Impact of electoral volatility and party replacement on voter turnout levels

Abstract

While elections are viewed as the lynchpin of modern democracies, few works have adequately assessed the role played by political parties in mobilizing voters. Much of the extant work has relied on the number of parties in a party system to estimate the impact on voter turnout; not surprisingly, the voluminous literature on voter turnout has arrived at a theoretical impasse regarding the relationship between party systems and voter turnout. We argue that in order to better understand the relationship between party systems and voter turnout, researchers should consider other relevant party system measures. In particular, several scholars have surmised that party system stability holds numerous implications for democracies, but there has yet to be an empirical analysis of this claim. In this study, we anticipate that lower volatility and replacement rates – both indicating more stable party systems – should have a positive impact on aggregate turnout. Even when including several control variables, the results of our cross-sectional time-series analyses confirm our hypotheses.

Keywords

elections electoral volatility party systems party system institutionalization stability/instability

Introduction

The role of party systems in the electoral process has been well documented. Despite the burgeoning literature on voter turnout, however, scholars have arrived at a theoretical crossroads in discussing the importance of party systems to electoral participation. Some researchers find that larger or multiparty party systems increase turnout levels (Blais and Carty, 1990; Blais and Dobrzynska, 1998; Franklin, 1999; Gosnell, 1930; Jackman and Miller, 1995; Ladner and Milner, 1999; Powell, 1980, 1986; Tingsten, 1937), while others discover that multiparty systems may be inversely related to turnout rates (Fornos et al., 2004; Kostadinova, 2003; Kostadinova and Power, 2007). In these studies, researchers have typically used the number (or effective number) of parties in aggregate analyses to gauge the impact of party systems on the polity.

Those arguing that larger party systems correspond with lower levels of turnout have generally emphasized the higher informational costs that arise from multiparty systems. In contrast, in those studies that find multiparty systems increase turnout, the rationale commonly given relies on the fact that more parties can appeal to broader swathes of the population. Therefore, voter turnout appears to be primarily contingent upon informational costs and ties with party systems. If this is the case, simply looking at the number of parties within a system may provide inadequate information or evidence to capture the full effect of party systems on political participation; it is therefore useful to consider other relevant party system attributes that may also influence electoral participation (Mainwaring, 1999; Mainwaring and Scully, 1995; Sartori, 1976). We contend that the number of parties, while laudable, only captures a portion of party systems' importance vis-à-vis political participation, and thus we analyse how electoral volatility and the emergence of new parties (or party replacement) – two standard measures of party system institutionalization (PSI) (Mainwaring, 1999; Mainwaring and Scully, 1995) – impact turnout levels.¹ Specifically, we argue that stable party systems, characterized by lower volatility and party replacement levels, should be positively related to turnout levels.

This article therefore offers two primary contributions to the continuing debate on electoral participation. First, while several studies posit that party system stability influences voter turnout rates (Cox, 1999; Crewe, 1981; Powell, 1980, 1986), there has yet to be a systematic analysis of this claim. In addition to these theoretical arguments, other studies provide anecdotal evidence to support the idea that voter turnout is conditioned by party system stability (Aldrich, 1993; Birch, 2005; Bratton, 1999; Farrell and Webb, 2002; Marsh, 2004; Polhemus, 1983; Rosenstone and Hansen, 1993).² And while recent works by Hankla (2006) and Robbins (2010) have conjectured that party system stability has numerous implications for policymaking, few scholars have examined how party system stability may impact political participation.

The second primary contribution of this article is its scope. While quite sophisticated, other macro-level analyses have often relied on one country or region when studying turnout levels.³ Previous studies have examined turnout levels in advanced industrial democracies (Jackman, 1987; Jackman and Miller, 1995), Latin America (Fornos et al., 2004), former Communist states (Kostadinova, 2003) and Africa (Bratton, 1999; Kuenzi and Lambright, 2007; Polhemus, 1983). However, most of these studies focus on only one region and thus questions remain as to the accuracy of these theories in a broader, cross-national setting. This is especially apparent when we review turnout levels by region (see Table 1). We see that both volatility and party replacement levels display wide variations across the different regions from the information presented in Table 1. With a broader sample, the results should offer a more comprehensive view of both volatility and party replacement’s effect on voter turnout.

Table 1.

Turnout, Volatility and Replacement by Region

	Parliamentary Turnout ^a	Volatility	Replacement
OECD	76.99%	11.94	9.68
Latin America	67.52%	17.55	18.22
Former Communist	68.81%	27.94	38.39
Asia	66.80%	16.17	25.70
Africa	73.51%	9.97	6.46

^a Information on turnout levels was gathered from the International Institute for Democracy and Electoral Assistance. (2008).

Party system stability’s role in turnout levels

Parties and party systems are the primary mechanism through which citizens' demands are funnelled into the policymaking process. Without these agents, citizens' voices may be inaudible and rendered virtually meaningless. Thus, we believe that party systems serve at least two significant functions that are germane to this discussion. First, we suspect that unstable party systems hinder voters' ability to utilize informational short-cuts at the ballot box. Downs (1957) argues that a large number of individuals form political opinions and decide to vote based on information acquired by accident or chance. Therefore, citizens with a better understanding of parties and party systems are more likely to vote (e.g., Aldrich, 1993). Recent work tests this theory and discovers that those individuals who can readily distinguish between parties' policy platforms are more likely to vote (Goodman and Murray, 2007).

However, party labels are not constant across all countries. In some developing states, such as in Latin America or among former Communist states, these labels are hardly stable. As can be seen in Table 1, both volatility and replacement levels remain high, which means that these systems do not convey the same consistent messages to would-be voters as is seen in advanced industrial democracies. Downs (1957) conceptualizes voter turnout in terms of the costs versus the benefits presented to citizens; individuals are predicted to vote as the expected benefits outweigh the costs. According to the model, later extended by Riker and Ordeshook (1968) and Aldrich (1993), when the costs of voting are high, individuals will abstain from voting. Similarly, Aldrich (1993) posits that where informational costs are relatively low, turnout is more likely. He notes, further, that parties can enhance citizens' knowledge of candidates through their various functions. The same logic can be extended to party systems and their ability to convey certain information to voters; hence, we expect that the higher informational costs stemming from unstable party systems will result in a decrease in voter turnout levels.

