Abstract
Abstract Recent high-stake occurrences of polling error led some analysts to conclude that the polls are “broken.” One question is whether domain knowledge can help identify the direction and magnitude of polling error before the fact, which would reduce uncertainty and improve accuracy. The present study compares the accuracy of a poll average to 4,494 expert forecasts, obtained across three German federal elections. Experts’ domain knowledge was unable to improve upon the accuracy of polls. On average, experts’ forecast error was 34% higher than the corresponding polling error, and in more than half of the cases, experts’ forecasts pointed in the wrong direction. Polls remain to be an important source of information when it comes to forecasting elections, despite their limitations and underlying uncertainty.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
