Voter in 2016 Survey Methodology

The Joint Center for Political and Economic Studies commissioned and analyzed the results of this survey conducted in partnership with the Nielsen Scarborough Company. This survey results from a nationally representative sample of 1,500 registered voters regardless of vote intention, with an intentional oversample of African Americans and Latinos (600 whites, 600 African Americans, and 300 Latinos). Though the report focuses upon sentiments and views expressed by African American voters, efforts were made to poll Latino and white voters for comparison.

The Joint Center acknowledges the absence of Asian American, Native Hawaiian, Pacific Islander, multiracial, and Native American data. Unfortunately, limited in-language resources prevented the Joint Center and Nielsen Scarborough from polling these voters.

Using survey questions developed by the Joint Center, the Nielsen Scarborough Company collected this data through the Nielsen Scarborough panel between September 1 and September 15, 2016. The Nielsen Scarborough panel consists of 200,000+ U.S. adults drawn from a random probability selection process that includes random-digit-dialing (RDD) and address-based sample methods. The panel offers statistically reliable projections to the total U.S. adult population and is designed to ensure the representativeness of Hispanic and African American populations.

Due to the rounding of responses to the nearest percentage point, not all the data in tables and graphs add up to 100 with variations including 99 or 101.  Groups broken down by political candidate reflect only those individuals within that demographic group who expressed support for that candidate. Questions broken down by racial subgroup without reference to candidate support reflect all respondents of that group, regardless of vote intention or candidate preference.

Respondents were asked to self-report race, and whether they were Latino or of Hispanic origin (which the Joint Center uses interchangeably in this report to be more inclusive of those who identify as being of South or Latin American descent).^[1] For results that compare racial and ethnic groups, non-Hispanic whites and non-Hispanic blacks were compared with Latinos overall. For breakdowns of black respondents we include only non-Hispanic blacks.

The margin of error for individual racial and ethnic groups is +/- 5 percentage points. In comparing differences within racial and ethnic subgroups, the margin of error grows larger. The Joint Center cautions readers to interpret group differences with care. It is also important to note that margins of error are calculated on individual proportions and not on the difference. The margin of error also shrinks significantly as a number approaches zero, allowing us to be more confident in some results than others.

While the Joint Center does not assume the overall support levels are identical to their current state, the survey’s results were generally similar to other national polling, with non-Hispanic white voters more likely to support Trump, and Latino and African American voters more likely to support Clinton than whites. This survey is not a snapshot of the horserace between two candidates just before the national election. Instead, with significant oversamples of African Americans and Latinos, the survey provides insight into trends in differences of opinion among different African American demographic groups (age, gender, income, educational attainment, geographic region).

Among black respondents included in the poll, the weighted sample reflects the following demographics:

The large gender gap among African American respondents reflects the fact that the registered voter pool for all black voters is more female than male. This is due in part to high incarceration rates and felon disenfranchisement for black men.