The second reason stable party systems should engender higher turnout has to do with their effectiveness in mobilizing voters (Aldrich, 1993; Bratton, 1999; Caldeira et al., 1990; Karp and Banducci, 2007; Kramer, 1970; Kuenzi and Lambright, 2007; Marsh, 2004; Powell, 1980, 1986; Rosenstone and Hansen, 1993; Vowles, 2002; Wattenberg, 2002). Generally speaking, party systems that more effectively reach out to voters are linked with higher turnout levels since they are inclined to exert considerable effort mobilizing would-be voters (Karp and Banducci, 2007). Predictably, higher voter turnout levels in many advanced democracies have been attributed to the party systems' ability to encourage greater participation among the electorate. Indeed, many Western European party systems continue to evince greater skill in conducting campaigns (Farrell and Webb, 2002), but party systems in developing democracies are less organized and lack the essential resources that can foment mobilization efforts (Birch, 2005). Similar conclusions have been drawn about the relationship between voter turnout and party systems elsewhere (Bratton, 1999; Kuenzi and Lambright, 2007; Polhemus, 1983; Powell, 1980, 1986; Vowles, 2002; Wattenberg, 2002).

Electoral volatility, party replacement and voter turnout

For advanced industrial democracies the number of parties may reflect stable representation, but in many developing states the number of parties is not an accurate signal of party system stability (Mainwaring, 1999). In other words, the number of parties may be similar in both settings, but this does not ensure the same level of stability across both developing and advanced democracies. For instance, in our dataset, the mean value for the effective number of parties in the United States and Albania is similar at roughly two effective parties. However, the volatility rates exhibit tremendous variation: the average volatility rate in Albania is nearly 30 percent compared with an average volatility rate of only 3 percent for the United States. Thus, counting the number of parties, or the number of coalition parties, or calculating the relative weight of these parties completely ignores consistency in representation, and one way to address this issue is to rely on other variables that can better approximate party system stability, such as volatility and party replacement.

Institutionalized systems are characterized by their legitimacy, relatively low levels of electoral volatility or turnover, established internal organizations and ties with voters or social groups.⁴ Furthermore, stable or institutionalized party systems are essential to democratic consolidation for several reasons, but their most attractive feature stems from the level of accountability found therein. Also, these systems should be more apt to quell uncertainty in political outcomes given the greater constancy in support levels. Previous work has noted that stable party systems maintain connections with social and/or political groups (Hankla, 2006; Robbins, 2010). Furthermore, internal organizations that are better organized and can disseminate information are also inherent in institutionalized systems. A relevant example comes from the Russian party system.⁵ This system has been, and continues to be, characterized by the absence of ideologically cohesive parties (Smyth, 2006). The relevance for voters is that if their elected officials have ambiguous policy platforms, voters' ability to evaluate candidates may be impaired.

The final component – stability in inter-party competition – should be the end-result of these criteria, and thus systemic measures can approximate the presence of these attributes. It is therefore reasonable to conclude that while systemic measures such as volatility and party replacement may not directly tap into each of Mainwaring’s components, they do reflect the existence of these items either directly or indirectly. Volatility levels, for instance, should be higher where party systems lack ties with social groups and/or have fluid internal structures, since consistent electoral support is lacking. If party systems rely on stable support from social groups, then electoral volatility should be dramatically lower than where party systems lack these ties. And, without internal organizations that can disseminate information and solicit support for its members, a party system’s electoral base should be more fluid. As these characteristics of individual parties are aggregated, we can gain a better picture of the party system’s stability.

To measure the relationship between voter turnout and party system stability, we therefore rely on electoral volatility and party replacement, as these variables have been used extensively in previous work (Birch, 2001; Birnir, 2005; Dix, 1992; Ishiyama, 2007; Kuenzi and Lambright, 2001, 2007; Mainwaring, 1998, 1999; Mainwaring and Scully, 1995; Roberts and Wibbels, 1999; Stockton, 2001; Tavits, 2005, 2006). Electoral volatility represents the aggregate change in party vote-shares from one election to the next. Here, we are able to see how parties' vote-share changes from one election to the next. If there are large fluctuations in volatility, then it is reasonable to conclude that voters are weakly attached to the party system and that party organizations are struggling to proselytize voters. And if party systems sustain voter attachments and can convincingly solicit support from would-be voters, turnout levels should be noticeably higher. Thus, high volatility levels are indicative of inchoate party systems. Formally stated, we hypothesize that:

Hypothesis 1: Volatility levels should be inversely related to turnout levels.

The second hypothesis also measures systemic changes in vote-shares. While volatility is useful in many ways, it does not measure the entrance of new actors.⁶ This is of particular concern since new parties frequently emerge in states with amorphous or incipient party systems and is certainly the case for many post-communist states. Although there are issues with measuring new party entry (Selb and Pituctin, 2010; Sikk, 2005), it has become a laudable complement to volatility. Here, we include the percentage of votes won by new parties.⁷ The theoretical relevance is much the same as volatility and therefore we anticipate that:

Hypothesis 2: Party replacement levels should be inversely related to turnout levels.

Data and methods

In order to test the hypothesis that volatility and party replacement influence voter turnout levels, a large-N pooled cross-sectional time-series approach is employed. For this analysis, we utilize a dataset that includes numerous democracies from Africa, Asia, the OECD, East Europe and Latin America. In order to be consistent with previous studies on turnout, a pooled cross-sectional approach is used.⁸ The countries included reached at least a ‘6’ on the Polity2 measure found in the Polity IV database (2000).⁹ A list of the countries and years in the analysis can be found in the appendix. The data collected span 1975–2004, except for states that were not democratic for the entire period (e.g. former Communist states).

