Polling is broken. Here is how the Dorworth Poll will be different
The methods that once made polls reliable no longer work the way they did. The answer is to show readers everything: the methodology and the weighting behind every poll.
Opinion pieces reflect the author’s views.
In brief
The big picture: A random sample of telephone users, the foundation of modern polling, can no longer be reached at scale.
Why it matters: Response rates have collapsed and calls are flagged as spam, so the people a pollster reaches are not a random slice of anyone.
The bottom line: Some published polls are not what they claim to be, and readers have no way to tell.
What's next: The Dorworth Poll will publish its methodology and its weighting with every result.
For decades, the promise of a poll rested on one idea: call a random sample of people, and the answers of the few will reflect the views of the many. The idea still holds in theory. The trouble is that a random sample of telephone users can no longer be reached at scale.
Fewer people answer calls from numbers they do not know, and response rates have collapsed. Carriers and phones now flag unfamiliar calls as spam, so many calls are never seen at all. The people who do pick up and stay on the line are not a random slice of the electorate. They are the people who pick up.
What that does to the numbers
Pollsters make up the difference with weighting: adjusting the answers they collect so the sample resembles the population on age, sex, education, region, past vote and more. Weighting is necessary, and it is also where much of a poll’s result is decided. Two pollsters with the same interviews can publish different numbers because they weighted them differently.
Readers almost never see those choices. And some published polls are not what they claim to be: surveys described as scientific that were not conducted as described, or results released without the information anyone would need to judge them.
Two pollsters with the same interviews can publish different numbers because they weighted them differently.
What the Dorworth Poll will do
The answer is not a better promise. It is transparency. Every Dorworth Poll will be published with its methodology: who was surveyed, how they were reached, when, and what they were asked, in the order they were asked it. And every poll will be published with its weighting: the targets the sample was adjusted to, where those targets came from and how the adjustment changed the result.
No one should have to take a poll on faith, including ours. When the methods are on the page, readers can judge the numbers for themselves.
Publisher and Editor
Chris Dorworth is the founder and publisher of Dorworth News Corporation and the editor of Dorworth News. He served in the Florida House of Representatives from 2007 to 2012, representing House District 34, and was Speaker-designate. He holds an MBA from Duke University and is a former managing partner of a Florida government-affairs firm. He is an author and the host of The Contrarian Raconteur podcast.
More by Chris Dorworth