As we head into the 2022 midterm U.S. elections we are seeing a lot of confusion with the polling system. For 60+ years we have had political polls at the center of our elections. They called up thousands of random people with a quick survey to research the election and make predictions. Until we had cell phones the surveys could depend on the responder answering their phone at home and taking the time to answer the questions. With enough random calls completed, a demographic and geographic balance (or weighting) could predict within 5% of the election results - which is the sampling error.
BUT - now we have the cell phone, with caller ID built-in, we carry it everywhere and even drive with it. It even shows us telemarketing and spam calls so getting a 'random sample' is almost impossible as very few are answering the call. We also don't have a reliable 'phone book' with most of us keeping our cell numbers hidden from public view. Most researchers use random number systems to try and make calls.
In the end pollsters and researchers struggle to get respondents and then even when they answer the time is limited as they walk, drive, shop, ride elevators, and who knows what else while talking on the phone. Some have tried to go to online research patterns but then how do you obtain a truly random sample?
All of these drawbacks and 'excuses' have left most pollsters and political watchdogs confused. Here on election eve, some are making big predictions, but are also walking them back after they have seen the last 4 elections end up with results that the polls missed.
Here in radio land we also have become dependent on a lot of research and polls to determine who we program and relate to the audience. Is this research helping, or just bringing in lots of false information that we end up acting on and failing with?
The same realities that are casting confusion and doubt on the election polling are also baked into our music research, audience-level ratings, and perpetual studies done for years and years.