Dependent variable: Voter turnout

The dependent variable in this study is the percentage of eligible registered voters who voted in a legislative election (Blais and Dobrzynska, 1998; Franklin, 1999; Kostadinova, 2003).¹⁰ Our analysis focuses on legislative elections for multiple reasons. First, with very few exceptions, most studies have also emphasized turnout in legislative elections (Blais and Dobrzynska, 1998; Kostadinova, 2003; Kostadinova, and Power 2007; Powell and Whitten, 1993). Second, the number of presidential elections pales in comparison to the number of legislative contests. We ran models for both types of contests, but lost numerous observations when we ran presidential races separately. Finally, there are also coding difficulties with some presidential regimes, such as: should researchers include systems where the president is indirectly elected? If not, how should we code these contests? Given these limitations, our study calibrates turnout as the percentage of registered voters who voted in legislative elections. Our turnout data were gathered from the International Institute for Democracy and Electoral Assistance’s (IDEA, 2008b) website.

It is also worth noting that the sequence of elections may have an important effect on turnout levels, whereby founding elections may witness especially high turnout levels. Therefore, it is possible that founding elections – an issue especially relevant in post-communist and Latin American states – may alter the results of these findings. Fortunately, this is not a serious threat to this study, since both volatility and replacement levels (discussed below) are based on electoral returns from two consecutive elections. Because of this calculation, founding elections are omitted from our analyses.

Measuring volatility and party replacement

As discussed above, Mainwaring’s work alleges that PSI or party system stability is a multifaceted concept. To that extent, we assess the impact of electoral volatility and party replacement as two key indicators for party system stability. And, to be clear, our study focuses on party systems not individual parties. Although looking at individual parties could be an interesting approach, previous studies on institutionalization have relied upon systemic measures, and thus we follow this same path (Birch, 2001; Birnir, 2005; Casas-Zamora, 2005; Ishiyama, 2007; Kuenzi and Lambright, 2001; Mainwaring and Scully, 1995; Mainwaring and Zoco, 2007; Przeworski, 1975; Roberts and Wibbels, 1999; Stockton, 2001).

Electoral volatility is measured as the absolute change in vote-share among those parties that have participated in consecutive elections. The summed change is then divided by the total vote-share accrued by those same parties in consecutive elections. The volatility calculation can be expressed formally as:

V = \frac{Σ |c_{i, t + 1} - c_{i, t}|}{Σ c_{i, t + 1} + Σ c_{i, t}}

Where V denotes the aggregate volatility and c_i the vote-share for party i at election t; c_i , _t+1 indicates the vote share of party i at the second, or next, election. To be clear, this measure is used instead of the Pedersen’s volatility score, as it permits more variance and provides a more accurate measure of change in volatility.

The range in volatility runs from the least volatile state, the United States (average volatility: 2.78 percent), to the most volatile state, Latvia (average volatility: 46.64 percent). These figures were collected by the authors who took advantage of various sources. For post-communist data, the University of Essex provides electoral returns for these states, whereas data for Africa are gathered from the African Elections Database (2007). The Political Database of the Americas (2007) has information on Latin American elections. Election results for the remaining countries are available through other sites (e.g. Adam Carr’s website, ElectionWorld and electoral commissions from various states).¹¹ Based on the arguments above, electoral volatility should inversely affect turnout levels (Hypothesis 1).

Though useful, volatility alone cannot fully account for the influence of new parties. Therefore, party replacement is the second systemic measure used in this study. Previous work has tabulated the number of new parties entering a system (Birnir, 2005; Tavits, 2006), while others have calculated the total vote-share won by new parties (Birch, 2001). The number of new parties is a useful initial attempt to study party systems, but simply counting the number of parties does not provide accurate information as to the electoral weight attributed to new parties. Because of this, we measure party replacement as the total percentage of votes won by new parties and we predict that higher replacement rates should depress turnout levels.

It is possible – though unlikely – that our models are potentially plagued by the endogenous relationship between our primary institutionalization variables and turnout levels; therefore, multiple provisions have been undertaken to ameliorate this potential threat. First, we ran a Durbin–Wu–Hausman test to detect if this was a legitimate concern. The results indicated that such a problem was not affecting the statistical tests, and thus not skewing the analyses.¹² Second, we use lagged measures of volatility and party replacement in the models below.¹³

Additional controls

In line with the previous literature, we also include a set of standard economic, political and social controls: concurrent elections, compulsory systems,¹⁴ GDP per capita, change in GDP per capita, inflation, unemployment, district magnitude, effective number of parties (ENP), federalism, presidential systems, Polity2, total population and percent urban population (Aguilar and Pacek, 2000; Almond and Verba, 1963; Blais and Dobrzynska, 1998; Fornos et al., 2004; Franklin, 1999; Gray and Caul, 2000; Inglehart, 1988; Jackman, 1987; Jackman and Miller, 1995; Kostadinova, 2003; Kostadinova and Power, 2007; Kuenzi and Lambright, 2007; Norris, 2004; Pacek and Radcliff, 1995b; Powell, 1986; Powell and Whitten, 1993; Power and Roberts, 1995; Putnam, 1993; Remmer, 1991). Concurrent elections and states with compulsory voting laws are calibrated with binary indicators and both were taken from IDEA’s website (2008a). Unless stated otherwise, the economic controls were obtained from the World Bank’s WDI (2007) database, while the institutional measures were taken from Beck et al.’s (2004) Database of Political Institutions. Each economic variable was lagged a year, as is commonly seen. The total population (in millions) metric was obtained from the Correlates of War project (COW Dataset, 2008).¹⁵ Finally, we also include a trichotomous indicator for federalist states that is taken from Wibbels' (2000) coding scheme. Table 2 presents the summary statistics for the variables used in this study.

Table 2.

