Sebastian Reyes ’19
In October 2016, the people of Georgia took to the polls to elect a new parliament in what was only the eighth contest held since the founding of the modern republic after the fall of the Soviet Union in 1991. While generally considered free and fair, the results of the election were immensely disproportional, with the ruling Georgian Dream party winning 116 out of 150 or 77% of available seats while receiving a much smaller percentage of the vote. This paper aims to provide a remedy to the current levels of disproportionality by proposing alternative electoral rules that the country may wish to adopt. Though, by modeling the results of the 2016 election under both mixed-member proportional (MMP) and proportional representation (PR), this paper finds that both systems have the potential to reduce disproportionality in Georgia’s elections dramatically, ultimately closed-list PR along with the use of a single, nationwide district and a threshold around 3% is recommended, given the particular concerns of Georgian voters about the shortcomings of their electoral system.
Summary for Web
While free and fair, Georgia’s October 2016 parliamentary elections had a remarkably disproportional outcome. By modeling the results of the election under alternative electoral rules, this paper provides a recommendation as to how the country might reduce its high levels of disproportionality. It is suggested that a closed-list proportional-representation system would be a wise choice for Georgia to adopt.
Disproportionality, the discrepancy between the amount of votes in an election and subsequent seats in a legislature that a party receives, has long been noted as a potential flaw of many electoral systems, especially when present at high levels. This paper is concerned with the extreme disproportionality that has long been present in electoral contests in the country of Georgia. After briefly detailing the history of Georgia’s electoral system, the background to the 2016 election, and the results thereof, this paper seeks to provide potential solutions to lessen the disproportionality of the country’s elections. This goal is pursued by modeling the results of the 2016 election under alternative electoral rules. While models under both mixed-member proportional (MMP) and proportional representation (PR) rules produce more proportional results, ultimately this paper recommends the latter along with the use of a single, nationwide district and a threshold around 3% given a few unique characteristics of Georgian politics.
On October 8, 2016, the Georgian people took to the polls for the first round of voting in what was only the eighth parliamentary elections held since the founding of the modern republic in 1991. Unlike a number of other post-Soviet states, Georgia has been moving steadily towards the consolidation of its democracy, at least since former president Eduard Shevardnadze was ousted in 2004. The 2016 elections were another step in the same direction, and an important one—these were the first contests held after the only successful and peaceful transfer of power between parties in the country’s history in the wake of the 2012 parliamentary elections, a momentous occasion for both Georgia and the region as a whole (Mueller 2014). And yet, despite this progress, public opinion polling suggests Georgians are unhappy with their government and elections, especially their constituency representatives and the high levels of disproportionality produced by the mixed-member majoritarian (MMM) system (“Statement,” 2016), the latter of which is the primary focus of this paper.
In order to most effectively address these issues, this paper suggests that it may be in the best interest of the Georgian people to adopt new electoral rules. To elect its parliament, at present the country utilizes a MMM system. This type of electoral rules blends together elements of proportional systems, which allocate seats in parliament based on the number of votes won, and majoritarian systems, which allocate seats simply to the individual or party who wins the largest vote-share. After exploring Georgia’s current electoral rules in greater depth below, this paper makes the case that a closed-list proportional representation (PR) system with a single, nationwide district and a threshold around 3% may be the country’s best option moving forward. As its name might suggest, PR is the most basic proportional system; it allocates seats to parties based on the number of votes they won in a district. Closed-list PR requires that parties provide ranked lists of candidates, instead of allowing voters themselves to choose candidates on a given party list. A threshold is a share of votes a party must surpass to gain seats in the legislature.
Georgia’s Electoral System
Like a few other post-Soviet states (Birch 2003), Georgia utilizes a MMM system to elect members of the country’s 150-seat, unicameral legislature. The overall number of seats in the parliament has been altered a few times in the country’s short history. For the first free elections held in 1992 there were 225 seats available, a number which increased slightly to 235 from 1995 to 2004, before finally being reduced to the current level of 150 in 2008. The breakdown in seats has also been subject to change. Originally there were 150 proportional and 75 majoritarian seats, before the latter was subsequently increased by 10 (Birch 2003). When the total number of seats was reduced to 150 in 2008 (“Elections,” 2016), there was a perfect 50-50 distribution between proportional and majoritarian seats, but in 2012 the former was increased to 77 and the latter decreased to 73 (Mueller 2014).
