I'm generally pretty skeptical of any polling this far out from an election. I am, however, very skeptical of the value of general election head-to-head match-ups this far out. 538's Harry Enten has a good articulation of why, in case it's not self-evident, but consider the situation. We're asking people who they'd pick before each party has settled on its candidate -- meaning while there's still internal partisan tension -- before any campaigning and before a year's worth of economic shifts happen.
There is some value though in one subset of head-to-head data, I think: Looking at how demographic groups view different candidates relative to one another.
On Tuesday, McClatchy and Marist released a new poll looking at the Republican field. Included in the results were a phalanx of head-to-head match-ups, pitting Hillary Clinton and Bernie Sanders against six different Republicans: Jeb Bush, Ben Carson, Ted Cruz, Carly Fiorina, Marco Rubio and Donald Trump.
Carson does better against the Democrats than the other five but, you know, 364 days until the election.
Here are all of the juxtapositions. Notice what jumps out.
The data on the independents is perhaps the most interesting. Carson does much better with independents against Clinton than Sanders -- perhaps as a function of her name ID or his political ideology. But independents are all over the map.
What else jumps out? Clinton's advantage with women isn't that much larger than Sanders's (though Sanders does better with men). Both Democrats do far, far better with non-white voters than do the Republicans -- as do less-wealthy voters.
In other words: the Democratic base supports the Democrats.
There's gray area in particular contests, but there's also a margin of error that isn't factored in. Demonstrating why we shouldn't worry too much about these polls: It's hard to know what is meaningful here and what isn't, beyond the big picture.