- Figure 5.5. Predicting Change in the Invalid Vote Using Features of Null Vote Campaigns Source: Original data collection, electoral management bodies. Note: The figure shows coefficient estimates from an OLS regression model regressing change in the invalid vote compared to a recent baseline election on indicator variables for each campaign attribute (N = 36). The figure shows 90% confidence intervals, with 80% confidence intervals (two-tailed) at the vertical hash marks. I control for year, and standard errors are clustered at the country level.
|