The African American Research Collaborative completed 1215 interviews with African Americans with roughly 200 interviews per state in 3 Rust Belt states (Michigan, Pennsylvania and Wisconsin) and 3 Southeastern states (Florida, Georgia and North Carolina).  An oversample of 18-29 year olds (half of all respondents) was used to learn more about this group which has lower turnout rates and less commitment to the Democratic party than other age groups of Black voters. The sample starts with a larger/inclusive frame, and then all respondents self-report race or ethnicity at the outset and those who are African American continue with the full survey.  The survey was completed 100% online. Resulting data was weighted to match the most recent Census estimates of the African American population with weights included for age, gender, education and region. The poll was in the field from July 1 to July 9, 2020 and has a margin of error of ± 2.8% (full sample) and ± 6.9% (each battleground state).

The African American Research Collaborative completed 604 interviews with African Americans across the United States.  The sample starts with a larger/inclusive frame, and then all respondents self-report race or ethnicity at the outset and those who are African American continue with full survey.  The survey was completed 75% online and 25% via live phones (to ensure that people without internet access, a potential barrier due to the pandemic were interviewed).

Resulting data was weighted to match the most recent Census estimates of the African American population with weights included for age, gender, education and region.

​

The poll was in the field from May 1 to May 7, 2020 and has a margin of error of +/- 3.9%.

In partnership with Asian American Decisions and Latino Decisions, the African American Research Collaborative completed 9,400 interviews with black, Latino, AAPI, Native, and white registered voters who had already voted early, or were certain to vote in the November 6, 2018 general election.

​

Interviews were conducted through a combination of cell phone and landline telephone with live callers and self-completed online. Latino and AAPI voters were given an opportunity to complete their interview in their language of choice at the start of the interview. Respondents were randomly selected from a statewide, or districtwide sample frame, giving all voters an equal opportunity to be selected for interviews. Respondents were reached on landline and cell phone-only households, from October 31 to November 5, 2018 and interviews averaged 14 minutes in length.

​

Voters were pre-screened based on their vote history in previous midterm elections, and date of registration to include a mix of new registrants and first-time voters as well as those who had confirmed vote history in 2010 and 2014 midterms, or newer voters who first voted in 2016 and even newer registrants who were first-time voters in 2018. Respondents were asked if they had already voted early, and if not, if they were 100% certain they would vote on November 6th, and approximately half of voters indicated they had already voted early, while half were Election Day precinct voters. The interview for any respondent who was not certain was terminated. Using this same methodology in 2014, 92% of the interviewed sample was subsequently confirmed (validated) as having voted when examining official vote records, with no meaningful deviation from reported totals.

​

For individual states, a minimum of 400 interviews were completed to provide state-specific reliable estimates and carry an overall margin of error of 4.9% for each racial group. For the 70-district congressional district battleground sample, 600 total interviews were conducted per each racial group for a total of 2,400 interviews. Native American voters were selected from a nationwide sample frame and include 600 total interviews. Final data were compared to known census estimates for demographic profiles of voters and weighted to be representative within each state or congressional district.

2,045 registered voters were interviewed online in the 61 congressional districts deemed most competitive by Cook, CNN and Crystal Ball, and the overall poll has a margin of error of +/- 2.1%. A minimum of 400 registered voters were interviewed from each racial and ethnic group: Latinos, African Americans, Asian Americans or Pacific Islanders, Native Americans, and white non-Hispanics, which each carry a margin of error of +/- 4.8%. Due to sample size considerations, Native Americans were drawn from a 50-state national sample, while the other four groups were drawn only from the 61 competitive districts. The overall data are weighted to the most recent Census estimates for the registered voter population for age, gender, education, and nativity, and then weighted proportionately to match the overall racial demographics of the combined 61 districts which is 68% white, 15% black, 12% Latino, 4% Asian, 1% Native American. Respondents were randomly selected for interviews through online panels, and verified to be registered to vote, and to live in one of the 61 competitive house districts, in order to participate in the survey. The survey was in the field July 5-July 14, 2018.