Assessing the Freeness and Fairness of National Elections

Assessing the freeness and fairness of national elections: Implementing a framework for the systematic study of election quality

Andrew Reynolds University of North Carolina, Chapel Hill, USA Jørgen Elklit University of Aarhus, Denmark

Summary of funding application

In this proposal we are requesting financial support to fund two graduate research assistants for two years, one based at UNC Chapel Hill and one sub-contracted at the University of Aarhus, Denmark. The graduate students will work under the direction of Professors Reynolds (UNC) and Elklit (Aarhus) in managing the panels and data collection and will have access to the data for their own research purposes. Monies are needed for the panels and an initial expert steering group meeting to finalize the method and codebook. Monies will also be expended on the design and maintenance of a website to both collect the data from panels and publish the data for the scholarly community and practitioner world. The project, for which funding is applied for here, will last for three years and we request summer support for the research. This will allow the co-PIs to focus on this research exclusively over the summers of the project period although project management (and some direct research activity) will continue throughout the year.

Scope of project: How to assess election quality

The project will implement a methodology for assessing the quality of a national election: its freeness, fairness, and administrative efficacy. There is a large vacant space within academia and policy analysis that we feel is ripe to be filled, i.e., the lack of a systematic method for assessing election and election management quality which can be applied in both developed and developing world cases, first elections and subsequent ones. The endeavor goes beyond previous work in the field of election and electoral administration assessment by suggesting a more operational and empirically oriented approach as we consciously pay more attention to down-to-earth detailed assessments of all phases of the electoral process and also pay much more attention to the election management process than has been done ever before. We have already developed an assessment framework which we now seek to further operationalise beyond the successful pilot projects we have undertaken so far. Elections take place in specific institutional, political, social, and cultural contexts, as indicated in Figure 1 below.1 The political and social context will – together with the specific political culture – have a strong impact on how elections are perceived by political actors (including the voters) and how they are administered. It also impacts on how losers react to losing (Anderson et al., 2005) a feature, which is probably the best single indicator of a country’s political culture. The figure outlines the phases of an electoral process which form the basis of our assessment methodology in Table 1. The assessment of electoral process and election management quality requires all phases to be assessed, not only because they are all important in their own right, but also because they are intimately interwoven.

Figure 1: The electoral process within its political, social, and institutional framework

POLITICAL AND POLITICAL SOCIAL CONTEXT CULTURE

ELECTORAL LAW (with details of the electoral system)

1. Boundaries E L E C

8. Election result 2. Voter T O

implementation registration R M A E L T

7. Disputes 1.-8. Voter A

S 3. Nomination

education N Y A T G R E A 6. Counting M P 4. Campaign and tabulation E N T 5. Polling

CONSTITUTION (with principles of the electoral system)

The space left by the lack of a robust and comprehensive framework of analysis has been filled by two equally unsatisfactory outcomes. Either election observers make judgments on the basis of impressionistic and incomplete evidence focused on the conduct of the vote and count on election day, or observation missions (often sent from abroad and with their own government’s lead) call an election in a politicized way, detached from any relation to the truth of the process itself (a case in point being the pronouncement by the official South African observation mission that the Zimbabwe elections of 2002 were ‘free and fair’). The literature on election monitoring and assessments of individual elections is to extensive to even attempt to summarise here. A recent overview of some tendencies is Bjornlund (2004). Kelley (2006 a, b) motivates and justifies a very interesting research project, which, however, is quite different from the one, for which support is applied for here. See also Elklit and Svensson (1997).

1 We owe the first version of this figure to Palle Svensson of the University of Aarhus, while William C. Kimberling should be credited for having suggested the splitting of the entire electoral process into phases in this particular way.

