The data displayed in these graphs come from daily or weekly tracking surveys conducted by YouGov.
For each survey, a representative sample of 1,500 respondents is selected from YouGov’s U.S. research panel. Panelists are invited to each survey, based upon their age, gender, race, and education, in proportion to their frequency to the frequency of adult citizens in the most recent American Community Survey.
Responding panelists are then weighted according to their demographics, voter registration status, and 2016 Presidential vote. The margin of error for adults in each survey is approximately 4% and for registered voters is approximately 5%. Trend lines are computed using a LOESS smoother (Cleveland and Devlin, Journal of the American Statistical Association, 1988) to reduce random fluctuations in the data due to sampling variability.
Each day YouGov asks topical questions to get the public’s first judgement on the biggest issues of the day. These launch each weekday morning, and the results are released each afternoon.
It’s possible to respond to these questions in multiple places, including the YouGov website and app. Once you have responded you can also view the live, unweighted results collected so far.
However, these numbers are just an early indication, and it’s important to be cautious with them. Unlike a traditional poll they are not yet weighted to be representative of the full population – an adjustment that we do to ensure that the results are an accurate depiction of the views of all US Adults.
It also means that, because anybody can take part in the poll through the YouGov website even if they are not regular panellists, people with a particular interest in a topic may be more likely to take part, skewing the results.
The final results will be published into the survey results archive each day, send out via the YouGov Daily email newsletter, and also tweeted from the YouGov's Twitter account. Unlike the live results, these results are weighted to be representative of the United States population by:
For the final nationally representative results, we only include responses from people who answered the questions as part of a traditional YouGov survey, and not those who are just responding through the mobile app or the YouGov website. Our testing has shown that this methodology produces the most accurate results that are in line with standard YouGov polling.