The Old Models, Revisited

Traditional turnout models were built on a relatively stable electoral universe. They assumed that older voters would vote at consistently high rates, that younger voters would turn out at lower rates, and that midterms would draw a different electorate than presidential years. Those assumptions worked well in an era when party affiliation and voter behavior were predictable. The 2022 midterms already delivered a warning shot. Turnout among voters under 35 surged well beyond historical norms in several key states, and it was disproportionately concentrated in the first two weeks of early voting — a window that traditional GOTV models were not designed to capture. The 2024 presidential cycle then shattered records across nearly every demographic group, raising a fundamental question: what does “normal” turnout even mean anymore?

The composition of the eligible electorate has changed dramatically since the last midterm cycle. The baent years. These shifts are not marginal adjustments; they represent a fundamental reweighting of the electorate. Beyond sheer numbers, behavioral patterns have diverged from historical norms in ways that complicate projection models. Mobile voters — those who change addresses frequently — now represent a significantly larger share of the electorate than they did in 2014 or 2016. Many of these voters are quietly dropped from rolls between cycles, often without their knowledge, and finding them again requires entirely different data strategies than the door-knocking and phone-banking operations that defined past campaigns.

Written by Amara Osei, Africa Bureau Chief