Top Line: The Republican coalition in Maryland took a sharp turn toward less educated voters and remains heavily white.

The above is a quickly done ordinary least squares (OLS) regression analysis of aggregate support for Trump and the change in the vote from 2012 to 2016 by county. The numbers in parentheses are standard errors around the coefficients shown above them.

The coefficients are easy to interpret. For example, the coefficient of .77 on Percent Non-Hispanic White in the model of Percent Trump indicates that 1% increase in the non-Hispanic white population resulted in a .77% increase in the Trump vote. Similarly, an increase of 1% in the share of college graduates reduced the Trump vote by .54%.

The second model of the share of the Trump vote in Maryland also includes a control for how counties voted on the MD Dream Act in 2012 as a further control for immigration attitudes. While the results are at the edges of statistical significance, this model indicates that, even after controlling for race and education, that a 1% increase in opposition to the Dream Act resulted in a .67% increase in support for Donald Trump.

The third and fourth models assess the impact of the same variables on the net change in the vote from 2012 to 2016. The change in the vote equals (%Trump – %Romney). Educational levels dominate changes in the share of the vote won by the Republican and Democratic nominees. While neither race nor the opposition to the Dream Act matter much, a 1% increase in the share of college graduates resulted in a loss of .32% by Trump.

The final model shows the relationship between race and college education to opposition to the Dream Act. It suggests that a similar coalition supported Trump in 2016.

The votes are still being counted in Maryland and across the country. While Trump has clearly won the Electoral College, whether he will carry the popular vote remains in doubt. In the meantime, let’s look at preliminary results in Maryland after last night’s political earthquake.

Though the earth didn’t move nearly as much in Maryland as elsewhere, there were significant changes in support patterns, which Clinton carried easily as expected. Here are the 2016 preliminary and 2012 final percentages side-by-side:

As you can see the preliminary results, suggest that both candidates lost ground in Maryland. In order to more easily examine shifts across the state, I’ve created a table showing vote shifts by county with counties reordered by net change.

As you can see, Maryland shifted about a net 1.4% to Trump but that masks major differences across the State. The two counties with the most wealthy, well-educated, liberal white populations–Howard and Montgomery–shifted heavily to Clinton.

In these two counties, the Democrats had stunning net gains of over 10% of the vote, as Clinton gained 3% or more and Trump lost more than 7% in both Howard and Montgomery. Howard went 2-1 and Montgomery 3-1 for Clinton.

Another set of counties reveals the opposite pattern. In 14 counties, Clinton lost ground over Obama while Trump gained compared to Romney. Among the 14 counties, 11 have median incomes below the Maryland average–Calvert, St. Mary’s and Queen Anne’s are the exceptions, though none has a median income higher than Howard or Montgomery.

In addition to a tendency toward lower incomes, these areas tend to have fewer college-educated whites and be socially more conservative than Howard or Montgomery. Perhaps critically, pro-Trump counties tend to be far more suspicious of immigrants based on how they voted in the 2012 referendum.

The lowest rates of change were in the State’s two majority black jurisdictions of Baltimore City and Prince George’s, as well as more moderate jurisdictions like Frederick and Anne Arundel that tend to be closely divided between the two parties.

Overall, the key takeaway is greater geographic polarization between the Democratic and Republican areas of the State. In general, the areas that lean the most Democratic became more Democratic or stayed much the same. On the side, areas that already listed heavily Republican became more so.

PollyVote surveyed 673 political scientists over the past four days and asked them to predict the outcome of the election in their state. These forecasts suggest that Clinton will win with 358 electoral votes with just 180 going to Trump:

Here are the predictions for the swingier states. The second column is the number of political scientists who responded to the survey. The third column is the predicted chance that Clinton will carry the state and the final column is the share of the two-party vote.

Needless to say, Maryland is not a swing state with the 26 respondents gauging the likelihood of Clinton winning at 100% and guessing her share of the two-party vote at 67.0%.

Read Adam Pagnucco’s critique of the Democratic response to the Hogan/Franchot proposed post-Labor Day school start posted this morning.

The Washington Post played monkey-see, monkey-do this morning in its coverage of last night’s debate. On page A1 (yes, I still get the print edition) and on its homepage, the Washington Post covered Matt Lauer’s questions about Clinton’s email that dominated the forum supposedly about national security.

It buried on A6 and left off the homepage the story that Clinton was absolutely right about Colin Powell’s advice regarding the use of her cell phone. As The Hill reported in a story titled “Powell told Clinton how to bypass State security measures:”

Powell says that he used a private phone line to keep his communications out of the State Department servers.

“What I did do was have a personal computer that was hooked up to a private phone line (sounds ancient.),” Powell wrote. “So I could communicate with a wide range of friends directly without it going through the State Department servers.”

“I even used it to do business with some foreign leaders and some of the senior folks in the Department on their personal email accounts. I did the same thing on the road in hotels,” he said.

Powell was responding to a question from Clinton about the restrictions on using a BlackBerry while in office. 

[Rep. Elijah] Cummings added that Republicans’ pursuit of Clinton over her server was politically motivated, otherwise they “would be attempting to recover Secretary Powell’s emails from AOL.”