Polling is Harder Than Ever, But Accurate Results are Still Possible

lavarr policy insightsA recent article in a special section of the Wall Street Journal addressed the future of polling, predicting that “big data” analytics will be the next frontier in public opinion research.

So will traditional polling go away? Survey research is critically important in politics. It’s also very important to Utah Policy Daily (UPD), as the only Utah news organization commissioning consistent polling. It’s also obviously very important to polling firms like Dan Jones & Associates, which conducts surveys for UPD. 

Advanced technology is disrupting many industries and it is having a big impact on polling. Dan Jones incorporates new tools of technology in the research he conducts.

The WSJ article noted that everything anyone does in the digital world leaves data behind. Since we spend so much time using digital tools for reading news, shopping, researching, communicating with friends, tweeting, posting, etc., using algorithms to analyze that data can reveal our opinions.

To a point.

Certainly, big data analytics can reveal consumer preferences. Those ads that follow you around the Internet pop up because you’ve spent time looking at similar products or services. But can data analytics reveal whether you support Doug Owens or Mia Love? Certainly, data algorithms can’t measure values, beliefs and concerns very well.

Computers can track what news stories you read and how much time you spend on various topics. They can also track what you and your friends post and tweet. Retailers and data firms are learning more and more about us. But the tools are still fairly dull. They can provide indications, but probably can’t accurately tell if you support a tax increase for education or if you oppose guns being allowed on campus.

And there are still a lot of old-timers (like me) who vote in high numbers, but may not reveal much about themselves in limited digital activity.  

A big problem in polling is that new communications and aggregation tools allow all manner of surveys that don’t really reflect public opinion. They do reveal the opinion of the people taking the surveys, but they don’t represent an accurate cross-section of targeted audiences.

An on-line survey of 10,000 people who self-select to take the survey won’t be as accurate as a survey of 600 respondents drawn as a true representative sample. 

The WSJ article concludes that public opinion research in the future will combine big data analytics with traditional polling, taking the best from both worlds.

The hardest thing about polling remains drawing a proper sample and then reaching that sample for responses. That’s where many pollsters fall short. Dan Jones works incredibly hard to get the right sample, the right demographics. He uses a combination of landline calls, cell phone calls, email and on-line panels to reach the right demographic groups in proper proportions. Sometimes it takes several days to get a representative sample and complete a survey.

It’s very hard work. Much harder than it used to be. Corners can’t be cut. It’s a tribute to Dan Jones that after 55 years of polling he’s adapted to the new tools of technology, still gets great samples, and still produces accurate results.