2 The greatest failing of election assessment has been the tendency to see election quality in bimodal terms. The election is either good or it is bad; or, when a fudge is required, it is ‘substantially free and fair.’ But there is no doubt that the quality of elections across cases and across time exists on a continuum. Election management within a country can be strong in some areas and weak in others. The playing field which regulates the campaign can vary subtly in both de jure and de facto ways and elections clearly can improve as well as decline on a number of dimensions over time. In essence, one needs to look at both process and outcome to gauge the full picture of election quality.

It would be too simplistic to apply a rigid methodology which gave single scores to try and capture all of the nuances of all that goes into allowing for good elections. At the same time, however, there are clear clusters of election elements which we can assess and one is able to offer an overall assessment of election and election management quality which is more rooted in the evidence than previous, rather impressionistic offerings.

There is also the important question of whether the election failings are great enough to effect the final result? Is an election where only one per cent of the votes are lost or manipulated and the winner wins by half a per cent any worse than an election where thirty per cent of the votes were irregularly cast or treated but the winner wins by thirty- five per cent? This raises the question if election quality should be assessed primarily on the basis of the electoral process or the electoral outcome, or both?

Our intent has been not to offer a foolproof method for categorizing election quality but rather to lay out a framework which is more comprehensive and meaningful than what has come before. Using a consistent and over-arching election and election management quality assessment model allows not only for cross country comparisons but also for internal comparisons of elections within a single country over time. We believe our framework identifies patterns of success and failure in the fairness of elections and is therefore able to spotlight the weak areas of election administration that a government might reasonable focus its subsequent quality improvement efforts on.

We are also consciously aiming for the development of a logically sound and applicable method, which is easy to understand and able to map, and quantify, elections and electoral management systems. The use of the framework in very different environments is in itself a strong argument for not applying complex statistical methods in the construction of individual country/election scores, which may not be appropriate in all cases. This form of modesty is also warranted when the phenomenon under scrutiny is characterized by a considerable number of constantly changing variables, many of which are difficult to measure in a precise, valid, and reliable way. But it goes without saying that any comparative analysis of the data gathered must make use of all appropriate statistical (and other) techniques to test the hypotheses we suggest. The eventual size of the data sets obviously points in the same direction.

The framework as developed will continue to be refined before the data gathering begins. Part of our application deals with the expert panel/steering group which will offer final improvements to the method. Readers might usefully take an election they are familiar

3 with, and have access to the data, and score the case using our methodology. Eyeballing the results alongside the examples we offer here will hopefully give a good feel for the strength of the framework.

From our preliminary work we have come to expect that established democracies might tend to lose points on areas of election management such as transparency, voter education, voter registration (only if they do not use automatic voter registration procedures), campaign regulation, and appeals processes. Our guess is that effective provisions covering these areas have atrophied as public trust in the system has grown over time. The framework therefore identifies a potential Achilles heel in elections even within stable democracies. A thorough assessment of voter registration procedures and com- plaint procedures in Florida prior to the 2000 US presidential election (or in Ohio prior to the 2004 election) would most likely have identified the issues, which later marred the electoral process.

In contrast in fledgling democracies the niceties of electoral law may be robust, at least in the first competitive election, but the playing field of electoral competition is often deeply skewed in favor of dominant parties and elites. This is something which even becomes more problematic over time. Our framework attempts to capture both sides of the equation, the de jure rules and de facto activities which shape elections, the written laws and the practical realities, the freeness of the vote and the fairness of the campaign, as well as the chance to win and the ability to lose.

We are aware of the complexities following from including legal formalities and practical realities and processes in the same index; we intend to experiment with different solutions to that problem, as we are convinced that they should be simultaneously included in the same index as they so evidently interact during the electoral process.

Introducing the framework

One of the chief questions when trying to gauge the freeness and fairness of an election is where to draw the boundary when it comes to deciding what issues are relevant to that question? The boundary lines are murky. While it is important to go beyond polling day and the vote count we exclude from our analysis the very broad determinants of political competition which speak only more indirectly to elections and voting. For example, we include questions of access to public media and boundary delimitation while excluding more general issues of party funding and candidate selection.

