
Additionally, Clinton won Indiana voters who made their choice in the last week, by 56-44 percent. She did less well among those who decided earlier.
Key in Indiana were working-class voters; 65 percent lacked a college degree, compared with an average of 53 percent in all primaries to date. While Obama tried to improve his appeal among working-class whites, the exit poll found a 64-35 percent Clinton advantage in this group in Indiana (and 71-26 percent in North Carolina); she won them by 61-32 percent in all previous primaries to date.
Continuing the same socioeconomic division that's marked the primaries all year, white college graduates divided evenly, 49-50 percent in Indiana, and 53-46 percent in North Carolina.
As noted, the racial makeup of the electorates themselves played a major role. Blacks accounted for 34 percent of North Carolina voters and 91 percent of them supported Obama. Against that voting bloc Clinton would have needed 70 percent of non-black voters; she fell well short, with 59 percent.
Obama also won nine in 10 blacks in Indiana. They accounted for 18 percent of voters there – far fewer than in North Carolina, but a record nonetheless for Indiana, surpassing 15 percent in 1988.
Clinton won white men in Indiana by 58-42 percent, and in North Carolina by 55-42 percent, both better than her total across all primaries this year, an essentially even 47-46 percent. However, she's done as well with white men in some states before, including 57-43 percent in Pennsylvania and 58-39 percent in Ohio.
Indeed, race largely eclipsed sex as a factor in vote preferences: Clinton won white men and women by similar margins in both states; Obama, ditto.
Clinton had a broad advantage -- 66-34 percent over Obama -- among rural and small town voters in Indiana, who accounted for one in six voters. That was just below her previous bests in this group, in Tennessee, Arkansas and Ohio. But Obama won in mid-sized and larger cities by a broad 61-39 percent, better than his usual.
Another result, mirroring one in Pennsylvania, indicated a small but negative impact of racially motivated voting.