The United States Commission on Civil Rights has identified five state-level policies that can serve to suppress votership, and only one state has put into effect every one of them: Georgia.
Some polling forecasters, including FiveThirtyEight, have begun to factor in questions of ballot access when rendering their formulas, using resources like the Cost of Voting Index to reflect the potentially negative impact on Democratic turnout that voting restrictions tend to have. But it is hard for pollsters or forecasters to dial into the specifics of vote suppression.
In the past 10 years, Atlanta was the fourth-fastest-growing metropolitan region in the country, but polling places have not been added to keep up with the growth in population; in many places, poll sites have in fact been closed or become overrun when voting machines malfunctioned. As a result, hourslong lines became a symbol of the state’s political dysfunction in the 2018 election.
In the Georgia primary in June, with the coronavirus causing a surge in poll closures, Atlanta accounted for the vast majority — 83 percent — of all polling places statewide that were forced to stay open late because of long lines, according to an expert study. This led to lower turnout in these areas, the study found, as some voters simply gave up waiting.
Statewide, hourlong wait times were 10 times as likely to occur in heavily Black areas compared to heavily white ones, according to the study.
There’s also a possibility that many voters who think they are registered will find that their names have been purged from the rolls. The 2017 purge included over 100,000 people who would have otherwise still been eligible to vote, according to one analysis.
As an election draws near, pollsters move toward using what is known as a “likely voter” model to predict who will show up on Election Day, often depending on respondents to report their own likelihood to vote, and on their past voting behavior. But when voters who fully intended to cast a ballot find themselves unable to do so, this introduces an element of entropy not easily detected in polls.