Summary Statistics of Variables Used

Variable	Mean	St. Dev.	Minimum	Maximum
Turnout	73.46	14.86	21	98.9
Volatility	15.06	12.02	.45	87.12
Party Replacement	15.81	20.98	0	99.9
Concurrent Elections	.25	.43	0	1
Compulsory Voting	.41	.49	0	1
Bicameral	.56	.50	0	1
President	.40	.49	0	1
Federalism	1.07	.77	0	2
District Magnitude	17.99	34.08	1	150
Effective Number of Parties	3.46	1.73	1	13.18
GDP per capita_t-1	9,448.62	9,659.71	128.68	52,182.86
Δ GDP per capita	193.11	376.95	−1,847.77	3,010.33
Inflation _t-1	58.65	498.43	−23.48	13611.63
Unemployment	8.54	4.79	.60	35
Urban Population (%)	61.70	17.88	5.8	97.2
Total Population	42,109.89	112,022.10	218	1,013,865
Polity2	7.42	2.37	6	10

Methodology

Given the temporal element of our data, we use a pooled cross-sectional time-series estimation in order to remain consistent with previous turnout studies.¹⁶ For a number of reasons, we use a regression analysis with random effects and panel-corrected standard errors (Beck and Katz, 2004). Because we want to gauge country-to-country variations, a random effects model is preferred (Beck and Katz 2004, 2006). In addition, because there are a number of time invarying covariates in our models (e.g. compulsory voting systems, district magnitude, bicameral systems, etc.), a random effects estimation is better suited for this study (2004, 2006).¹⁷ We have run additional estimations that should mitigate any concerns that our findings are driven by particular cases. Specifically, we have tested, though do not report, our models with a series of country-specific measures to simulate a fixed-effects model (see Radcliff, 1992; Pacek and Radcliff, 1995a, b, c). Our hypotheses are still in the expected direction and statistically significant.¹⁸ Additional robustness checks are enumerated at the end of the next section.

In Table 3 we present six separate empirical models to test our hypotheses. The two primary independent variables are run separately so as to avoid any potential collinearity (which appears to be a concern based on our analyses).¹⁹ We also run separate models for presidential systems (Models 3 and 6) and we separately test both GDP per capita and change in GDP per capita as the previous literature has yet to reach a consensus on the significance of these variables.

Table 3.

Time Series Regression of Parliamentary Turnout Levels

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Volatility	−.38 (.09)***	−.35 (.07)***	−.33 (.09)***	−	−	−
Party Replacement	−	−	−	−.10 (.05)**	−.12 (.05)**	−.11 (.05)**
Concurrent Elections	2.88 (2.09)	1.12 (1.98)	3.62 (2.29)	2.99 (2.16)	1.26 (1.89)	3.80 (2.34)
Compulsory System	8.93 (1.47)***	12.31 (1.34)***	11.11 (2.10)***	9.76 (1.59)***	12.30 (1.37)***	11.45 (2.12)***
Bicameral	−1.89 (2.11)	−7.07 (1.61)***	−3.91 (1.91)**	−1.65 (2.19)	−6.69 (1.67)***	−3.39 (2.08)
Federalist	5.46 (1.38)***	5.77 (1.24)***	3.15 (1.22)**	5.32 (1.37)***	5.34 (1.33)***	2.68 (1.34)**
District Magnitude	.00 (.02)	.02 (.01)	−.03 (.02)	.01 (.02)	.02 (.01)	−.03 (.02)
President	−	−	−5.53 (2.77)**	−	−	−6.13 (2.86)**
ENP	1.06 (.58)*	1.06 (.50)**	.45 (.62)	.95 (.63)	.80 (.53)	.32 (.68)
GDP Per Capita _t-1	−	−.00 (.00)***	−.00 (.00)	−	−.00 (.00)*	−.00 (.00)
Δ GDP Per Capita	−.00 (.00)	−.00 (.00)	.00 (.00)	−.00 (.00)	.00 (.00)	.00 (.00)
Inflation_t-1	.26 (.75)	.13 (.92)	−.20 (.78)	−.04 (.87)	−.01 (1.01)	−.53 (.84)
Unemployment Pct_.t-1	−.32 (.15)**	−.23 (.17)	−.45 (.18)**	−.44 (.18)**	−.30 (.19)	−.55 (.19)**
Urban Population	−	.20 (.05)***	.20 (.06)***	−	.24 (.06)***	.24 (.07)***
Total Population	−.00 (.00)***	−	−.00 (.00)*	−.00 (.00)***	−	−.00 (.00)
Polity2	3.57 (1.01)***	4.83 (1.14)***	3.19 (1.13)***	3.88 (.99)***	4.78 (1.10)***	3.44 (1.12)***
Groups	39	39	38	39	39	38
Observations	157	174	155	157	173	155
R²	.55	.52	.61	.51	.55	.59
Probability X ²	.00***	.00***	.00***	.00***	.00***	.00***

*p<.10; **p<.05; ***p<.01, one-tailed tests; Panel-corrected standard errors in parentheses

Table 4.

Subsequent Changes in Turnout Levels

	Volatility	Replacement
Minimum	75.73%	74.60%
Mean	73.52%	73.20%
Maximum	65.78%	67.28%

Discussion and results

The results of our analysis are provided in Table 3. The tables report our results of the relationship between voter turnout and electoral volatility and party replacement. As expected, both measures are significant, negative and affirm our hypotheses. Furthermore, each model is statistically significant, revealing that these variables jointly influence turnout levels.

Do stable party systems play a role in turnout levels? Clearly, the answer is yes. We see that both volatility and party replacement measures are statistically significant and verify our hypotheses. Volatility levels (Models 1–3) are in the expected direction and exert a sizeable effect on turnout. That is, more volatile systems correspond with lower turnout levels as every one unit increase in volatility results in a drop in turnout of 0.38 (Model 1). Another way to interpret this finding is to consider the minimum, mean and maximum volatility values and the subsequent effect on aggregate participation.²⁰ At its minimum value, volatility is associated with a turnout level of nearly 76 percent. In contrast, the most volatile systems generated a sizeable decline in turnout (around 66 percent). Therefore, there is a 10 percent point difference between the most volatile systems compared with the most stable ones. The effect of replacement levels echoes this finding. The most fluid systems depress voter turnout by a notable margin. For each increase in party replacement (or, as new parties displace older parties by a larger margin) turnout levels fall by 0.10 (Model 4). At its mean, party replacement yields turnout rates of approximately 73 percent. Where the vote-share won by new parties is negligible, turnout levels are higher (74.6 percent). Predictably, in the most unstable systems – when replacement is set at its maximum value – turnout drops to 67.3 percent. Therefore, turnout levels are impacted by replacement levels, since there is a difference of roughly 8 percent between the most stable party systems and the least. Based on the theoretical expectations explicated earlier, these results are unsurprising. In countries where parties are weakly rooted in society, frequently appear/disappear from national politics, and lack the internal structures that could facilitate voter mobilization efforts, turnout levels appear to suffer even when controlling for other variables. We have also considered changes in turnout levels when volatility and party replacement levels are set to 1 standard deviation above and below their mean values. When volatility is set to 1 standard deviation below its mean, turnout levels are approximately 71 percent. Yet, when volatility is set to standard deviation above the mean, turnout levels fall to 65 percent. Similar findings emerge when we consider changes in party replacement levels. Turnout levels vary based on changes in party replacement levels: when set to 1 standard deviation below, the mean turnout levels are 70 percent and when set to 1 standard deviation above the mean turnout levels are 66 percent.