The 77 seats of the proportional tier are elected through a single, national district with a 5% threshold. The seat-allocation method employed is rather atypical. Any proportional system requires a clearly defined allocation method to translate vote-share into seat-share, because there are only few seats available relative to the amount of votes a given party may receive. One of the most common methods in use is the d’Hondt formula, which allocates seats depending on which party has the “highest average,” meaning that party’s vote-share divided by a number in a sequence that indicates the seats that the party has already received (“Appendix A,” 2005). The d’Hondt formula uses the sequence: 1, 2, 3, 4, 5, and so on; another notable “highest average” formula, Saint-Laguë, uses the sequence: 1, 3, 4, 7, 9, and so on (“Appendix A,” 2005). Other allocation formulae, called “largest remainder” methods, operate by denoting a quota based on votes cast and seats available; parties are then allocated seats depending on the amount of quotas they fill, with extra seats awarded to parties with the most votes remaining. The Hare formula, for example, uses a quota that is simply the number of votes divided by seats (“Appendix A,” 2005). In contrast to all these allocation methods, Georgian law stipulates that the seats a party wins is equivalent to the “number of votes garnered by a party/bloc multiplied by 77 and divided by an aggregate sum of votes cast for all the parties and blocs” (“Elections Guide,” 2016). This formula occasionally prevents parties which just barely pass the threshold from receiving 6 seats; when this occurs, seats are redistributed from larger parties to ensure that any party which breaks the threshold receives at least 6, the minimum number necessary to form a faction in parliament.
Perhaps more interestingly, Georgia’s electoral system utilizes the two-round system (2RS) rather than the more typical single-member district (SMD) system in its majoritarian tier. SMD is the type of system used in the United States and the United Kingdom, in which multiple candidates compete in a single district for a single seat. 2RS is similar except, as its name suggests, this system utilizes two rounds of voting rather than one. Specifically in the Georgian case, for each of the 73 2RS districts, if no candidate wins over 50% of the vote in the first round, a runoff between the top-two vote-getters occurs no more than 25 days after the first round (“Elections Guide,” 2016). Previously, the electoral rules held that only if no candidate won over 30% of the vote would a second round be triggered; otherwise, the top vote-getter would win in the first round. Up until the most recent elections, districts used in the 2RS tier were previously based on traditional administrative boundaries, resulting in huge population discrepancies amongst districts, from 6,000 in the smallest to 150,000 in the largest. The Georgian supreme court ruled these districts unconstitutional and mandated the country’s election commission to draw new lines. Consequently, for the 2016 election, population discrepancies were reduced to only 12,000 at most, with 41,000 people in the smallest district and 53,000 in the largest (“Elections Guide,” 2016).
2016 Parliamentary Election: Background
The context surrounding the 2016 parliamentary elections transpired to be markedly different from those four years prior in many ways other than the results. While only fourteen parties and two blocs were registered to participate in the 2012 elections (“Political Parties, 2012) these numbers increased significantly to nineteen and six, respectively (“Election Subjects,” 2016). The predominant competitors remained the same; both elections were contests between Georgian Dream, a bloc founded in 2012 by billionaire and former Prime Minister Bidzina Ivanishvili, and the party it was founded to compete against, namely the United National Movement, a generally center-right party founded by another former Prime Minister Mikheil Saakashvili in anticipation of the 2004 elections. Only a few other parties were sizable enough to be even somewhat competitive in the 2016 elections, among them the Alliance of Patriots, a pro-Russian party (Antidze 2016); the Free Democrats, formed by former members of the United National Movement and focused on constitutional reform (“Georgian Opposition,” 2009); and the Democratic Movement, a generally center-right party formed by former acting president and Speaker of the Parliament Nino Burjanadze (“About Us,” n.d.).
Despite the proliferation of parties and blocs, public opinion research suggests that Georgians are generally dissatisfied with their performance. They perceive parties to be concerned with personal disputes and petty rivalries rather than the issues of greatest concern to ordinary people, like unemployment and poverty (“Statement,” 2016). Another perennially salient issue in Georgian politics is foreign policy, particularly the dynamic of the country’s relation with both the West, via NATO and the EU, and Russia. Substantial majorities of citizens continue to support closeting of ties to the West, and all major parties, with the exception of the Alliance of Patriots, reflect such sentiments (“Pre-Election Watch,” 2016).
