Woven throughout each individual FRAME are statistics and data-driven statements based on research collected by Barna Group as a part of the FRAMES project. This research provided the foundational work for producing FRAMES.
During 2013, Barna Group conducted four national studies, interviewing 4,495 adults ages 18 or older. The surveys were quantitative in nature and conducted using online and telephone, including cell phone sampling. To ensure a nationally representative study, minimal statistical weighting was applied to create a sample that reflects known national demographic averages based upon U.S. Census data sources.
For the online portions of the FRAMES research, Barna Group used a research panel known as KnowledgePanel®, created by Knowledge Networks. This panel is based on probability sampling rather than volunteers. The sampling method used by KnowledgePanel® covers 97% of U.S. households, including online and offline households, as well as cell phone only households. Recruited households that do not have Internet access are provided a netbook computer and free Internet service so they may participate as online panel members.
The phone interviews took place at various times throughout the day. If the individual selected to participate in the survey was unavailable on the first call attempt, up to five additional callback attempts were made. This effort maximized the possibility of contacting the selected individual, which in turn increases the reliability of the survey results. Most of the phone interviews took place by dialing respondent’s home telephone. Thirty percent of the respondents were contacted by dialing their mobile number. All of the phone interviews were conducted with a live interviewer. To ensure proper skip patterns and accuracy in recording data, a CATI system (Computer Assisted Telephone Interviewing) was used.
The accuracy of a survey is noted in its sampling error. The sample error number refers to the degree to which a survey’s results may be inaccurate due to interviewing a sample rather than the whole population. Sample size and the degree to which the result is close to 50 percent or the extremes, 0 percent and 100 percent, affect the sampling error estimate. The table above provides the maximum sample error range for each study.
While reading research results it is important to bear in mind that there are several other factors, beyond sampling error, that may affect the accuracy of a study, for example, biased question wording, question sequencing, inaccurate recording of the responses provided, inaccurate data tabulation and so on. These factors cannot be statistically estimated