This finding suggests that the importance of party systems goes beyond simply counting the number of parties. It reveals that volatility and party replacement are significant predictors as well. These proxies of party system stability show that voters may be influenced by more than just the sheer number of parties. In countries beleaguered by high volatility and/or party replacement levels, voters are more likely to refrain from voting. This finding has numerous implications for the study of party system stability and its relationship with democracy.

Other results are of interest as well. The two dichotomous measures for bicameralism and compulsory voting systems comport with previous literature: bicameral legislatures significantly depress turnout (Models 2, 3 and 5), while more citizens flock to the polls when voting is mandatory. The other four institutional variables, however, failed to reach statistical significance. Perhaps most telling is the insignificance of district magnitude. Previous studies have indicated that district size significantly shapes turnout levels, yet Table 3 does not support the conventional wisdom. Meanwhile, for concurrent elections the effect was correctly signed, though insignificant. Furthermore, institutional studies have also demonstrated that multiparty systems tend to result in lower turnout levels, but our findings are incongruous with the extant literature. Since we have tested for collinearity and find that this is not a threat to our models, we are certain that these results are not driven by this pathology. We also control for executive type and federal systems. The former reveals a significant and negative effect on the dependent variable, while the federalist measure is positive and significantly related to turnout.

The empirical tests have also gauged the influence that macro-economic indicators have on electoral participation. These findings are not entirely consistent with previous work. Although unemployment rates significantly affect turnout, the effect is not robust across our models. The direction of the coefficient also indicates that turnout rates decrease as unemployment rates increase, though these findings are not consistent in all models. As such, our findings reveal that unemployment rates are the most consistent predictor of participation levels in parliamentary contests.

The remaining measures used were the total population, percent urban population and the Polity2 indicator. Because other measures of a country’s intellect (e.g. literacy rates, newspapers per capita, etc.) were unavailable, we have relied on the urban population measure and find that the variable is in the expected direction. Driven by the assumption that urbanites are more politically savvy than their rural counterparts, other studies have relied on this measure to approximate the influence that cultural effects have on turnout (Fornos et al., 2004; Kostadinova, 2003; Pacek et al., 2009). Conversely, the total population metric is inversely related to turnout (Models 1, 3 and 4). Finally, we also discover that development does matter, as Polity2 ratings are significantly correlated with higher participation.

Robustness checks

Aside from testing for collinearity, we also evaluated the potential distorting effects of any outliers and re-estimated our models using two additional approaches. To detect the presence of outliers, we use Cook’s D test to assess the leverage exerted by potential outliers. After identifying these cases, we ran additional tests without these cases and our results – while they improved slightly – were substantively similar to those seen in Table 3.

Additionally, some have opted for another estimation procedure when the number of countries exceeds the time periods in a sample (e.g. Fornos et al., 2004). Since this aptly describes our dataset, we felt it necessary to run a time-series analysis with a generalized least squares estimation (Liang and Zeger, 1986; Zeger and Liang, 1986).²¹ Again, our findings hold: unstable party systems depress aggregate turnout levels. In the third set of tests, we omitted the time-series specification and ran an ordinary least squares (OLS) regression, but our primary results were unchanged. Therefore, based on these alternative specifications, we are confident that the relationship between party system stability and voter turnout levels is quite robust and is not predicated on the estimation method chosen.

Finally, we have also estimated, though do not report, our equations with a series of regional dummy variables. Predictably, the findings from these tests affirm our conclusions that both volatility and party replacement remain important predictors of voter turnout.²²

Conclusion

Political parties are vital to modern democracy. They serve a number of functions, and several works have alleged that they even factor into whether individuals participate in the political process. Despite this contention, we are still just beginning to understand the various ways in which these organizations influence aggregate participation. This article has offered an attempt to unearth the ways in which parties can foment turnout by looking at how the level of PSI plays into this important calculus. This is a significant finding, since few previous works have looked beyond the number of parties when assessing party system attributes. Clearly, more work is needed to fully comprehend how voters respond to parties and party system stability, and, hopefully, the present study offers a modest contribution to this endeavour.

Overall, this work has discovered that where volatility and party replacement levels are low, turnout rates rise. We believe that this can be attributed to the assets inherent in institutionalized party systems. Namely, that these organizations provide clearer policy platforms to voters, thus making it easier for citizens to discern between parties. Also, because stable party systems typically have stronger ties with social, ethnic, economic or other groups, it should be easier for them to mobilize potential voters. Even when controlling for other institutional, economic and social factors, our results remain statistically significant and in the hypothesized direction. Accordingly, we are convinced that PSI plays a fundamental role in understanding turnout levels.

Footnotes

Notes

Appendix

Table 5.

Countries and Years Used

Country	Parliamentary Elections Used
Albania	1995-2004
Argentina	1985-2004
Australia	1983-2004
Austria	1983-2004
Belgium	1981-2004
Bolivia	1985-2002
Botswana	1989-2004
Brazil	1990-2002
Bulgaria	1992-2001
Canada	1980-2004
Chile	1993-2001
Colombia	1982-2002
Costa Rica	1982-2002
Czech Republic	1996-2002
Denmark	1981-2001
Dominican Republic	1986-2002
Ecuador	1984-2002
El Salvador	1988-2003
Estonia	1996-2004
Finland	1983-2003
France	1981-2002
Germany	1980-2003
Greece	1981-2004
Honduras	1981-2001
Hungary	1992-2004
India	1984-2004
Ireland	1982-2002
Israel	1981-2003
Italy	1983-2001
Japan	1983-2003
Latvia	1995-2002
Lithuania	1996-2004
Luxembourg	1984-2004
Mexico	1988-2003
Moldova	1995-2001
Mozambique	1999-2004
Namibia	1994-2004
Netherlands	1981-2003
New Zealand	1981-2002
Norway	1981-2001
Panama	1999-2004
Paraguay	1993-2003
Philippines	1992-2004
Poland	1993-2001
Portugal	1980-2002
Romania	1996-2004
Russia	1995-2003
Slovakia	1994-2002
Spain	1982-2004
Sweden	1982-2002
United Kingdom	1983-2001
United States	1982-2004

Table 6.

