Hey guys, I noticed something quite dramatic in the cross-tabs for the North Carolina presidential poll carried out by National Research, sponsored by Civitas. The poll is available from their website here. Basically the poll is great for Obama from just the headline comparison but is ZOMGWTFBBQ amazing once Gary Johnson is included.
So let's go through the poll after the squiggle.
The poll asks people who they would vote for and then for each candidate the available answers are 'definitely', 'probably' or 'lean' i.e.
The results for the presidential race when it's just Romney/Ryan against Obama/Biden with leaners i.e. with all 'definitely' + 'probably' + 'lean' is:
With leaners Romney/Obama 45/49
Take away the the voters who answered 'probably' or 'lean' and the results become:
Without leaners Romney/Obama 40/44
So basically both candidates lose about 5%, so that's pretty equivalent to the result with 'leaners' included and nothing is particularly amazing about it.
So let us look what happens when Gary Johnson is included in the poll. First including the leaners the results are:
With leaners Romney/Obama/Johnson 43/47/2
i.e. both Romney and Obama both lose a couple of percentage points but the result is pretty much in line with the headlne result of Obama leading by 45/49. But now what happens to the result without leaners?
Without leaners Romney/Obama/Johnson 36/43/1
Adding Gary Johnson to the mix makes Obama's lead amongst committed voters be massive over Romney. I doubt that all those uncommitted Romney voters are likely to switch to Obama but it is likely that at least some of them will either not bother to vote at all or make a protest vote against Obama by voting for Johnson.
Unfortunately the other recent National Research polls are not available online, and other pollsters are either not giving out cross-tabs or don't include Johnson in the question. I would love to know if this pattern of soft support is in other states as Johnson is on the ballot on every state except Oklahoma, Pennsylvania and Michigan.