Polls, Schmolls, They’re All Wrong and Here’s Why
He’s up, He’s down, He’s up and down at the same time! WTF? What is going on here?
Apart from the usual idiocy of the horse-race instead of actually focusing on the issues (and I don’t believe Benghazi-gate is actually an ISSUE!) but that is fodder for another discussion, the polling for the horse-race in this election has been even more a comedy of errors than usual. It will only get worse.
The first problem is that the Gallup organization chose to change its methodology in the middle of the campaign for some reason. Usually it is not a good idea to do that. It seems to have been a particularly egregious decision when the results of the change now put their results about 6 points to the far right of even the far right leaning discredited Rasmussen polling organization.
The second problem is one we have heard about before – cellphone users vs landline users. In 2008, the percentage of cell-only users was just above 31%. Currently that number hovers around 38%. So the problem has gotten worse. Robo-calling of cellphones is prohibited, so the only way polling organizations can get cellphone numbers is by acquiring them through some voluntary way by the user. I recently read an article by Andy Kroll at Mother Jones Pollster: Undercounted Cellphone Users Hide Obama’s Lead. There is an excellent graph at the original piece showing how cell users favor Obama by an up to 11 point margin, while landline only users favor Romney by about a 14 point margin. Dual users are tilted somewhat toward Romney by 3 points but of the dual users who use their landlines more than their cells, Obama comes out on top by 3.
The point is, polling that is done on landline users only – no matter the leanings of the polling organization will be tilted toward the right, and will hide any voters leaning left due to non-polling of the cell-only crowd who are mostly younger, and mostly voting Democrat. This has ramifications not only for the Obama campaign but for a lot of down-ticket races as well.
The third problem is sheer under-response that is not reported by the polling organizations. According to an article by Salvatore Babones at Truthout, Political Polling Is No Longer Meaningful, The Pew Research Center recently did a survey looking at response rates to calling and got a 9% response rate. Pew is a first-rate non-profit and this rate is probably higher than what an impatient, for-profit political polling organization gets for a response rate. In the words of Babones,
When response rates fall this low, polls tell us less about public opinion than about who answers the phone
This problem of lack of responders has been with us since the beginning of public polling, and has also provided us with such huge failures of public polling as the headlines “Dewey Wins over Truman” and the like. The fact that not one of the polling organizations tell us what the percentage of non-responders is becomes quite telling. That margin of error of 3-5% is a joke when the percentage of non-responders is probably higher than 91%. It is hardly a representative sample of the country when most people refuse to answer the phone for an unknown private number or and 800 number they are unfamiliar with. Even if they do answer the phone when they find out it is a political survey they refuse and hang up, it also doesn’t improve the situation. Or in some cases, the polling organization isn’t even using humans to call but robo-calls where the person on the other end is far more likely to just hang up!
All these polling organizations keep assuring us that the people surveyed have been scientifically chosen according to some algorithm to be a representative sample. Given these obstacles I would think it is highly doubtful that algorithm would stand up to the pragmatic realities of today’s population and their relationship to their phones and who is calling.
The fourth problem is under-representation of non-English-speaking persons in the samples. I believe someone here at the Lake wrote a diary on this very subject recently although I didn’t have time while it was still up to read it I would be really appreciative if someone would link it in the comments please? While these pollsters may have assumed these individuals were not citizens – in many cases they are. These people may be naturalized citizens who are registered voters, but who may be far more comfortable speaking their native languages in their homes and on the phone with persons who usually call them. But the pollsters, hearing a “foreign” hello, just hang-up, and then do not include this population in their survey at all. This group includes not just Latino voters but new citizens from all parts of the world.
The fifth problem is the group of people who use VOIP (Voice Over Internet Protocol) phones. I have no idea how big a group that is, I have seen no data on this group and as far as I know – no one is looking at it. This group may be negligible, or it may be large, no idea.
The sixth problem is the group of people who are hearing impaired and who may or may not use special TDD equipment or not depending on the extend of their hearing loss. Again, I have no idea how large this group is, or whether anyone is looking at this one either.
All of this stuff convinces me that watching the polls before the election does nothing to tell me one way or the other what the outcome of the election is going to be. Whoever your candidate is – you need to get out there and work. Get out the vote, get to the polls, get others to the polls. And Vote! And after the election, get active and stay active. Get your county and/or state election boards to get rid of these electronic voting machines and get the vote counting back in the hands of your public officials and out of the hands of Diebold and ESS and these private companies. Prosecute the voter officials who commit election fraud (that’s different from voter fraud). And make the position of Secretary of State (the person responsible for elections) a non-partisan position.
And finally, on the subject of Exit Polling Most of the polling organizations and the media have begun to say that exit polling is unreliable and that they are going to start to phase it out. WRONG WRONG WRONG!! Exit polling is the only kind of polling that is actually still mostly accurate. And it is the only way we really have to have some kind of a check on all these hinky electronic voting machines and the behind the scenes rigged counting and all the rest of the election fraud that is running rampant in our so-called democratic process. After all, as a wise man once said – it doesn’t matter if you vote, it matters who counts the votes!