Safe States, Silent Voters: A Six-Cycle Analysis of Partisan Predictability and Voter Turnout in U.S. Presidential Elections (2004–2024)
Revised May 2026 · An OLS analysis across six presidential cycles, an evaluation of the National Popular Vote Interstate Compact, and the case that ~89 million non-votes in 2024 reflect structural disenfranchisement, not apathy.
Cody L. Hall
This study examines whether state-level electoral predictability, commonly described as a state being “safely” Democratic or Republican, is associated with lower voter participation. Using certified Federal Election Commission (FEC) vote totals and Voting-Eligible Population (VEP) turnout estimates from the United States Elections Project, we estimate Ordinary Least Squares (OLS) regressions of VEP turnout on absolute partisan lean for all 50 states plus the District of Columbia across six consecutive presidential elections (2004 to 2024). In every cycle the relationship is negative, statistically significant (p < 0.001), and explains between 41% and 55% of the cross-sectional variance in state-level turnout (R² = 0.410 to 0.553). The regression slope is strikingly stable across cycles, averaging approximately −0.41 percentage points of turnout per unit of absolute partisan lean. These findings are consistent with a large body of peer-reviewed research showing that electoral closeness and campaign attention raise participation (Blais, 2006; Geys, 2006; Cox and Munger, 1989; Gimpel et al., 2007; Enos and Fowler, 2018). We situate the result within rational-choice theories of abstention, demonstrate its structural persistence across candidates and national environments, and argue that the pattern constitutes empirical evidence of a participation cost imposed by the winner-take-all geography of the Electoral College. We close by evaluating the National Popular Vote Interstate Compact (NPVIC) as a proposed remedy, presenting both the case advanced by its proponents and the principal scholarly objections.