2016 Parliamentary Election: Results
The best and perhaps only way to characterize the results 2016 parliamentary elections is as a decisive victory for the Georgian Dream; the party was able to drastically shore up the already substantial majority that it had achieved after the 2012 contests. As Table 1 demonstrates, Georgian dream won only plurality of the PR vote, although a sizable one with 21% more than the runner-up, the United National Movement. Nevertheless, the party achieved a majority of PR seats, winning 44 out of 77.
Georgian Dream’s success, however, is even clearer in the results of the 2RS tier, presented in Table 2. Once again, the party won only a plurality of votes in the first round of voting, but remained the highest vote-earner in almost all districts and was able to proceed to the second round of voting in 72 out of 73 contests, of which it won 71. Clearly, it was Georgian Dream’s dominance in the 2RS tier that propelled it to a vast supermajority of seats.
2016 Parliamentary Election: Index of Disproportionality
As would be expected, Georgian Dream’s sweeping win in the 2RS contests resulted in an incredibly high level of disproportionality for that tier. Using the Least Squares Index, the disproportionality of the 2RS tier was found to be a whopping 28.65, as shown in Table 3. Developed by Gallagher, the Least Squares Index is a measure that produces the overall
|Party / Coalition||Round 1 Vote Totals||Round 1 Vote Percent||Round 2 Vote Total||Round 2 Vote Percent||Decisive Round Vote Totals||Decisive Round Vote Percent||Seats||Seat Percent||Total Seats (PR and 2RS)||Total Seat Percent (PR and 2RS)|
|United National Movement||450670||26.01%||212984||24.01%||347166||23.89%||0||0.00%||27||18.00%|
|Alliance of Patriots||94202||5.44%||0||0.00%||26910||1.85%||0||0.00%||6||4.00%|
|Party / Coalition||Vote Totals||Vote Percent||Seats||Seat Percent||Total Seats (PR and 2RS)||Total Seat Percent|
|United National Movement||477053||27.11%||27||35.06%||27||18.00%|
|Alliance of Patriots||88097||5.01%||6||7.79%||6||4.00%|
disproportionality of a system by first calculating the difference between a party’s share of votes and share of seats, the disproportionality of that single party, and then compiling this value with those of all other parties who competed for votes in the election. To quote Gallagher directly, the step-by-step process of calculating the Least Squares Index is thus:
- 1. For each party, take the difference between its percentage share of the seats and the percentage share of the votes
- 2. Square each of these values
- 3. Add these squares
- 4. Divide the resulting total by 2
- 5. Take the square root of this number (“Appendix B,” 2005)
In short, this index demonstrates disproportionality because the greater the discrepancy between a given party’s seat and vote share, the greater that party’s calculated value will be within the index. But because the Least Squares Index cannot consider multiple rounds of voting, determining the disproportionality for this tier presented some challenges in the context of Georgia’s 2RS tier. One’s initial instinct might be to use the sum of the vote totals from the first and second rounds in order to calculate the index, but such an approach is flawed given that it counts first round votes that were ultimately useless in the sense that they did not directly contribute to the election of a candidate to win the seat for a given district. The method ultimately opted for, then, was to calculate the Least Squares Index for each district, and then average these only in the round that directly elected a representative, the “decisive” round, whether it was the first or second. This method appears to be the most accurate and most useful in understanding the true disproportionality of the tier. As Table 2 also demonstrates, Georgian Dream, for example, won only 64.71% of the 2RS vote in the decisive round, yet won 97% of 2RS seats.
Much simpler to calculate was the disproportionality of the PR tier, which was determined to be 10.05 (see Table 3). As will be discussed at greater length in the following section, this level of disproportionality is still noticeably high for a proportional system.
|Least Squares Index (Proportional Tier)||10.05|
|Least Squares Index (2RS Tier, Decisive Round)||28.65|
Discussion: Extreme Disproportionality and Possible Remedies
As mentioned briefly in the previous section, the results of the elections at hand were disproportionate to a remarkable degree. While the mere fact that Georgia’s MMM tier is capable of producing such levels of disproportionality is enough to give one pause, more concerning is the fact that the results of this year’s contests were by no means a one-off. Figure 1 presents the Least Squares Index for the PR tier of all Georgia’s elections since 1992 aside from 2003, the results of which were ultimately annulled due to widespread fraud (Jones 2005). The lack of data disaggregated by round and precinct for all elections prior to this year’s elections prevented me from creating a similar graph for the 2RS tier, but the numbers for the PR tier are striking in their own right, peaking at 39.93 in 1995. Moreover, while the disproportionality of the PR tier has decreased since the replacement of the original 7% threshold with a 5% one in 2008, a look at Table 4 shows that the disproportionality of both Georgia’s PR and 2RS tiers remain high in comparison to similar systems in other countries.
