Analyzing the Effect of Regional Modality in Polling Surveys: A Case Study of the 2020 U.S. Presidential Election Results in Florida

Abstract

Pre-election polls have been conducted using different modes of data collection with varying degrees of accuracy. As media’s appetite for horserace polling increased, the demand for all types of data collection increased, raising the price of achieving a representative sample. Mixing modes of data collection have been popular due to the increased cost of polling using traditional methods such as face-to-face interviews, mail-in surveys, or live operator phone surveys. Due to technological advances in self-administered surveys, other forms of data collection provide more cost-effective solutions. Florida Atlantic University’s Business and Economics Polling Initiative conducted political polls throughout the fall of 2020 in anticipation of the general election for President of the United States. This unique dataset from polls conducted in September and October of 2020 showed interactive voice response (IVR) landline polling to be the most accurate statewide and across most regions of Florida. The survey data collected via opt-in online panels and via mobile phone text messaging exhibited results outside of the margin of error for certain regions. However, given the trend of declining landline phones and greater instances of exclusive mobile phone usage, polling by way of online and mobile phone surveys is becoming increasingly important.