The accuracy of political polling has been trending downward over the last ten years and is now beginning to lose its credibility. Unfortunately it seems that both pundits and the general public seem to feel that polls are reflective of the probable outcome of an event. In fact, most of all of the current political news begins by citing the latest poll, and we now know that there is a strong likelihood that the poll is wrong!
A good example is in the presidential contest of 2012. Pollsters were predicting a virtual tie between Obama and Romney and well known and respected Gallup predicted a Romney win. But Obama won by over 4 % with 5 million more votes. And the problem with poll accuracy is worldwide. Pollsters got the referendum on Scottish independence wrong and it was clearly rejected. But the same problems have been seen in Britain and in Israel. So what has gone wrong with the polling system and does it reflect the same concerns in other polls? Please be aware that I am discounting the fact that 93 % of dentists believe that toothpaste A cleans 98 % better than toothpaste B as marketing BS.
The prevailing wisdom about opinion poll failure seems to come down to the convergence of two events. The first is the growth of cellphones. Ten years ago 6 % of the population relied on cellphones and by 2014 the number was close to 60 % using primarily non-landline communications. In the past, opinion researchers would autodial landlines and if they picked up they went to a live interviewer. But Federal law prohibits autodialing cellphones so researchers have to manually dial landlines and this shifts the outcome. How? The landlines are mostly answered by those over 50 and this leaves out the young and mobile millennials. It also discounts the less affluent and minority population that uses mobile phones only. The other compounding factor with cellphones is that when people move, they take their number with them so following the area code becomes much more problematic. The likelihood of a poll capturing a given part of a state becomes risky at best. So, pollsters use a system of compensation called “weighting” to adjust the results. But that is another can of worms that I don’t want to open in this post!
The second factor involves the continued use and abuse of telemarketers. They have poisoned the well for people who used to answer a landline phone. Even the residents who have added their number to the “Do Not Call” list have discovered that they are not immune to being annoyed by unwanted calls for aluminum siding and lower credit rates. With so many options for telephone marketers to hide their locations, like spoofing, they are difficult to identify and stop. So, people with landlines screen their calls or let them go to answering machines or voicemail.
As far as I am concerned the polls are a waste of time and we will not know who the next president will be until after the election. So, the news programs can continue to present polls in colorful splendor, and I will continue ignoring them. Oh wait! Another two dentists say that toothpaste A really is better. I should start buying that brand…
A good example is in the presidential contest of 2012. Pollsters were predicting a virtual tie between Obama and Romney and well known and respected Gallup predicted a Romney win. But Obama won by over 4 % with 5 million more votes. And the problem with poll accuracy is worldwide. Pollsters got the referendum on Scottish independence wrong and it was clearly rejected. But the same problems have been seen in Britain and in Israel. So what has gone wrong with the polling system and does it reflect the same concerns in other polls? Please be aware that I am discounting the fact that 93 % of dentists believe that toothpaste A cleans 98 % better than toothpaste B as marketing BS.
The prevailing wisdom about opinion poll failure seems to come down to the convergence of two events. The first is the growth of cellphones. Ten years ago 6 % of the population relied on cellphones and by 2014 the number was close to 60 % using primarily non-landline communications. In the past, opinion researchers would autodial landlines and if they picked up they went to a live interviewer. But Federal law prohibits autodialing cellphones so researchers have to manually dial landlines and this shifts the outcome. How? The landlines are mostly answered by those over 50 and this leaves out the young and mobile millennials. It also discounts the less affluent and minority population that uses mobile phones only. The other compounding factor with cellphones is that when people move, they take their number with them so following the area code becomes much more problematic. The likelihood of a poll capturing a given part of a state becomes risky at best. So, pollsters use a system of compensation called “weighting” to adjust the results. But that is another can of worms that I don’t want to open in this post!
The second factor involves the continued use and abuse of telemarketers. They have poisoned the well for people who used to answer a landline phone. Even the residents who have added their number to the “Do Not Call” list have discovered that they are not immune to being annoyed by unwanted calls for aluminum siding and lower credit rates. With so many options for telephone marketers to hide their locations, like spoofing, they are difficult to identify and stop. So, people with landlines screen their calls or let them go to answering machines or voicemail.
As far as I am concerned the polls are a waste of time and we will not know who the next president will be until after the election. So, the news programs can continue to present polls in colorful splendor, and I will continue ignoring them. Oh wait! Another two dentists say that toothpaste A really is better. I should start buying that brand…
RSS Feed
