It is still very early in the 2020 campaign, but we are starting to see some movement as the Democratic candidates become more known to the public. Looking at changes in how the top four Democratic candidates match up against Trump since the start of July, a clear pattern emerges: Trump is loosing ground.
The prediction method
I started tracking the state level polls to predict which candidate will win in the electoral college. As we learned from 2016, national polls can mask what is happening at the state level and our system is predicated on state level outcomes. In fact, my predictions have all the major Democratic candidates winning the national vote (as they did in 2016) but not all would currently win the Presidency.
A challenge in predicting voting patterns at the state level is the polling data are spotty and there are no polls for many states. In these cases, I use the 2016 election outcomes as a proxy for the expected current voting patterns. The states for which there is no current polling data tend to be highly partisan. Pollsters concentrate their efforts on the swing states. A change in the electoral college outcomes for highly partisan states would require a major shift in voting patterns. Therefore, I contend that using 2016 voting patterns to predict the 2020 electoral college outcomes for states such as Oklahoma is highly likely to be accurate.
shifts in the 2020 Predictions
As the graph below shows, if the election were held today, Joe Biden and Bernie Sanders would beat Trump handily. Biden holds the largest electoral college margin and it has remained steady. Sanders has widened his electoral college margin over the last 6 weeks. In contrast, Trump would beat Kamala Harris and Elizabeth Warren, However, his margins have shrunk substantially over the last 6 weeks. He holds on a 2 electoral vote margin over Harris and a 17 vote margin over Warren, making these races basically toss-ups.
In addition to Trump losing ground based on electoral college predictions, he is also losing ground in the overall popular vote percentages based on the state polls. As the tables below show, the polls suggest he has lost ground relative to all four of the leading Democratic candidates since July 1, 2019. (On the second table, declines in support are noted in red and increases are noted in green), This is a result of his poll numbers sagging in both swing states and states that are have historically voted Democratic. That is, in Democratic states he is becoming even more unpopular and he is struggling in swing states such as Michigan, Pennsylvania, and Wisconsin.
Dividing the country may prove to be a bad strategy
These shifts in the polls may simply reflect Harris and Warren becoming more known to the electorate. But, Trump continues to have an antagonistic relationship with a large share of the electorate. He has embraced a win strategy that entails attacking and belittling those that disagree with him and stoking racial tensions and xenophobic fears. Based on the past six weeks of state polling data, this strategy is not working.