Morning Consult conducted surveys with 1,198,588 registered U.S. voters from Jan. 20, 2017 to May 31, 2018 to determine the approval ratings for President Donald Trump in each of the 50 states and Washington, D.C., for each month.
In each poll, Americans indicated whether they approve or disapprove of the job performance of Trump. For each question, they could answer strongly approve, somewhat approve, somewhat disapprove, strongly disapprove, or don’t know/no opinion. The surveys also included about 30 demographic questions.
The results use a statistical technique called multilevel regression and post-stratification (MRP) to estimate state-level public opinion from the national survey data. MRP has been widely used in industry and in academia, and MRP estimates of state- and congressional district-level public opinion have generally been shown to outperform national polling, especially when there are few respondents in smaller geographic areas.
Responses to the Trump approval question are modeled via multilevel regression as a function of both individual level and state-level variables. Our models use age, gender, education and race as individual-level predictor variables. For our state-level variables, we chose variables that may influence state-level vote choice such as the percent change in state gross domestic product (GDP), state unemployment rates, state median household income and state-level outcomes from the 2016 presidential election.
Morning Consult obtained population parameters for registered voters from the November 2012 Current Population Survey. We applied post-stratification weights at the state level based on gender, age, educational attainment and race using the American Community Survey (ACS).
Standard errors for our estimates for each state were calculated by taking 100 bootstrap samples with replacement from our full national dataset for each hypothetical matchup and then assessing this empirical distribution at the state level. The distribution of these predictions at the state level allows us to construct a predictive interval, which gives us a sense of the spread of MRP estimates. The 95 percent predictive intervals range from 1 percentage point in larger states such as California, Florida, New York, Pennsylvania and Texas, to 5 percentage points in the smaller population size of Alaska.