Control Variables and Predicted Results

Variable	Rationale for Inclusion based on Previous Literature	Hypothesized Effect
Concurrent Elections	In concurrent elections, voting becomes more salient since there are more contests at stake, which increases voter turnout.	+
Compulsory Systems	Mandatory voting laws provide disincentives for failing to vote, thus increasing turnout rates.	+
GDP Per capita & Change in GDP Per Capita	Lower levels of GDP per capita are expected to increase turnout as citizens push for policy changes to stimulate the economy. Negative changes in GDP per capita should produce a similar situation where more voters cast a ballot in order to facilitate changes in economic policies.	−/−
Inflation	We expect inflation to be positively related to turnout since voters are likely to respond to high levels of inflation rather than low or modest inflationary levels.	+
Unemployment	Higher unemployment rates are expected to engender higher turnout levels as more citizens head to the polls with hopes of improving their economic situation.	+
District Magnitude	The literature is divided on the effect of this variable with some arguing that district magnitude should be positively linked with turnout since larger electoral districts often produce multiparty systems thus providing more parties from which voters can choose. Conversely, others maintain that since larger districts produce more parties, voters face greater informational costs as they attempt to become informed about each competitor. As a result, this may discourage voters from participating in elections.	+/−
Effective Number of Parties (ENP)	Again, previous work on turnout has been divided on the effect of this variable. Some contend that as the ENP increases so should turnout levels since more parties could offer more choices to voters, thus increasing turnout. On the other hand, other studies have noted that ENP may be inversely related to turnout since voters may be confused by polities with numerous political parties.	+/−
Federalism	Federalism has been linked with lower turnout levels as it may create ‘voter fatigue’ among the electorate.	−
Presidential Systems	Presidential systems are likely to increase the number of overall elections in a democracy thereby overwhelming voters and depressing turnout.	−
Polity2 (Country−level indicator that measures levels of democracy )	This variable captures political and democratic development. More advanced or development states are expected to produce environments conducive to higher levels of voter participation.	+
Total Population	Countries with larger populations should have lower turnout levels as population size decreases the likelihood a voter can make a difference in an election.	−
Percent Urban Population	More urbanized states should witness higher turnout levels since education and literacy rates are higher in more urbanized areas.	+

References

‘Adam Carr’s Election Archive’ (2007) Available at: http://psephos.adam-carr.net (accessed 20 June 2007).

‘African Elections Database’ (2007) Available at: http://africanelections.tripod.com (accessed 31 July 2007).

Aguilar

Edwin Eloy

Pacek

Alexander C.

(2000) ‘Macroeconomic Conditions, Voter Turnout, and the Working-Class/Economically Disadvantaged Party Vote in Developing Countries', Comparative Political Studies 33: 995–1017.

Aldrich

John H.

(1993) ‘Turnout and Rational Choice’, American Journal of Political Science 37: 246–78.

Almond

Gabriel A.

Verba

Sidney

(1963) The Civic Culture: Political Attitudes and Democracy in Five Nations. Princeton, NJ: Princeton University Press.

Beck

Nathanial

Katz

Jonathan

(2004) ‘Time-Series Cross-Section Issues: Dynamics.’ Working paper.

Beck

Nathanial

Katz

Jonathan

(2006) ‘Random Coefficient Models for Time-Series Cross-Section Data: Monte Carlo Experiments', Political Analysis 15: 182–95.

Beck

Thorsten

Clarke

George

Groff

Albert

Keefer

Philip

Walsh

Patrick

(2004) ‘New Tools and New Tests in Comparative Political Economy: The Database of Political Institutions', World Bank Economic Review 15: 65–176.

Birch

Sarah

(2001) ‘Electoral Systems and Party System Stability in Post-Communist Europe.’ Paper presented at the Annual Meeting of the American Political Science Association, San Francisco.

10.

Birch

Sarah

(2005) ‘Single-member District Electoral Systems and Democratic Transition’, Electoral Studies 24: 281–301.

11.

Birnir

Johanna Kristin

(2005) ‘Public Venture Capital and Party Institutionalization’, Comparative Political Studies 38: 915–38.

12.

Blais

André

Carty

R. K.

(1990) ‘Does Proportional Representation Foster Voter Turnout’, European Journal of Political Research 18: 167–81.

13.

Blais

André

Dobrzynska

Agnieszka

(1998) ‘Turnout in Electoral Democracies', European Journal of Political Research 33: 239–61.

14.

Bratton

Michael

(1999) ‘Political Participation in a New Democracy: Institutional Considerations from Zambia’, Comparative Political Studies 32: 549–88.

15.

Bremer

Stuart

(1992) ‘Dangerous Dyads: Conditions Affecting the Likelihood of Interstate War, 1816–1965’, Journal of Conflict Resolution 36: 309–41.

16.

Caldeira

Gregory A.

Clausen

Aage R.

Patterson

Samuel C.

(1990) ‘Partisan Mobilization and Electoral Participation’, Electoral Studies 9: 191–204.

17.

Casas-Zamora

Kevin

(2005) Paying for Democracy: Political Finance and State Funding for Parties. ECPR Press.

18.

Cox

Gary W.

(1999) ‘Electoral Rules and the Calculus of Mobilization’, Legislative Studies Quarterly 24: 387–419.

19.

Crewe

Ivor

(1981) ‘Electoral Participation’, in Butler

David

Ramsey

Austin

(eds) Democracy at the Polls. Washington, D.C.: American Enterprise Institute.

20.

