đź“Š What This Paper Asks:
Estimates how nonvoters would have affected U.S. presidential election outcomes from 1952–2000 by applying a propensity score reweighting technique borrowed from the economics-of-discrimination literature to National Election Study (NES) data.
🧠How Turnout Was Reconstructed:
- Uses NES survey data for the 13 presidential elections between 1952 and 2000.
- Applies a propensity score reweighting approach to simulate a full-turnout electorate by statistically reweighting observed voters to represent nonvoters.
🔎 Key Findings:
- Nonvoters are, on average, slightly more likely to support the Democratic Party.
- Across the 13 elections studied, reweighting produces no change in the ultimate winner in all but two possible exceptions: 1980 and 2000.
- A modest increase in turnout—on the order of two percentage points—can be decisive when elections are very close.
- Compulsory or substantially higher turnout would not, in most cases, radically alter the overall partisan distribution of the vote.
⚠️ Limitations and Method Comparison:
- Limitations of NES data likely lead to an underestimate of the true impact of nonparticipation.
- The propensity score reweighting approach is compared with other econometric techniques to assess robustness and relative strengths.
đź’ˇ Why This Matters:
- Results align with much previous research on participation: expanding turnout changes vote margins but usually does not overturn most presidential outcomes.
- Findings are used to offer informed speculation on why the Democratic Party has not pursued broader get-out-the-vote or registration drives on a national scale, given the limited partisan payoff implied by the estimates.
