Blue Wall Voices Project
The Blue Wall Voices Project was designed and analyzed by public opinion researchers at the Kaiser Family Foundation (KFF) in collaboration with the Cook Political Report. The survey was conducted September 23rd – October 15th, 2019, among a representative random sample of 3,222 registered voters in four states (767 in Michigan, 958 in Minnesota, 752 in Pennsylvania, and 745 in Wisconsin) constituting the “Democratic Blue Wall” – the area in the Upper Midwest that were previously considered Democratic strongholds, and where state polls performed poorly in 2016 and underestimated support for President Trump. All registered voters included in the sample were sent an invitation letter including a link to complete the survey online, a toll-free number that respondents could call to complete the survey with a telephone interviewer, and $2 pre-incentive. Respondents who were living in Census block-groups identified as low-education and respondents identified as likely Hispanic were offered $10 post-incentive if they completed the survey. A random half of respondents received a QR code on their invitation letters. All respondents were then sent a reminder postcard, which included a QR code for all respondents. Respondents who were flagged in the voter file as both a) speaking Spanish and b) speaking either primarily or only their native language, received bilingual mailings, including text in both English and Spanish.
The sample was designed to reach respondents less likely to complete surveys online, by oversampling areas with a relatively low percentage of college graduates. Sample that could be matched to telephone numbers and that had not yet completed the survey online or by inbound computer-assisted telephone interview (CATI) were called by CATI interviewers to attempt to convert this sample to completed interviews. A total of 2,763 respondents completed the questionnaire online, 255 by calling in to complete, and 204 were completed as outbound CATI interviews. Data collection was carried out in English and Spanish by SSRS of Glen Mills, PA. The registered voter sample was provided by Aristotle. KFF paid for all costs associated with the survey.
A series of data quality checks were run on the final data, which resulted in 40 completes being removed from the data. Weighting involved multiple stages: First, each state sample was weighted to account for the sampling methodology including the oversampling of low-educational attainment areas and to the proportions of the voter file reachable or unreachable by outbound phone-call. Second, each state’s sample was weighted to match the voter file distribution of 2016 voter status, party identification, gender, race/ethnicity, home-ownership, Metropolitan status, state region, as well as the 2018 CPS Voter Supplement for age-by-gender, educational attainment and race. To address non-response among partisans not accounted for by demographics, the weight was adjusted so that the self-reported 2016 vote in each state matched the actual state outcome. The final weight combined each state’s weight and balanced the combined total sample to state distributions. All statistical tests of significance account for the effect of weighting.
The margin of sampling error including the design effect for the full sample is plus or minus 2 percentage points. Numbers of respondents and margins of sampling error for each state are shown in the table below. For results based on other subgroups, the margin of sampling error may be higher. Sample sizes and margins of sampling error for other subgroups are available by request. Note that sampling error is only one of many potential sources of error in this or any other public opinion poll. Kaiser Family Foundation public opinion and survey research is a charter member of the Transparency Initiative of the American Association for Public Opinion Research.