Dalton

Russell J.

Weldon

Steven

(2007) ‘Partisanship and Party System Institutionalization’, Party Politics 13: 179–96.

21.

Dix

Robert H.

(1992) ‘Democratization and the Institutionalization of Latin American Political Parties', Comparative Political Studies 24: 488–511.

22.

Downs

Anthony

(1957) An Economic Theory of Democracy. New York: Harper and Row.

23.

Edwards

Martin S.

Thames

Frank C.

(2007) ‘District Magnitude, Personal Votes, and Government Expenditures', Electoral Studies 26: 338–45.

24.

Farrell

David M.

Webb

Paul

(2002) ‘Political Parties as Campaign Organizations', in Dalton

Russell J.

Wattenberg

Martin P.

(eds) Parties withoutPartisans: Political Change in Advanced Industrial Democracies, pp. 102–28. Oxford: Oxford University Press.

25.

Fornos

Carolina A.

Power

Timothy J.

Garand

James C.

(2004) ‘Explaining Voter Turnout in Latin America, 1980–2000’, Comparative Political Studies 37: 909–40.

26.

Franklin

Mark N.

(1999) ‘Electoral Engineering and Cross-National Turnout Differences: What Role for Compulsory Voting?’ British Journal of Political Science 29: 205–24.

27.

Goodman

Craig

Murray

Gregg

(2007) ‘Do You See What I See? Perceptions of Party Differences and Voting Behavior’, American Politics Research 35: 905–31.

28.

Gosnell

Harold F.

(1930) Why Europe Votes. Chicago, IL: University of Chicago Press.

29.

Gray

Mark

Caul

Miki

(2000) ‘Declining Voter Turnout in Advanced Industrial Democracies, 1950 to 1997’, Comparative Political Studies 33: 1091–121.

30.

Hankla

Charles R.

(2006) ‘Party Strength and International Trade: A Cross-National Analysis', Comparative Political Studies 39: 1133–56.

31.

Huntington

Samuel P.

(1968) Political Order in Changing Societies. New Haven, CT: Yale University Press.

32.

Inglehart

Ronald

(1988) ‘The Renaissance of Political Culture’, American Political Science Review 82: 1203–30.

33.

International Institute for Democracy and Electoral Assistance (2008a) ‘Compulsory Voting’. Available at: http://www.idea.int/vt/parl.cfm (accessed 2 April 2008).

34.

International Institute for Democracy and Electoral Assistance (2008b) ‘Parliamentary Elections’. Available at: http://www.idea.int/vt/parl.cfm (accessed 30 March 2008).

35.

Ishiyama

John

(2007) ‘Electoral Systems, Ethnic Fragmentation and Party Systems Volatility in Sub Saharan African Countries: Implications for Ethiopia.’ Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL.

36.

Jackman

Robert W.

(1987) ‘Political Institutions and Voter Turnout in the Industrial Democracies', American Political Science Review 81: 405–23.

37.

Jackman

Robert W.

Miller

Ross A.

(1995) ‘Voter Turnout in the Industrial Democracies During the 1980s', Comparative Political Studies 27: 467–92.

38.

Karp

Jeffrey A.

Banducci

Susan A.

(2007) ‘Party Mobilization and Political Participation in New and Old Democracies', Party Politics 13: 217–34.

39.

King

Gary

Tomz

Michael

Wittenberg

Jason

(2000) ‘Making the Most of Statistical Analyses: Improving Interpretation and Presentation’, American Journal of Political Science 44: 341–55.

40.

Kostadinova

Tatiana

(2003) ‘Voter Turnout Dynamics in Post-Communist Europe’, European Journal of Political Research 42: 741–59.

41.

Kostadinova

Tatiana

Power

Timothy J.

(2007) ‘Does Democratization Depress Participation? Voter Turnout in the Latin American and Eastern European Transitional Democracies', Political Research Quarterly 60: 363–77.

42.

Kramer

Gerald H.

(1970) ‘The Effects of Precinct-Level Canvassing on Voting Behavior’, Public Opinion Quarterly 34: 560–72.

43.

Kuenzi

Michelle

Lambright

Gina

(2001) ‘Party System Institutionalization in 30 Africa Countries', Party Politics 7: 437–68.

44.

Kuenzi

Michelle

Lambright

Gina M. S.

(2007) ‘Voter Turnout in Africa’s Multiparty Regimes', Comparative Political Studies 40: 665–90.

45.

Laakso

Markku

Taagepera

Rein

(1979) ‘Effective Number of Parties: A Measure with Application to Western Europe’, Comparative Political Studies 12: 3–27.

46.

Ladner

Andreas

Milner

Henry

(1999) ‘Do Voters Turn out More Under Proportional Than Majoritarian Systems? The Evidence from Swiss Communal Elections', Electoral Studies 18: 235–50.

47.

Leeds

Brett Ashley

(1999) ‘Domestic Political Institutions, Credible Commitments, and International Cooperation’, American Journal of Political Science 43: 979–1002.

48.

Liang

Kung-Yee

Zeger

Scott L.

(1986) ‘Longitudinal Data Analysis Using Generalized Linear Models', Biometrika 73: 13–22.

49.

Mainwaring

Scott P.

(1998) ‘Electoral Volatility in Brazil’, Party Politics 4: 523–45.

50.

Mainwaring

Scott P.

(1999) Rethinking Party Systems in the Third Wave of Democratization: The Case of Brazil. Stanford, CA: Stanford University Press.

51.

Mainwaring

Scott

España

Annabella

Gervasoni

Carlos

(2008) ‘Extra System Volatility and the Vote Share of Young Parties.’ Paper presented for the Annual Meeting of the American Political Science Association, Boston, MA.

52.

Mainwaring

Scott P.

Scully

Timothy

(1995) Building Democratic Institutions: Party Systems in Latin America. Stanford, CA: Stanford University Press.

53.

Mainwaring

Scott

Zoco

Edurne

(2007) ‘Political Sequences and the Stabilization of Interparty Competition: Electoral Volatility in Old and New Democracies', Party Politics 13: 155–78.

54.