When it comes to the election indicators, our rule of thumb is not to pronounce upon the inherent fairness of an electoral system or regulation (if it is generally perceived to be a legitimate democratic option) but rather to assess whether the rules, as written, are applied fairly and without partisan bias. For example, Kenya would not lose points because it uses a majoritarian rather than a proportional election system but because its majoritarian single member districts are so massively malapportioned in a manner which gives rise to partisan bias. A recent presentation of the Kenya malapportionnement

4 problems is Barkan et al. (2006), which also provides an interesting suggestion for a solution.

After settling on relevant areas of election regulation and administration the issue becomes, which questions does one need to ask to gain a clear view of the workings of the given area and what data will serve as good indicators of election performance? In our model we have eleven2 steps ranging from the initial legal framework to the closing post-election procedures. We incorporate a number of areas the Electoral Management Body (EMB) – realizing that in some countries some of these functions rest with more or less parallel bodies – usually has responsibility over: Districting, voter education, registration, the regulation and design of the ballot, polling and counting along with some broader areas such as campaign regulation, complaints procedures, and the implementation of election results. Each step includes 3-10 questions, the answers to which will gauge the quality of election administration and conduct for that step. In sum there are 54 questions which act as our indicators.

Some steps are analyzed primarily through reference to data specific efforts, e.g. on voter education, while others are by necessity scored on the basis of expert judgments (i.e., the perceived legitimacy of the election management body, even though this variable can also, at least in some cases, be gauged from survey data). These answers will be based on survey data, where available, but in most cases on expert readings and assessments of events and the domestic political climate.

The framework is in some ways akin to methodologies used for comparative democracy, human rights, and corruption measures such as Polity, Freedom House, and Transparency International. However, it is also different in at least two important ways: (1) Other democracy indices only include election assessments in rather crude, superficial ways, while our framework is more detailed and requires that closer attention is paid to the actual processes, and (2) the inclusion of election management quality allows us not only to include that important factor but also to connect it much more directly – and probably causally – to the voters perception of the process and through that to political legitimacy also. Table 1 ELECTION ASSESSMENT STEPS AND PERFORMACE INDICATORS

Step Performance indicators How to measure 1. Legal 1.1. Is a consolidated legal foundation easily available? Expert panel framework 1.2. Is a comprehensive electoral time table available? 1.3. Were the elections held without extra-legislative delay? 1.4. Can the electoral legislation be implemented? 1.5. Is the electoral framework broadly perceived to be legitimate? 2. Electoral 2.1. What is the perceived degree of legitimacy/acceptance Polling evidence for management of the EMB by parties and voters? perceptions 2.2. What is the perceived degree of the EMBs impartiality?

2 The difference from the “electoral cycle” model in Figure 1 above is that we now include the establishment of the legal framework around elections as well as the electoral management body (EMB) as separate steps.

5 2.3. What is the perceived quality of the EMBs delivery of Expert panel for de service in these elections? jure and de facto 2.4. What is the perceived degree of the EMBs analysis of EMB transparency? impartiality

Survey of stake- holders for EMB quality and transparency 3. Constituency 3.1. Is the constituency structure reasonable and broadly Expert panel and polling accepted? district 3.2. Is information about constituencies and lower level Stakeholder surveys demarcation districts (demarcation, sizes, seats) easily available? 3.3. Are fair and effective systems for boundary limitation and seat allocation in place used according to the rules? 4. Voter 4.1. What percentage of voters in need of voter education is ‘In need’ is here education exposed to voter education which facilitates their effective operationalized as participation? first time voters. 4.2. Have at risk groups been recognized and their identified needs addressed? ‘At risk’ are 4.3. What percentage of ballots cast is valid? historically 4.4. In terms of voting age population, what percentage of marginalised groups. those eligible to vote for the first time in this election actually voted? Voter education outreach assessed through surveys