The question, then, remains: what should be done to ameliorate the problem of high disproportionality in Georgia’s electoral system, if anything? This question is addressed by modeling the 2016 election under alternative electoral rules.
The first calculation conducted was to determine the results under a mixed-member proportional (MMP) system rather than its sibling, MMM. Such a change should produce more proportional results because MMP allocates seats by distributing the proportional seats in a manner so as to compensate for the difference between the number of seats won by a party in the majoritarian tier and those to which it is entitled by its portion of the total vote. Table 5 demonstrates that this expected result is indeed achieved. While it is difficult to determine the exact disproportionality of the MMM system given the parallel nature of the two tiers, it seems to reasonable to suggest that the is approximately the mean of the two numbers, or about 19. The disproportionality under the modeled MMP scenario, coming in at 10.59, is certainly an improvement.
And yet disproportionality of 10.59, as Table 4 suggests, remains on the high end for a system that purports to be very proportional. Why is this so? The answer is no other than the 5% threshold and the provision for a minimum of 6 seats to any party that surpasses that threshold. Table 5 presents the results of the 2016 election under MMP both with and without these two aforementioned features. A quick look at Table 7 will show that the disproportionality significantly diminishes in the latter case.
A move to MMP seems as though it would be a natural path for Georgia to follow, given that such a system is so similar to the one that the country currently uses while having the capacity to mitigate the most contentious aspects of the current system. And yet Georgia is unlikely to adopt an MMP system for two reasons. First, the same public opinion research cited above suggests that voters are dissatisfied with their representatives elected by 2RS, whom they consider removed and unrepresentative of their interests (“Statement,” 2016). Second, voters largely agree that the MMM system adversely impacts the representativeness of the legislature, perhaps in large part thanks to public discussions of a potential switch to straight PR.
Given this, Table 6 displays the results of the 2016 election under list-PR rules, with and without the 5% threshold. In both cases, the results are very similar to those of their respective counterparts under MMP. When a threshold is in place, the only difference is that neither any
|Georgia (PR Model w/ No Threshold, 2016)||0.89|
|Georgia (MMP Model w/ No Threshold, 2016)||1.31|
|Georgia (PR Model w/ 3% Threshold, 2016)||3.23|
|Georgia (PR Tier, 2008-2016)||6.983|
|Georgia (MMP Model w/ 5% Threshold, 2016)||10.59|
|Trinidad and Tobago†||12.644|
|Saint Vincent and the Grenadines†||15.303|
|Georgia (PR Tier, 1992-2016)||15.511|
|Saint Kitts and Nevis†||17.661|
|Party/Coalition||Vote % (PR)||Total Seats (MMM)||% of Seats (MMM)||Total Seats (MMP, 5% Threshold)||% Of Seats (MMP, 5% Threshold)||Total Seats (MMP, No Threshold)||% Of Seats (MMP, No Threshold)|
|United National Movement||27.11%||27||18.00%||50||33.33%||41||27.33%|
|Alliance of Patriots||5.01%||6||4.00%||9||6.00%||8||5.33%|
|Vote % (PR)||Total Seats (MMM)||% of Seats (MMM)||Total Seats (PR, 5% Threshold)||% Of Seats (PR, 5% Threshold)||Total Seats (PR, 3% Threshold)||% Of Seats (PR, 3% Threshold)||Seats (PR, No Threshold)||% of Seats (PR, No Threshold)|
|United National Movement||27.11%||27||18.00%||50||33.33%||43||28.67%||41||27.33%|
|Alliance of Patriots||5.01%||6||4.00%||9||6.00%||8||5.33%||8||5.33%|
Independents or the Industrialists party enter parliament under list-PR, because the 2RS tier, through which both won a seat, is removed. Without a threshold, the only difference between the results under MMP and PR is that, once again, no Independents win a seat; the Industrialists still win just enough of the vote to gain a single seat. These discrepancies between the two systems, as Table 7 demonstrates, make the PR system less proportional when the 5% threshold is in place, while the opposite is true without a threshold.
