How to make sense of the electoral divisions in this year's Republican primaries and caucuses? The contours of Donald Trump's support and opposition don't fall on traditional lines.
There's not a regional division, for example. Trump's best states have been Massachusetts, Mississippi and Arizona. We're not seeing the divide between evangelical Christians and others apparent in the 2008 race between Mike Huckabee and John McCain.
We're not seeing the suburbs/countryside division of 2012, when in crucial primaries Mitt Romney carried a million-plus metro areas and Rick Santorum carried almost all other counties. Trump carried metro Detroit and Chicago — Romney country last time — but lost to Ted Cruz in Raleigh-Durham and Kansas City.
So what factor distinguishes Trump and non-Trump voters? My answer is social connectedness or, Robert Putnam's term in Bowling Alone, social capital. Socially connected people have strong family ties and wide circles of friends, are active in churches and voluntary organizations and work steadily.
Putnam's thesis is that social connectedness has declined sharply since the 1950s. But as Charles Murray notes in Coming Apart, that decline is uneven. Whites in the top third of income and education scale still have plenty of social capital. But there's been a precipitous decline among whites in the lowest fifth of those scales. They work less steadily, attend church less often and participate very little in voluntary organizations.
Looking over the election returns, I sense that Trump's support comes disproportionately from those with low social connectedness. My first clues came from the Dutch. Heavily Dutch-American counties in northwest and central Iowa and western Michigan around Grand Rapids were Huckabee and Santorum territory in past years.
This year, unlike surrounding territory, they voted for Ted Cruz, with Trump a poor third. Dutch-Americans have dense networks of churches and civic groups, unusually high social connectedness.
I saw something similar in strong Huckabee/Santorum southern Missouri. Southeast Missouri voted heavily for Trump, but southwest Missouri for Cruz. Southeast Missouri has high rates of disability insurance, an indicator of low workforce participation and low social connectedness. Southwest Missouri, headquarters of the Assemblies of God, has dense networks of civically active churches.
Similarly, exit polls show Trump doing worse with evangelicals who attend church weekly than with those who don't. This helps explain why Trump carried South Carolina and lost Oklahoma, where church attendance is higher.
Then there is majority-Mormon Utah, whose Mormons have very high social connectedness compared to other Americans. Only 14 percent of Utah Republicans voted for Trump.
Putnam reports that social connectedness is highest in states with large Scandinavian and German-American populations, and in Utah. It's lowest in — no surprise — Nevada, one of Trump's best states.
In the 13 states highest in social connectedness, Trump has gotten just 21-35 percent in primaries and caucuses. In the 11 states lowest in social connectedness (excluding Cruz's home state of Texas), his percentages ranged from 33-47 percent.
In states with medium social-connectedness but many retirees who vote in Republican primaries — Florida, Arizona — Trump has run in the high 40s. Similarly in Massachusetts, where only a sliver of voters are registered Republicans, their social connectedness may be limited to listening to Howie Carr on talk radio.
The good news for Trump opponents is that all but West Virginia of the 11 low-social-connectedness states have already voted. Seven of the 13 high-social-connectedness states do, including Wisconsin April 5. There, Milwaukee Journal Sentinel reporter Craig Gilbert finds Trump getting massive unfavorable ratings in the heavily German-American area around Milwaukee, which unlike other northern suburbs remains heavily Republican. Cruz has led Trump in two recent Wisconsin polls.
All remaining contests but one are in states with high social connectedness (Colorado, Oregon, Washington, the Dakotas, Nebraska) and medium levels (the Northeast, New Mexico, California). Many states choose most delegates by congressional districts, and there are no sufficiently granular metrics of social connectedness for precise forecasting.
Still, social connectedness strikes me as the most useful explanation I've seen yet of the variations in Trump's appeal. It's plausible that people with few social connections and inclined to blame elites for their problems might see in Donald Trump, who promises singlehandedly to make things great again, "a sense of collective identity," as Clare Malone of fivethirtyeight.com writes.
So it remains unclear whether a socially unconnected minority will be able to impose their leader on the Republican party and the nation, or whether the socially connected will rally and reject him.