Other data from register, polling, and election results 5. Voter 5.1. What proportion of the voting age population is Data from register registration registered to vote? 5.2. Is the register free from serious bias based on gender, age, ethnic or religious affiliation, or region? 5.3. Are qualified people able to be registered with a minimum of inconvenience? 5.4. Are there appropriate mechanisms for ensuring that the information in the register is accurate? 5.5. Are there appropriate mechanisms for ensuring that the public can have confidence in the register? Expert panel analysis 5.6. Are the criteria for registration fair and reasonable and compliant with accepted international standards? 6. Access to 6.1. Are parties allowed, and can parties and candidates Expert panel and design of who fulfill the requirements of registration be registered assessments ballot paper. without bias? Party and 6.2. Are Independent candidates allowed and registered if candidate they fulfil legal requirements? nomination 6.3. Is the method of voting or the design of the ballot paper and non-discriminatory? registration 7. Campaign 7.1. If there is a system to provide access to state-owned Expert panel regulation media, is it implemented equitably? assessments 7.2. If a system for allocation of public funds to political parties is in place, it is implemented? 7.3. Is there an independent mechanism for identifying bias

6 in the state media and is identified bias subject to swift correction? 7.4 Are state resources by and large used properly by the political parties and candidates? 8. Polling 8.1. What is turnout as a percentage of total registration? Data from elections 8.2. What is turnout as a percentage of the voting age results and observer population?3 reports 8.3. Is there a low level of serious election related violence? 8.4. In how many polling stations did polling happen Expert panel according to rules and regulations? assessments based on 8.5. Are there systems in place to preclude and/or rectify data fraudulent voting? 8.6. Is polling accessible, secure, and secret? 8.7. If there is substantial desire for election observation, is the desire satisfied? 8.8. If there is substantial desire for political party election observation, is the desire satisfied? 8.9. Are there systems in place to preclude vote buying? 8.10. Is the level of intimidation so that voters can express their free will? 9. Counting 9.1. Is the count conducted with integrity and accuracy? Expert panel and tabulating 9.2. Is the tabulation transparent and an accurate reflection assessments based on the vote of the polling booth count? data from observer 9.3. Are results easily available to interested members of reports the general public? 9.4. Does counting take place with no undue delay? 9.5. Are parties and candidates allowed to observe the count? 10. Resolving 10.1. Are serious complaints accepted for adjudication? Expert panel election related 10.2. Is there an appropriate dispute resolution mechanism assessments complaints. which operates in an impartial and non partisan manner? Verification of 10.3. Are court disputes settled without un-due delay? Reports final result and 10.4. Do election observation organisations confirm that the certification elections were without serious problems? Legislation 10.5. If legislation prescribes a timeframe for the constitution of parliament, is this timeframe met? Expert panel 10.6. Is a person with a reasonable case able to pursue their assessments case without unreasonable personal or financial risk? 10.7. Are seats taken only by those persons properly elected? 11. Post- 11.1. Are properly documented elections statistics easily Expert panel election available without serious delay? assessments procedures 11.2. Are EMBs audited and the results made publicly available? 11.3. Is there capacity for election review?

The framework also shares the various qualities of the three indices mentioned as well as their problems, which we will not elaborate on here. A fine methodological discussion of tremendous interest here is Munck & Verkuilen (2004), who demonstrate the problems of conceptualizing and measuring democracy connected with different indices of

3 The importance of including proper vote participation in the measurement of democracy was recently demonstrated by Moon et al. (2006).

7 democracy. A comparison with their stringent approach is somewhat discouraging, but we are confident that the framework suggested here will allow us to develop measures, which can also live up to their high standards – at least at the completion of the project. The three scoring systems mentioned all depend on both objective data indicators and subjective expert assessments and they are all, as David Beetham (2004: 2-3; see also Munck & Verkuilen, 2004) aptly categorises them, democracy assessment comparisons based on ‘league tables of human rights and democracy’. An overly simplified scoring system proved inadequate to capture the differing pressures pertaining to established versus fledgling democracies. We therefore developed weightings to reflect step importance relative to each of the two types of polity. Our rule of thumb was: If this element fails will that cause the catastrophic breakdown of the election process?

