Methods & Data
You can access the data from the Baylor Religion Surveys waves 1-3 here:
WAVE 5 METHODS:
The 2017 administration of the Values and Beliefs of the American Public survey was fielded using a self-administered pen and paper methodology with mail based collection only. Gallup mailed out an initial 11,000 surveys with an invitation letter, return envelope, and $1 USD cash incentive on February 2nd. Reminder postcards were mailed to all on February 13th, and a full cover letter, survey, and return envelope package was sent to those addresses remaining open on February 28th. Collection of completed interviews finished on March 21st.
Sample Design and Selection
The sample for this study was selected using ABS (Address Based Sample) methodology based on a simple stratified sample design. The mode of data collection was mail. Gallup obtained the ABS sample from the Marketing Systems Group (MSG), a database and survey sample vendor, who has access to the latest DSF (Delivery Sequence File) frame and generated the Sample for Gallup. The DSF of USPS (United States Postal Service) is a computerized database that contains all delivery point addresses, with the exception of general delivery where carrier route or P.O. Box delivery is not available and mail is held at a main post office for claim by recipients. The choice ABS sampling method was made to address the evolving coverage problems associated with telephone-based samples. Gallup obtained a total sample of 11,000 addresses nationwide from MSG and a total of 1,501 surveys were completed with an overall response rate of 13.6%.
The target population (of all adults nationwide) was stratified into 12 strata described Table 1 below. As shown in Table 1, strata were formed based on density of specific subgroups (like Hispanic, African, Younger population (18-34)) to ensure minimum coverage of these sub-populations in the sample. The determination of high density Hispanic, African American or younger population was based on information available at the census block group level. Sampling was done independently within each stratum.
Weighting of Sample data
Sample data were weighted to minimize bias in the survey based estimates. The base weight assigned to each respondent in each stratum was equal to the inverse of the probability of selection (or the sampling fraction) for that stratum. The base weights were then adjusted for non-response by a non-response weight adjustment factor equal to the ratio of the sample size and the number of completed surveys in each stratum. The final step involved post-stratification weighting to restore proportionality among groups of the population that may have been overrepresented or underrepresented in the survey due to differential non-response or representation on the sample frame. In the process of post-stratification weighting, Gallup weighted the actual respondent database to match the known demographic characteristics of the U.S. adult population by geographic region (census region), age, gender, race/ethnicity, and education. Trimming of extreme weights were carried out to minimize the effect of large weights on sampling error. The target data for post-stratification weighting were obtained from the latest Current Population Survey (CPS) 2015 population projections.
Sample Comparisons
Wave 5 of the Baylor Religion Survey (BRS5, 2017) has a final sample of 1,501 respondents, a sample that compares favorably with the 2016 sample of the General Social Survey on a number of key demographic, religious and political characteristics. The mean age of the BRS sample was 48.8, compared to 47.6 on the GSS. A slight majority of both samples were female, comprising 52.0 and 54.8 percent of the BRS and GSS, respectively. In terms of education, the BRS had slightly fewer respondents with less than 8 years of schooling than the GSS (1.4 and 4.1 percent, respectively), though shares of the samples with at least a college degree were more similar (32.7 and 31.2 percent, respectively). Slightly fewer respondents in the BRS were single and never married than in the GSS (21.1 and 27.4 percent, respectively), but about half of both samples were currently married. In terms of religious participation, a slightly higher share of BRS respondents never attend religious services compared to GSS respondents (28.0 and 25.0 percent, respectively) while similar proportions attend about once a week or more often (29.9 and 29.1 percent, respectively). Both the BRS and GSS samples included about 37 percent of respondents who reported to be political moderates. The BRS had a somewhat larger share of both conservatives (21.7% vs. 15.8%) and liberals (15.8 vs. 12.4%) than the GSS while having slightly lower shares of respondents among both those who leaned in either direction or considered themselves extreme in either direction.