic and 150 non-Hispanic Black, SSRS employed multiple approaches:  First, the total number of completed interviews in the High Hispanic and High Black strata were oversampled; meaning, if the respondent reached through the oversamples was neither Hispanic nor Black, the interview was terminated, and the respondent screened out. In total 31 Hispanic and 66 Non-Hispanic Black respondents were reached through oversampling. The landline sample included a small oversample of records in the frame that were matched to directory-records with a distinctively Hispanic surname. An additional 32 interviews were completed with respondents who had previously completed interviews on the KFF Health Tracking Poll six months ago or more and were called back for this month’s study. Finally, 35 interviews were completed with respondents who had previously completed an interview on the SSRS Omnibus poll (and other RDD polls) and identified as Hispanic (including 4 in Spanish) and 49 interviews were completed with respondents who had previously completed an interview on the on the SSRS Omnibus poll (and other RDD polls) and identified as non-Hispanic Black. SSRS Omnibus is a weekly RDD poll, employing an overlapping dual-frame design. In total, 206 respondents from the VM survey said they live in a zip code that matched the definition of rural counties, meaning their county was not in a metropolitan statistical area; 29 of these respondents were Hispanic, and 36 non-Hispanic Black.

In the course of the field period, SSRS also invited members of its probability-based online panel (SSRS Opinion Panel) to participate in the study. Invitees all self-reported living in rural zip codes. As with other sample components, Hispanic and Black respondents were oversampled.  The SSRS Opinion Panel is a nationally representative probability-based web panel. SSRS Probability Panel members are recruited randomly in one of two ways: (a) Through invitations mailed to respondents randomly sampled from an Address-Based Sample (ABS). ABS respondents are randomly sampled by MSG through the U.S. Postal Service’s Computerized Delivery Sequence (CDS). (b) from a dual-frame random digit dial (RDD) sample, through the SSRS Omnibus survey platform. Sample for the SSRS Omnibus is obtained through Marketing System Groups (MSG). In total 272 interviews in the rural supplement sample were completed via landline and 438 via cell phone, including 328 who could not be reached via landline. 291 respondents completed the survey online.

The combined landline, cell phone, and online rural samples were weighted to balance the sample demographics to match estimates for the national population using data from the Census Bureau’s 2019 U.S. American Community Survey (ACS), on sex, age, education, race, Hispanic origin, and region, within race-groups, along with data from the 2010 Census on population density. The sample was also weighted to match current patterns of telephone use using data from the January- June 2020 National Health Interview Survey. The rural sample was also weighted to match aggregate county level demographics for the rural counties. Population parameters were derived using Census-based estimates provided by Nielsen Pop-Facts through Marketing Systems Group based on data from data Census Bureau’s 2019 American Community Survey (ACS). Weighting parameters included race and Hispanic origin, race by gender, educational attainment, age, Census region and micropolitan status. Data were also adjusted to match internet-use estimates for rural areas based on ACS data. The weight takes into account the fact that respondents with both a landline and cell phone have a higher probability of selection in the combined sample and also adjusts for the household size for the landline sample, and design modifications, namely, the oversampling of prepaid cell phones and likelihood of non-response for the re-contacted sample. All statistical tests of significance account for the effect of weighting.

The margin of sampling error including the design effect for the full sample is plus or minus 3 percentage points. Numbers of respondents and margins of sampling error for key subgroups are shown in the table below. For results based on other subgroups, the margin of sampling error may be higher. Sample sizes and margins of sampling error for other subgroups are available by request. Note that sampling error is only one of many potential sources of error in this or any other public opinion poll. Kaiser Family Foundation public opinion and survey research is a charter member of the Transparency Initiative of the American Association for Public Opinion Research.

This work was supported in part by grants from the Missouri Foundation for Health, the Chan Zuckerberg Initiative DAF (an advised fund of Silicon Valley Community Foundation), the Ford Foundation, and the Molina Family Foundation. We value our funders. KFF maintains full editorial control over all of its policy analysis, polling, and journalism activities.