Today, the "Arkansas Poll" was released by the University of Arkansas’ Department of Political Science, and once again I have serious doubts about the quality of their polling data.
I often joke the UA’s ‘Arkansas Poll" is just a poll of educated, white Baby Boomers and you’ll see there is some truth to that jest.
This is the 13th annual poll conducted by UA’s political science department and I believe the people they interview are not truly representative of Arkansans. UA polled 800 people from October 14-19 from a random sample of Arkansans.
Here are examples of what I’m driving at when it comes to the poll sample and how it doesn’t truly doesn’t represent Arkansans.
In Arkansas, roughly 19% of people over 25 have a bachelor’s degree, but in the UA’s recent poll sample a whooping 35% of the people interviewed were college graduates. Also, roughly 15.4% of Arkansans are African-American, but the UA’s poll sample only had 8% African-American respondents. 77% of our population is white, but the UA’s sample was 81% white.
It seems the UA is polling people who aren’t reflective of the average Arkansan.
To UA’s credit, however, they do release the demographic data on their polling sample, which some polling organizations don’t do.
It’s not clear if the poll respondents are registered voters or even have a proven history of voting. The poll asks the respondents if they are registered, but I didn’t see on the poll summary of the 800 how many said they were actually registered. But when it comes to political polling, it’s better to use a random sample from a registered voter list because some respondents will claim to be registered when they are actually not.
Their most famous polling mistake was in late October of 2008 when they polled the Proposed Initiated Act 1, which would prohibit same-sex couples from adopting or serving as foster parents. The UA’s poll said that 55% of Arkansans opposed the measure and only 38% supported it, which meant it would surely be defeated in November.
However, Initiated Act 1 passed with a 57% majority. Ouch. The poll wasn’t even close and that almost 20% percent mistake can’t be explained away with the notion of a last minute surge. The UA just got it wrong.
The UA traditionally releases their poll in late October, which I strongly disagree since their polling data has an impact on the political narrative every election year. I personally have a problem with a taxpayer-funded institution releasing polling information on the eve of an election that creates media stories and causes candidates to respond to their questionable data.
As you can see, I am not a fan of the Arkansas Poll and don’t put much stock in their polling data.