Marsh

Michael

(2004) ‘None of That Post-Modern Stuff Around Here: Grassroots Campaigning in the 2002 Irish General Election’, British Elections and PartiesReview 14. London: Frank Cass.

55.

Norris

Pippa

(2004) Electoral Engineering: Voting Rules and Political Behavior. Cambridge: Cambridge University Press.

56.

Oatley

Thomas

(2004) ‘Why is Stabilization Sometimes Delayed? Reevaluating the Regime Type Hypothesis', Comparative Political Studies 37: 286–312.

57.

Pacek

Alexander C.

Radcliff

Benjamin

(1995a) ‘Economic Voting and the Welfare State: A Cross-National Analysis', Journal of Politics 57: 44–61.

58.

Pacek

Alexander C.

Radcliff

Benjamin

(1995b) ‘The Political Economy of Competitive Elections in the Developing World’, American Journal of Political Science 39: 745–59.

59.

Pacek

Alexander C.

Radcliff

Benjamin

(1995c) ‘Turnout and the Vote for Left-of-Centre Parties: A Cross-National Analysis', British Journal of Political Science 25: 137–43.

60.

Pacek

Alexander C.

Pop-Eleches

Grigore

Tucker

Joshua A.

(2009) ‘Disenchanted or Discerning: Voter Turnout in Post-Communist Countries', Journal of Politics 71: 473–91.

61.

Pedersen

Mogens

(1979) ‘The Dynamics of European Party Systems: Changing Patterns of Electoral Volatility’, European Journal of Political Research 7: 1–26.

62.

Polhemus

James H.

(1983) ‘Botswana Votes: Parties and Elections in an African Democracy’, Journal of Modern African Studies 21: 397–430.

63.

‘Political Database of the America – Georgetown University’ (2007) Available at: http://pdba.georgetown.edu/ (accessed 1 July 2007).

64.

Powell

Guy Bingham

, Jr. (1980) ‘Voting Turnout in Thirty Democracies: Partisan, Legal, and Socioeconomic Influences', in Rose

Richard

(ed.) Electoral Participation: A Comparative Analysis. Beverly Hills, CA: Sage.

65.

Powell

G. Bingham

, Jr. (1986) ‘American Voter Turnout in Comparative Perspective’, American Political Science Review 80: 17–43.

66.

Powell

G. Bingham

, Jr. and Guy D. Whitten (1993) ‘A Cross-National Analysis of Economic Voting: Taking Account of the Political Context’, American Journal of Political Science 37: 391–414.

67.

Polity IV Project (2000) Polity IV Dataset [Computer file; version p4v2000]. College Park, MD: Center for International Development and Conflict Management, University of Maryland.

68.

Power

Timothy J.

Timmons Roberts

(1995) ‘Compulsory Voting, Invalid Ballots, and Abstention in Brazil’, Political Research Quarterly 48: 795–826.

69.

Przeworski

Adam

(1975) ‘Institutionalization of Voting Patterns, or is Mobilization a Source of Decay’, American Political Science Review 69: 49–67.

70.

Putnam

Robert D.

(1993) Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press.

71.

Radcliff

Benjamin

(1992) ‘The Welfare State, Turnout, and the Economy: A Comparative Analysis', American Political Science Review 86: 444–56.

72.

Rae

Douglas W.

(1967). The Political Consequences of Electoral Laws. New Haven, CT: Yale University Press.

73.

Randall

Vicky

Svasand

Lars

(2002) ‘Party Institutionalization in New Democracies', Party Politics 8: 5–29.

74.

Remmer

Karen L.

(1991) ‘The Political Impact of Economic Crisis in Latin America in the 1980s', American Political Science Review 85: 777–800.

75.

Robbins

Joseph W.

(2010) ‘Party System Institutionalization and Government Spending’, Comparative Politics 42: 229–48.

76.

Roberts

Kenneth M.

Wibbels

Erik

(1999) ‘Party Systems and Electoral Volatility in Latin America: A Test of Economic, Institutional, and Structural Explanations', American Political Science Review 93: 575–90.

77.

Rosenstone

Steven J.

Hansen

John Mark

(1993) Mobilization, Participation, and Democracy in America. London: Macmillan.

78.

Sartori

Giovanni

(1976) Parties and Party Systems: A Framework for Analysis. Cambridge: Cambridge University Press.

79.

Selb

Peter

Pituctin

Sandrine

(2010) ‘Methodological Issues in the Study of New Parties' Entry and Electoral Success', Party Politics 16: 147–70.

80.

Sikk

Allan

(2005) ‘How Unstable? Volatility and the Genuinely New Parties in Eastern Europe’, European Journal of Political Research 44: 391–412.

81.

Smyth

Regina

(2006) ‘Strong Partisans, Weak Parties? Assessing the Assumptions of Cohesiveness within Russian Party Organizations’, Comparative Politics 38: 209 –28.

82.

Stockton

Hans

(2001) ‘Political Parties, Party Systems, and Democracy in East Asia: Lessons from Latin America’, Comparative Political Studies 34: 94–119.

83.

Tavits

Margit

(2006) ‘Party System Change: Testing a Model of New Party Entry’, Party Politics 12: 99–119.

84.

Thames

Frank C.

Edwards

Martin S.

(2006) ‘Differentiating Mixed-Member Electoral Systems: MMM, MMP, and Government Expenditures', Comparative Political Studies 40: 905–27.

85.

Tingsten

Herbert

(1937) Political Behavior: Studies in Election Statistics. London: P. S. King.

86.

Vowles

Jack

(2002) ‘Offsetting the PR Effect? Party Mobilization and Turnout Decline in New Zealand, 1996–1999’, Party Politics 8: 587–605.

87.

Wattenberg

Martin P.

(2002) Where Have All the Voters Gone? Cambridge, MA: Harvard University Press.

88.

Wibbels

Erik

(2000) ‘Federalism and the Politics of Macroeconomic Policy and Performance’, American Journal of Political Science 44: 687–702.

89.

World Bank (2007) World Development Indicators. [CD-Rom]. Washington, D.C.: World Bank.

90.

Zeger

Scott L.

Liang

Kung-Yee

(1986) ‘Longitudinal Data Analysis for Discrete and Continuous Outcomes', Biometrics 42: 121–30.