Francesca Hong was projected to win Wisconsin’s gubernatorial contest by as much as 22 percentage points according to the last five polls, but she lost narrowly. Similarly, Abdul el Sayed was shown to lead in the Michigan Senate Democratic primary but also won by a narrow margin. These outcomes have raised concerns about the reliability of polling, particularly in primary elections, which an analysis indicates can be off by an average of 13 points.
Several factors contribute to polling inaccuracies, including the difficulty in estimating turnout and the composition of the electorate, especially in closely contested states with open primaries. Pollsters are attempting to address these issues, and the American Association for Public Opinion Research has convened experts to analyze polling failures from previous elections.
Despite efforts to improve polling accuracy, recent failures have left a lasting impression on the public, with experts suggesting that inaccuracies may continue. A recent New York Times/Siena poll surveyed 3,659 likely voters across six competitive states, highlighting the challenges faced by pollsters in accurately capturing voter sentiment.
Additionally, government officials are concerned about declining response rates and the increasing costs associated with conducting surveys. The Federal Reserve's Survey of Consumer Finances, which has been conducted since 1992, faces similar challenges, with NORC exploring incentive strategies to improve response rates.
The failures in election polling may affect public willingness to participate in surveys, indicating a potential spillover effect on other important statistical surveys.