Reporting out TNR's Ohio Primer, the phrase I heard most often from pollsters and political scientists was, "Hillary should do better than Obama in that district, but not 60% better." Districts with an even number of delegates might have favored Hillary, but her lead wasn't supposed to be lopsided enough that she'd win an extra delegate.
Come March 5, colleagues at the office may have noticed my jaw dropping. Hillary slaughtered Barack Obama in several of these districts, picking up 4-2 leads in places that were supposed to remain 3-3. She also pulled off a massive 4-1 win in CD6 and cleaned up in college-student heavy districts like CD18.
Now, it turns out she won at least one of those delegates because of crossover Republicans. The Cleveland Plain Dealer has exit data on crossover voters in CD10:
A staggering 16,000-plus Republicans in Cuyahoga County switched parties when they voted in last week's primary.
That includes 931 in Rocky River, 1,027 in Westlake and 1,142 in Strongsville. More than a third of the Republicans in Solon and Bay Village switched. Pepper Pike had the most dramatic change: just under half of its Republicans became Democrats.
Subtract 16,000 from Hillary's vote total and she gets 57.4% of the vote in CD10--Dennis Kucinich's district--as opposed to the 62.4% she actually received. Plug that into the Ohio delegate calculator and you'll find Republican crossovers pushed her over the threshold, delivering her a 4-2 victory instead of a 3-3 tie.
While the Plain Dealer doesn't have data for other districts, it gives evidence of heavy crossover voting statewide. Considering the vote totals in Republican-leaning CD14 (which Hillary also won 4-2) were very similar, it's not a stretch to suspect a comparable pattern caused her win there.
An estimated 24% of Hillary's support in the Mississippi primary came from Republicans. In light of the evidence, it's probably safe to say Hillary's expectation-defying victory in Ohio benefited from a similar dynamic.