Polling Accuracy Reaching a Tipping Point as IVR Surveys Pollute Averaging
With dozens of polls on the presidential and congressional races out weekly in the closing phase of the 2012 campaign, many of which are IVR (robo-polls) as opposed to live caller polls, even taking an average of polls a la the Real Clear Politics average is becoming questionable.
Charlie Cook’s briefing for National Journal members this week was instructive in that he generally trashed the cacophony of polling in general and IVR polls in particular. Cook bemoaned the fact that in many Senate races, for example, “well over half” of the data reported in the media is IVR-oriented.
With the preponderance of IVR polls and response rates for live caller surveys down into single digits per 100 calls, the polling industry is seemingly at a crossroads as we close the 2012 election cycle.
Adding to the confusion is the cell phone mix of the sample, with different pollsters using different formulas to achieve what they feel to be the right percentage versus land line phones.
Until election day, we won’t be able to assess the accuracy of various data. Regardless, it seems like we’ve reached a fork in the road in terms of how pollsters will pursue their craft to achieve maximum accuracy at a time when so many variables add to the ongoing confusion.