In talking over the past few weeks to some very smart people about the primaries and caucuses, I have been surprised how many of them assume that Iowa and New Hampshire will be as predictive of the nomination for both parties. I think that’s probably wrong for the reason highlighted in this table.

GOP primary voters/caucus-goers are as white as the driven snow in all four contests (and probably in all the ones afterwards, though I didn’t check). Even in lily white Iowa, they are whiter than the general population. A Republican candidate who wins Iowa’s white Evangelical voters and New Hampshire’s white Establishment voters should have a ride to the GOP nomination that’s as smooth as mayonnaise on Wonder Bread .

But the Democratic primaries are different because the people who participate in them are racially and ethnically diverse. Iowa and New Hampshire are an unusual pair of starting states for Democratic candidates because they are much more monochromatic than the party as a whole. If we assume that race and ethnicity affect candidate preferences (which seems a good assumption this year at least), it’s easy to imagine an scenario in which a Democratic candidate does badly in the first two contests and well in the next two, or vice versa.

The data come with caveats, the biggest one being that because the Democratic nomination in 2012 was a lock from day one, news organizations didn’t invest in exit polling and we are thus stuck with 2008 data. I have linked all my sources below and I welcome crowdsourcing of better data — if you know of some please it in the comments.

Iowa: Caucus data from the past two cycles for Republicans and 2008 only for Democrats.

New Hampshire: 2012 exit poll for Republicans and 2008 exit poll for New Hampshire.

South Carolina: Exit poll data from 2012 for Republicans and 2008 for Democrats.

Nevada: 2012 Republican entrance poll and Democratic 2008 exit poll.

[Cross-posted at The Reality-Based Community]

Keith Humphreys

Keith Humphreys is the Esther Ting Memorial Professor at Stanford University. @KeithNHumphreys