Data trump assumptions

Well that's interesting. The Mythology Of Trump’s ‘Working Class’ Support by Nate Silver.

It’s been extremely common for news accounts to portray Donald Trump’s candidacy as a “working-class” rebellion against Republican elites. There are elements of truth in this perspective: Republican voters, especially Trump supporters, are unhappy about the direction of the economy. Trump voters have lower incomes than supporters of John Kasich or Marco Rubio. And things have gone so badly for the Republican “establishment” that the party may be facing an existential crisis.

But the definition of “working class” and similar terms is fuzzy, and narratives like these risk obscuring an important and perhaps counterintuitive fact about Trump’s voters: As compared with most Americans, Trump’s voters are better off. The median household income of a Trump voter so far in the primaries is about $72,000, based on estimates derived from exit polls and Census Bureau data. That’s lower than the $91,000 median for Kasich voters. But it’s well above the national median household income of about $56,000. It’s also higher than the median income for Hillary Clinton and Bernie Sanders supporters, which is around $61,000 for both.

[snip]

Trump voters’ median income exceeded the overall statewide median in all 23 states, sometimes narrowly (as in New Hampshire or Missouri) but sometimes substantially. In Florida, for instance, the median household income for Trump voters was about $70,000, compared with $48,000 for the state as a whole. The differences are usually larger in states with substantial non-white populations, as black and Hispanic voters are overwhelmingly Democratic and tend to have lower incomes. In South Carolina, for example, the median Trump supporter had a household income of $72,000, while the median for Clinton supporters was $39,000.

[snip]

Many of the differences reflect that Republican voters are wealthier overall than Democratic ones, and also that wealthier Americans are more likely to turn out to vote, especially in the primaries. However, while Republican turnout has considerably increased overall from four years ago, there’s no sign of a particularly heavy turnout among “working-class” or lower-income Republicans. On average in states where exit polls were conducted both this year and in the Republican campaign four years ago, 29 percent of GOP voters have had household incomes below $50,000 this year, compared with 31 percent in 2012.

Likewise, although about 44 percent of Trump supporters have college degrees, according to exit polls — lower than the 50 percent for Cruz supporters or 64 percent for Kasich supporters — that’s still higher than the 33 percent of non-Hispanic white adults, or the 29 percent of American adults overall, who have at least a bachelor’s degree.

Very interesting. Was I wrong in remembering exit polls indicating otherwise. Did I simply pick up on a narrative without really investigating it sufficiently. Or is Silver simply wrong. He has been wrong about Trump the entire election cycle.

I am pretty comfortable writing off the third explanation. Silver is a bright guy and has already done a postmortem on his own primary forecasting failure. Still, are his data samples wrong?

I don't know. But a great reminder to check your data before you rely on your assumptions. Particularly assumptions you may have picked up from the mainstream media.