Site MenuPotential Impacts of Extended Advance Voting on Voter Turnout
The Canadian Evidence: Does Increased Advance Voting in an Electoral District Increase Overall Turnout in that District?
Advance voting could increase overall turnout in at least two ways. First, the ability to vote on more than one day increases the ease of voting; citizens have a greater range of options about when they can cast their vote. Electors' cost of voting is reduced, and this should boost turnout. Second, advance voting creates additional opportunities for parties and candidates to mobilize voters. Because voting is extended over a number of days, local campaign organizations are given more opportunities to bring voters to the polls.
One way to ascertain the impact of advance voting is to look at electoral districts and determine whether there is a correlation between advance voting and turnout. However, the relationship between overall turnout and advance voting cannot be determined by just observing their correlation across electoral districts, because the same factors that theoretically increase election day voting should also increase advance voting. For example, electoral districts with higher mean levels of age could be expected to have higher turnout of both types. Similarly, competitive electoral districts should have higher turnout of both types than non-competitive electoral districts. What is needed is a method that controls for these confounding factors. We can hold these factors constant by looking at the relationship between increases in advance voting and overall turnout. If greater advance voting causes higher overall turnout, then electoral districts that experienced more advance voting in 2006 than 2004 should also have higher overall turnout.
We perform just such an analysis using data from each electoral district from the 2006 federal election. We examine the relationship between election day turnout in each constituency in the 2006 federal election and three variables:
- the difference in the number of those who voted at advance polls as a share of all voters between 2006 and 2004
- election day turnout in 2004
- the closeness of the race in 2006
All other variables, such as the socio-demographic composition of the constituency, are held constant, since constituency boundaries did not change between 2004 and 2006.
Table 4 presents the results when we regress election day turnout on change in advance voting, controlling for the closeness of the race and election day turnout in 2004. To ascertain the effect of increases in advance voting on overall turnout, we examine the coefficient measuring the impact of advance voting increase.
| Independent variables | Dependent variable: |
|
|---|---|---|
| OLS Regression coefficient | (Error) | |
| Election day turnout, 2004 | 0.70*** | (0.03) |
| Advance voting increase, 2006 | -0.44*** | (0.12) |
| Margin of victory, 2006 | -0.03*** | (0.01) |
| Constant | 21.24*** | (1.66) |
Number of cases 308
R2 0.66
* significant at 0.10 (two-tailed test)
** significant at 0.05 (two-tailed test)
*** significant at 0.01 (two-tailed test)
The results suggest that for every percentage point increase in advance voting in 2006 over 2004, election day turnout was 0.44 percentage points lower, taking into account election day turnout in 2004. Since total turnout is the sum of advance voting turnout plus election day turnout, the implication is a net positive effect of 0.56 percentage points (1 – .44). In other words, an electoral district that went from 5-percent advance voting in 2004 to 10-percent in 2006 is predicted to experience a 2.2-point (5 x .44) decrease in election day voting, but an overall increase in total turnout of approximately 2.8 percentage points (5.0 – 2.2). However, the average increase in advance voting was much smaller than this. In 2006, the average electoral district realized an increase in advance voting of 1.3 percentage points, compared with 2004. The overall effect of increased advance voting (between 2004 and 2006) on overall turnout was then, on average, just 0.7 percentage points. While this is a significant effect, it is also a modest one.