open Secondary menu

Potential Impacts of Extended Advance Voting on Voter Turnout

The Impact of Advance Voting, Holiday Voting and Two Consecutive Regular Days of Voting on Turnout: A Comparative Re-examination

In a study prepared for Elections Canada, Blais et al. (2003) examined why voter turnout is higher (or lower) in some countries than in others. The purpose of the study was to determine the impact of certain institutional variables on electoral participation, controlling for other important socio-economic and geographic factors. In particular, the authors examined the impact of holiday voting and administrative measures making voting easier. However, they did not directly address the issue of advance voting and number of voting days.

We begin by describing the initial study's methodology and main findings. Then we re-analyze the initial data set taken from Blais et al. (2003), using the same methodology, while improving some measures in order to directly address the questions highlighted above.

The study covered a total of 151 elections, held in 61 democratic countries, from 1990 until 2001 (for more details, see Blais et al. 2003).Footnote 8 Three sets of variables that may affect turnout were distinguished: the socio-economic and geographic environment and two types of institutional factors (macro-institutional and electoral administration).

The authors proceeded in three stages. They started with the first set of variables and examined which socio-economic and geographic factors influence turnout. They then explored the impact of macro-institutions related to the voting system and compulsory voting, controlling for the socio-economic and geographic environment. Finally, they considered the micro-institutional level of administrative measures and their independent effects on electoral participation. At this last stage, they examined whether turnout is higher in countries where registration is compulsory, where it is possible to register on the day of the election, where the government is responsible for taking the initiative to register electors, or where the electoral legislation facilitates the exercise of the right to vote (by allowing voting by mail, by proxy or in advance, and by holding the vote on a holiday).

Table 1 presents the results.Footnote 9 They show the relationship between turnout and electoral administration variables once the effects of socio-economic and geographic, as well as macro- institutional, variables have been neutralized. Two results are more directly relevant here. First, the authors find no independent impact of holiday voting. Second, they report a strong effect resulting from administrative measures designed to facilitate voting.

Table 1: Determinants of Voter Turnout: Socio-economic, Geographic
and Macro- and Micro-institutional Factors
Independent variables Dependent variable:
Turnout by number of people registered on the lists of electors
OLS Regression coefficient (Error)
Eastern Europe 0.18 (3.85)
Population (log) -3.34*** (1.22)
DGP per capita (log) 8.28*** (2.93)
Compulsory voting with penalties 13.28*** (2.80)
PR and mixed compensatory 4.20* (2.19)
Compulsory registration 3.79 (2.45)
Registration on polling day -4.80 (3.07)
Government initiative 0.85 (2.43)
Holiday voting -0.26 (2.72)
Easy voting 11.04** (4.35)
Mali -33.42*** (11.14)
Constant 55.54*** (13.74)

Number of cases 119
Adjusted R2 0.46


* significant at 0.10 (two-tailed test)
** significant at 0.05 (two-tailed test)
*** significant at 0.01 (two-tailed test)

Source: Blais et al. 2003

However, since a single variable was created (named Easy voting), encompassing three administrative measures designed to making voting easier,Footnote 10 these findings do not allow us to determine if advance voting as such has an independent effect on turnout. Neither do we know whether having two consecutive regular days of voting makes a difference. Below, we provide a re-analysis of the initial data, which allows us to address these questions directly.

We expand the initial analysis in three ways. First, we look separately at the impact of advance voting on electoral participation. We take apart the variable Easy voting and create three more specific measures: Proxy voting, advance voting and POSTAL VOTING. Second, we add a new variable, named Two consecutive regular days of voting,Footnote 11 to determine whether turnout is higher in countries where there are two consecutive regular days of voting. Finally, we investigate whether turnout is higher in countries where voting takes place on a holiday. We refine the operationalization of the variable in the present analysis.Footnote 12

The new findings are presented in Table 2. In countries where the electoral legislation makes it easier to vote by allowing advance voting, turnout is about 4 percentage points higher than in countries where such an option does not exist. Likewise, the presence of proxy voting enhances electoral participation by about 5 percentage points. Postal voting, on the contrary, does not seem to have a statistically significant effect. We should note that each variable has the predicted positive coefficient but that in all cases the results are not very robust. Advance and proxy voting are significant only at the 10-percent level, while postal voting does not quite reach statistical significance. When we combine the three measures into one, as was done in Table 1, the results are more significant.

Table 2: Re-analysis of Determinants of Voter Turnout: Socio-economic, Geographic and Macro- and Micro-institutional Factors
Independent variables Dependent variable:
Turnout by number of people registered on the lists of electors
OLS Regression coefficient (Error)
Eastern Europe -1.21 (3.95)
Population (log) -3.55** (1.35)
DGP per capita (log) 8.68*** (2.96)
Compulsory voting with penalties 12.26*** (2.85)
PR and mixed compensatory 2.81 (2.32)
Compulsory registration 3.70 (2.53)
Registration on polling day -5.11 (3.10)
Government initiative 1.19 (2.53)
Holiday voting 2.59 (2.93)
Advance voting 3.98* (2.39)
Postal voting 3.07 (2.49)
Proxy voting 4.89* (2.73)
Two consecutive regular days of voting 7.39 (5.56)
Mali -35.89*** (11.52)
Constant 54.27*** (13.74)

Number of cases 119
Adjusted R2 0.45

* significant at 0.10 (two-tailed test)
** significant at 0.05 (two-tailed test)
*** significant at 0.01 (two-tailed test)

The new estimation still reveals no significant impact of holiday voting, though we should note that the coefficient now has the predicted positive sign. The same pattern emerges with respect to two consecutive regular days of voting. We must keep in mind, however, that only six elections (out of 119 in our sample) had two consecutive regular days of voting; such a small number may not be enough to reveal any significant pattern.Footnote 13

In short, this re-analysis of cross-national variations in turnout yields somewhat ambiguous results. On the one hand, each of the three specific variables that are of direct interest here – holiday voting, advance voting, and provisions of two consecutive regular voting days – has the expected positive coefficient, but two of them do not reach the standard levels of statistical significance, and the third is significant only at the 10-percent level. Perhaps a prudent conclusion is that these data suggest that each measure may have a positive impact, but that the impact remains uncertain.


Footnote 8 The initial sample included all 74 democratic countries with populations over 100,000, to which Freedom House had given a highest score of 1 or 2 for political rights in 1996–1997 and that were included in the Massicotte et al. study on electoral legislation (2004). Nine countries had to be excluded because of missing information, and two others, Switzerland and the United States, had to be omitted because their electoral legislation varies across states or cantons.

Footnote 9 All analyses were performed using two different measures of turnout rates: turnout calculated on the basis of number of registered electors and turnout based on the population of voting age as reported in national population censuses. We have more confidence in results using the first measure, because the information is more complete. Thus, we focus here on turnout calculated on the basis of number of registered electors.

Footnote 10 Those measures were these: voting by mail, by proxy and in advance. The variable Easy voting takes the value of 0 when none of the options is available, 0.33 when only one option is available, 0.66 when two options are available, and 1.00 when all three options are available under the electoral legislation in a given country.

Footnote 11 The variable takes the value of 1.00 when election is held on two consecutive days, 0 otherwise.

Footnote 12 Initially, the variable Holiday voting indicated whether the polling day was a day of rest or a specially declared holiday, as opposed to either a weekday or a weekday and holiday together. In the present study, the variable takes the value of 1.00 whenever holiday voting is an option.

Footnote 13 The countries that allow this measure for the period covered in our study are these: Czech Republic, Slovakia and Namibia.