This report is based on results from two surveys – a national telephone survey of more than 35,000 adults that was the centerpiece of the Pew Research Center’s 2014 U.S. Religious Landscape Study, and a 2015 “recontact” survey in which 5,000 Landscape Study respondents were recontacted and asked additional questions. Full methodological details about the 2014 survey are included in Pew Research Center’s report “America’s Changing Religious Landscape.” The remainder of this section provides details about how the recontact survey was conducted.
Interviewing for the recontact survey was done under the direction of Abt SRBI March 17 to May 6, 2015. Interviews were conducted in English and Spanish; 2,013 respondents were interviewed on a landline telephone and 2,987 were interviewed on a cellphone.
Respondents who participated in the 2014 Religious Landscape Study’s original telephone survey were sorted into 15 recontact strata based on a variety of religious and demographic characteristics. In order to support the analytical goals of the project, certain groups (e.g., religiously intermarried respondents) were oversampled, while others were undersampled (see tables). After data collection, the data were weighted (as described below) such that all 15 strata are represented in roughly their proper proportion in the study’s final, weighted estimates.
The national results included in this report are based on weighted estimates. The weighting was conducted in two stages. The first stage in the weighting process was the calculation of base weights, which was accomplished by multiplying the final base weight from the 2014 Religious Landscape Study’s original telephone survey by the inverse of the subsampling rate in the recontact survey. This step was carried out separately for each of the 15 recontact strata. For details on how the base weights were calculated in the 2014 Religious Landscape Study, see the methodological appendix in “America’s Changing Religious Landscape.”
The second stage of the weighting calibrated the base-weighted data to demographic benchmarks for the population covered by the survey (U.S. adults). This was performed via iterative proportional fitting (or “raking”). The raking procedure aligned survey respondents to population benchmarks on the following dimensions:
- Gender by age
- Gender by education level
- Education level by age
- Telephone service
- Population density
- State of residence
- Religious affiliation
The demographic benchmarks for age, gender, education level, race, and state came from an analysis of the Census Bureau’s 2013 American Community Survey (ACS) one-year estimates, which was the most current data source available at the time the 2014 Religious Landscape Study data were weighted. The ACS parameters were calculated for adults ages 18 and older residing in households in the U.S., excluding those living in institutionalized group quarters. The population density parameter was derived from Census 2010 data and filtered on adults ages 18 and older residing in the U.S. The benchmark for religious affiliation was based on weighted estimates of self-reported data from the 2014 Religious Landscape Study. The telephone service parameter was constructed from model-based estimates released by the National Center for Health Statistics for July-December 2013, projected to the year 2014. These estimates were based on adults, ages 18 and older, living in households with a telephone (either landline or cell). The completion rate for this study was 37.2%. Taken with the response rate to the original 2014 Religious Landscape Study, the cumulative response rate of the new survey is 3.9%.
Design effect and margins of error
Weighting and survey design features that depart from simple random sampling tend to result in a loss of precision in survey estimates. This loss of precision, known as the design effect, is incorporated in all margins of error, standard errors and tests of statistical significance included in this report.
Generally speaking, larger sample sizes are associated with smaller margins of sampling error, and smaller sample sizes are associated with larger margins of error. The margin of error for national estimates from this survey is +/- 2.0 percentage points. Many of the analyses in this report as based on subgroups of respondents (e.g., those who provided complete information about the religion of their parents, or on respondents with a particular kind of religious background), which have larger margins of error. The accompanying table presents sample sizes and corresponding margins of error for some of the groups commonly discussed in this report. Sample sizes and margins of error for other groups are available upon request.
In addition to sampling error, one should bear in mind that question wording and practical difficulties in conducting surveys can introduce error or bias into the findings of opinion polls.