|Current MMM System, PR Tier||10.05|
|Current MMM System, Decisive Round, 2RS Tier||28.65|
|Proposed MMP System With 5% Threshold||10.59|
|Proposed MMP System Without Threshold||1.31|
|Proposed PR System With 5% Threshold||11.28|
|Proposed PR System With 3% Threshold||3.23|
|Proposed PR System Without Threshold||0.89|
In terms of addressing the high levels of disproportionality, the results under list-PR with no threshold all but resolve the matter. And yet it is both highly unlikely and would be highly unwise for Georgia to adopt such rules. Without a threshold, there are very few barriers obstructing parties from achieving a seat in parliament. For example, under the model above, the party For Georgia’s Peace was able to win a seat with just 3824 votes, or 0.21% of all votes cast. In general this is concerning because, while For Georgia’s Peace’s does not likely fit in such a category, PR facilitates the election of extremist parties, which are typically smaller in size (Farrell 2001). What seems more reasonable is the adoption of a threshold somewhere between 0% and 5%. Consequently, in Table 7, the results with a 3% threshold, a level used in many European countries, are modeled. This model produces an entirely reasonable level of disproportionality of 3.23.
Before concluding, it is necessary to address a few other considerations at play in the process of electoral reform, both in terms of the Georgian case and in general. First, a discussion of the role that allocation methods play in regards to this exercise is in order. In attempting to address the issue of disproportionality, a few different allocation methods were tested in order to determine which would produce the most proportional results. While there was no differentiation amongst the allocation methods in terms of seats when a 5% was maintained, there was some variation with a 3% threshold and more with no threshold (see Supplementary Materials). Overall, however, the Hare formula proved to be the most proportional (See Table 1 in Supplementary Materials), and, therefore, the results obtained with this formula are included in the tables above. And yet, it is dubious that the Hare formula will be that which Georgians chose if they decide to adopt a proportional system. Given Georgian Dream’s vote share of almost 50% in this election, the allocation formula may have the potential to either grant or deny that party a majority of seats in the parliament. Georgian Dream would likely, then, demand a formula such as D’Hondt or Imperial which is more favorable to large parties (Gallagher and Mitchell 2008) to ensure the best chance that it gains a majority. At the very least, it appears that the allocation formula may be a contentious issue if the switch to PR is ever made.
Second, it will be useful to briefly address the issue of districts. All the above models utilize a single, nationwide PR district, but it is entirely conceivable that, in moving to a PR system, Georgians decide to use a number of smaller, multimember PR districts. Indeed, such an option seems particularly feasible because the country has an institutional background for this type of system; 10 large multimember districts were used to elect representatives the PR tier up until 1999 (Birch 2003). Using current administrative boundaries and multimember districts proposed in a 2011 USAID policy paper (Dahl 2011), a model of what such a system might look like was considered, but this task quickly proved to be impossible given the new district lines that were drawn for the 2016 election. Regardless, a list-PR system with low-magnitude multimember districts may be ideal for Georgia as such a system would likely support a level disproportionality similar to that under PR with a single, nationwide district (Carey and Hix 2011).
Third, a quick word about closed-list versus open-list PR. Given that the current system uses a closed list for the PR tier, it seems likely that a switch to PR would simply maintain this rule. Moreover, a move to open-list PR might be unwise because it would only exacerbate the unease of many Georgians that their parties focus on interpersonal disputes rather than policy debates, given that open-list PR generates intraparty competition and, thus, demands that candidates rely more on personal vote-earning attributes, especially at higher magnitudes such as that of a single, national district (Shugart, Valdini, and Suominen 2005).
Lastly, a note about the limitations of this exercise. All of the above models were created under the assumption that the vote shares that each party achieved would remain the same under all of the proposed electoral rules. This assumption seems to have greater validity in this instance than if the results of Georgia’s elections were modeled under other electoral rules , given that MMP is closely related to MMM and also already includes a PR tier. But it is by no means entirely valid. Were the 2016 elections held under just list-PR, a few differences in the distribution of votes would be plausible. For one, it is likely that smaller parties would have received a larger vote shares, as their voters would be more incentivized to vote given that the barriers to winning a seat posed by the threshold and the majoritarian nature of the 2RS tier. However, turnout and, consequently, vote share for some parties might decrease due to the inability of most independent candidates to compete under list-PR. These are just two of the many ways the vote distribution of the 2016 election might differ under list-PR.
