From Pew:
In March, 82% of those who reported voting for Trump – and whom researchers were able to verify through voting records as having voted in 2016 – said they felt “warmly” toward Trump, with 62% saying they had “very warm” feelings toward him. Their feelings were expressed on a 0-100 “feeling thermometer.” A rating of 51 or higher is “warm,” with 76 or higher indicating “very warm” feelings.
The views of these same Trump voters had been quite similar in November 2016: At that time, 87% had warm feelings toward him, including 63% who had very warm feelings.
This report is based on surveys conducted on Pew Research Center’s nationally representative American Trends Panel. The Center tracked views of Trump among the same groups of Americans in March 2018 and at three points in 2016, including in November shortly after the election. In that survey, respondents reported whom they had voted for.
When state voter files – publicly available records of who turned out to vote – became available months after the election, respondents were matched to these files. Self-reported turnout was not used in this analysis; rather, researchers took extensive effort to determine which respondents had in fact voted. And unlike other studies that have employed voter validation, this one employs five different commercial voter files in an effort to minimize the possibility that actual voters were incorrectly classified as nonvoters due to errors in locating their turnout records.
This study also includes a detailed portrait of the electorate – which also is based on the reported voting preferences of validated voters. It casts the widely reported educational divide among white voters in 2016 into stark relief: A majority of white college graduates (55%) reported voting for Hillary Clinton, compared with 38% who supported Trump. Among the much larger share of white voters who did not complete college, 64% backed Trump and just 28% supported Clinton.
