Things looked pretty good for Mark Pryor a few months ago when several non-partisan polls showed him leading Tom Cotton by solid margins, a few even by double digits. Since then we have been hit by an onslaught of right wing affiliated pollsters that show Cotton ahead about 5-7 points. It gets a bit disheartening when you look at the HuffPo poll charts to see Cotton with a pretty hefty lead, albeit on the back of these republican pollsters:
2014 Pryor v Cotton
So today when yet another republican poll came out showing Cotton with a 4 point lead, I just had to look inside to see what was up.
I'm not a poll unskewer, but I would consider myself a poll skeptic. Small differences in the weighting of age, race, or party should generally not be fussed over, but when there are large discrepancies with reality, it needs to be pointed out. Take the latest in the Arkansas race, a poll by Impact Management Group (R):
2014 Impact Management Group (R) Survey
It shows Tom Cotton with a 47-43 lead over Mark Pryor, and while crosstabs are not provided, some of the demographics are. I usually don't like to look at party affiliation as they can pretty nebulous, but race and age are usually pretty steady, and their changes year to year are roughly predictable. For comparison I looked at the 2010 Arkansas senate election exit polls at CNN:
2010 Arkansas Senate Exit Poll
This should be a pretty good comparison as it represents the rock bottom for democrats: a midterm election in a particularly bad year for democrats with an only lightly contested race and a relatively unknown candidate.
The racial breakdowns in the poll seem pretty on par with the election exit polls, but what differs greatly is the age demographics. The 2010 exit polls breakdown age from 18-29, and from 30-39, while the poll does 18-34, but some comparisons can still be made. In 2010 8% of the electorate was 18-29 while 12% was 30-39. So we can reasonably figure the percent of 18-34 year olds is about 13-14%. This was not coincidentally the age group that the democrats did the best with in 2010. What is the percent of 18-34 year olds in the 2014 poll? 3%. Pretty much negligible and no where in the ball park of reality. On the flip side, in 2010 voters 65 and older comprised 32% percent of the electorate, and again not coincidentally were the group that went most heavily for the republican. In the 2014 poll? 42%, almost a one third increase from the 2010 exit poll.
The point of this isn't to say well this polls should really be 48-43 for Pryor if we adjust the demographics to look like 2010, that's just bunk. The point is to say that even with demographics that are markedly worse than 2010 it looks like Pryor is still in it (and hopefully we'll get demographics that are markedly better than 2010). That being said I would love to see some recent non-partisan polling to assuage my fears that this race may be slipping away. I don't think it is, I just hate seeing all this red ink. Any comments?
Edit: Not all of the recent (R) polls provide demographic breakdowns, but looking at a couple other recent ones that do show more realistic age breakdowns that generally jive with 2010 results. This just heightens my desire to see some more recent non-partisan polling. The Arkansas seat is one we can ill afford to lose.