Despite all these concerns, as Georgians continue to debate the shortcomings of their current electoral system and the merits of alternatives, it seems clear that a move to a closed-list PR system with a single, nationwide district and a threshold around 3% appears to be an excellent option for future elections. Such a change would drastically reduce the disproportionality of the current MMM system while eliminating majoritarian representatives, who are much disliked by ordinary voters. A closed-list PR system with a number of smaller, multimember districts may also be a viable option worth considering, given careful consideration of the size and boundaries of districts in such a case.
Antidze, Margarita (2016, October 11). “Pro-Russian party wins a toe-hold in Georgia’s new parliament.” Reuters. Retrieved from http://www.reuters.com/article/us-georgia-election-idUSKCN12B1XF
Birch, Sarah (2003). Electoral Systems and Political Transformation in Post-Communist Europe. New York: Palgrave MacMillan.
Carey, John M. and Simon Hix (2011). “The Electoral Sweet Spot: Low-Magnitude Proportional Electoral Systems.” American Journal of Political Science, 55(2), 383-397.
Central Election Commission of the Republic of Georgia (2016). “Election Subjects Participating in the Parliamentary Elections According to the Sequence of Submitting the Application.” Retrieved from http://cesko.ge/res/docs/PartiebiParlamentiEng29.09.2016.pdf
Central Election Commission of the Republic of Georgia (2016). “Elections 2016: Preliminary Results.” Retrieved from http://results.cec.gov.ge/eng/
Central Election Commission of the Republic of Georgia (2012). “Political Parties, Registered for the Parliamentary Elections of October 1, 2012.” Retrieved from http://cesko.ge/res/old/other/11/11041.pdf
Civil.ge (2016, October 3). “Elections Guide.” Retrieved from http://civil.ge/eng/article.php?id=29478
Civil.ge (2016). “Elections.” Retrieved from http://civil.ge/eng/category.php?id=32
Dahl, Robert (2011). “Georgia’s Parliamentary System: Options for Advancing Voter Equality.” USAID. Retrieved from https://www.ifes.org/sites/default/files/georgias_parliamentary_electoral_system_2011_0.pdf
Democratic Movement Party. “About Us.” Retrieved from http://www.democrats.ge/index.php/en/chvens-shesakheb/tavmjdomare/biograpia.html
Gallagher, Michael and Paul Mitchell (2008). “Appendix A.” In Michael Gallagher and Paul Mitchell (Eds.), The Politics of Electoral Systems (pp. 580-597)
Gallagher, Michael and Paul Mitchell (2008). “Appendix B.” In Michael Gallagher and Paul Mitchell (Eds.), The Politics of Electoral Systems (pp. 598-606)
Gallagher, Michael (2015). “Election Indices Dataset.” Retrieved from http://www.tcd.ie/Political_Science/staff/michael_gallagher/ElSystems/index.php
International Republican Institute (2016). “Pre-Election Watch—Georgia: October 8, 2016.” Retrieved from http://www.iri.org/web-story/pre-election-watch—georgia-october-8-2016
Jones, Stephen (2005). “Presidential and parliamentary elections in Georgia, 2004.” Electoral Studies, 24, 303-311.
Mueller, Sean (2014). “The parliamentary and executive elections in the Republic of Georgia, 2012.” Electoral Studies, 34, 342-346.
National Democratic Institute (2016). “STATEMENT OF THE NATIONAL DEMOCRATIC INSTITUTE PRE-ELECTION DELEGATION TO GEORGIA.” Retrieved from https://www.ndi.org/files/Georgia-PEAM-Statement-ENG-17-06-2016.pdf
Radio Free Europe (2009, July 17). “Georgian Opposition Leader Alasania Forms New Party.” Retrieved from http://www.rferl.org/a/Georgian_Opposition_Leader_Alasani_Forms_New_Party/1779224.html
Shugart, Matthew Søberg, Melody Ellis Valdini, and Kati Suominen (2005). “Looking for Locals: Voter Information Demands and Personal Vote-Earning Attributes under Proportional Representation.” American Journal of Political Science 49(2), 437-449.