The survey was designed and conducted by PRRI among a total sample of 22,612 adults (age 18 and up) living in all 50 states in the United States, 20,670 of whom are part of Ipsos’s Knowledge Panel. An additional 1,596 were recruited by Ipsos using opt-in survey panels to increase the sample sizes in smaller states. An additional 346 Hispanic Protestants were also recruited using opt-in survey panels. All oversamples are weighted into the full population. Interviews were conducted online March 8-30, June 7-23, August 9-30, and October 18-November 9, 2021. Each of the four waves is weighted separately, but a total weight is used for reporting the combined dataset.
Demographic information and religious affiliations are from a separate pull of 40,000 records from KnowledgePanel, sampled and weighted to be representative of the U.S. population.
Respondents are recruited to the KnowledgePanel using an addressed-based sampling methodology from the Delivery Sequence File of the USPS ‐ a database with full coverage of all delivery addresses in the U.S. As such, it covers all households regardless of their phone status, providing a representative online sample. Unlike opt-in panels, households are not permitted to “self-select” into the panel; and are generally limited to how many surveys they can take within a given time period.
The initial sample drawn from the KnowledgePanel was adjusted using pre-stratification weights so that it approximates the adult U.S. population defined by the 2019 American Community Survey (ACS). Next, a probability proportional to size (PPS) sampling scheme was used to select a representative sample.
To reduce the effects of any non-response bias, a post-stratification adjustment was applied based on demographic distributions from the ACS. The post-stratification weight rebalanced the sample based on the following benchmarks: age, race and ethnicity, gender, Census division, metro area, education, and income. The sample weighting was accomplished using an iterative proportional fitting (IFP) process that simultaneously balances the distributions of all variables. Weights were trimmed to prevent individual interviews from having too much influence on the final results. In addition to an overall national weight, separate weights were computed for each state to ensure that the demographic characteristics of the sample closely approximate the demographic characteristics of the target populations. The state-level post-stratification weights rebalanced the sample based on the following benchmarks: age, race and ethnicity, gender, education, and income.
These weights from the KnowledgePanel cases were then used as the benchmarks for the additional opt-in sample in a process called “calibration.” This calibration process is used to correct for inherent biases associated with nonprobability opt-in panels. The calibration methodology aims to realign respondents from nonprobability samples with respect to a multidimensional set of measures to improve their representation.
The margin of error for the national survey is +/- 0.8 percentage points at the 95% level of confidence, including the design effect for the survey of 1.7. In addition to sampling error, surveys may also be subject to error or bias due to question wording, context, and order effects. Additional details about the KnowledgePanel can be found on the Ipsos website: https://www.ipsos.com/en-us/solution/knowledgepanel