The Landmark Spirituality and Health Survey is a Nationwide Face-to-Face Survey of 3,000 respondents with this targeted age breakdown: 1,000 aged 18-40; 1,000 aged 41-64; and 1,000 aged 65 and older. This study includes a 70-minute interview in English or Spanish with some basic biomarker data collection. NORC will collect 3000 face-to-face interviews using a representative sample of housing units from a subset of the NORC 2010 National Sampling Frame. This frame has been updated based on the latest Decennial Census, and is being used for two of NORC’s flagship face-to-face data collection projects: the General Social Survey and the Survey of Consumer Finances.
Within each household, up to two adults are selected at random. This results in an average of 1.59 eligible respondents per household, based on American Community Survey (ACS) data. To achieve 3,000 interviews regardless of age, a total of 5,132 housing units are needed with assumed interview and screener response rates as well as housing unit and language eligibility rates.
While 33.3 percent of the interviews need to be with adults aged 65 and older, only 17.2 percent of the U.S. population is aged 65 and older. The ACS does show that the most difficult to interview respondents are younger, however, so we are more likely to see an age distribution like the ACS than the U.S. population. Based on the ACS, we expect approximately 21.9 percent of respondents to be 65 years and older, 34.5 percent to be 18-40 years old and 43.6% to be 41-64 years old. Therefore, to achieve 1,000 respondents who are 65 years old and older, we need to increase the number of housing units selected and subsample the selected respondents who are younger than 65. These numbers are summarized in Table 2 below.
To reduce the total number of housing units selected in order to achieve 1,000 respondents 65 years and older and also to increase the subsampling rates for younger age groups, we are currently attempting to oversample areas based on the percentage of adults aged 65 and older, so the sample size may be less than 7,823 housing units and the selection rates may be higher than shown in Table 2.