| Sumario: | The purpose of this paper is to examine the performance of the 2024 US presidential election polls. It is intended to consider the magnitude, directionality, and significance of polling errors using publicly available data and transparent statistical tools. It is demonstrated that presidential election polling accuracy remains compromised by systematic biases. Our aggregate analysis of 100 national and battleground state polls reveals that Republican vote shares have been widely underestimated. A concerning finding is the statistically significant directional bias that resides in several polling contexts. Interestingly, the primary source of these biases resides in the underestimation of Republican vote, rather than the overestimation of Democratic support.
|
|---|