This assessment enabled us to assign ‘essential,’ ‘important,’ or ‘desirable’ status to each step, as indicated in Table 2, where one also notes that assignments are only partially identical for the two types of polities. To take an example: The standard of election management per se is in our opinion essential in fledgling democracies, because of the nature of the problems surrounding the entire electoral process, while election management in established democracies has become more business as usual. It is still important (as the cases of Florida 2000 or Ohio 2004 made so abundantly clear) but failure does not have the same implications for stability as within democratizing post- conflict polities. Voter education is another example of an element, to which different importance should be attached in established and fledgling democracies.

Table 2 WEIGHTING SYSTEMS FOR ESTABLISHED AND FLEDGLING DEMOCRACIES

Essential Important Desirable ( weight factor: 3) (weight factor: 2) (weight factor: 1) Established 1. Legal framework 2. Election management 4. Voter education democracie 6. Access to ballot 3. Constituency 7. Campaign Regulation s 8. Polling demarcation 11. Post-election procedures 9. Counting the vote 5. Voter registration 10. Resolving disputes Fledgling 1. Legal framework 4. Voter education 3. Constituency demarcation democracie 2. Election management 5. Voter registration 7. Campaign Regulation s 6. Access to ballot 11. Post-election procedures 8. Polling 9. Counting the vote 10. Resolving disputes.

At least two issues – which have invariantly been raised during presentations of the framework – grows out of Table 2: (1) If established democracies do indeed differ from fledgling democracies in the salience of the various elements of the election process, how should the factors then be weighted in the measurement of quality? and (2) Should there be more categories than two? Are elections in post-conflict situations so different that they deserve a category of their own (see on this primarily Lyons, 2004)? Are there other

8 special types of democracies we need to cater for by allocating special weights to some of the steps, e.g. elections in authoritarian regimes? See on electoral authoritarianism in particular Schedler (2006) and Schedler (ed.,) 2006).

Data Collection: The Election Panels

Panel assessments of each legislative election held in an independent state will be conducted only after the release of final election results, but as soon as humanly possible, i.e. normally within a month. The universe of elections will be divided along continental lines with UNC being responsible for the Americas, Asia, and Oceania, and Aarhus being responsible for Eastern, Central and Western Europe and Africa. The RAs will collect the quantitative elements of the framework (primarily indicators: 4.1, 4.3, 4.4, 5.1, 8.1, 8.2) and process the expert panel indicators as they come in.

Code book and quantitative measures A comprehensive codebook will be prepared by the PIs and expert steering group to provide guidance to the expert panelists. Each of the 54 questions is scored on a 0 to 3 scale with 0 being worst and 3 best. Individual scores are then averaged for the panel as a whole (but with no decimals, using ordinary rounding rules) – but if important discrepancies occur panelists will have to go deeper into the issues. In the third step the sum of scores for each of the eleven steps (e.g. 7.1 – 7.4) is standardized relative to the value 10 to allow comparions among steps with different numbers of indicators attached to them. The intermediary step score for all 11 steps is then combined under consideration for the relevant weight factors mentioned above and again with a standardization process with a maximum value of 100 to have comparable values. Dimensionality issues will be addressed using appropriate statistical tests.

Examples:

5.1: “What proportion of the Voting Age Population is registered to vote?”

Countries with automatic registration automatically score 3, while other countries are assessed based on their registration levels compared to relevant regional and development levels. Accepted levels of deviance from such averages will be included in the coding manual.

8.3: “Is there a low level of serious election related violence?”

Here, panelists will base their scoring on a comparison with previous election processes, as there is no generally accepted “election violence scale” to apply. “None” is not difficult, but when is a level low or medium – and what is “serious”? When does an election rally scuffle become “serious”? So assessors must use their discretion, which is why there must be several assessors, and why most of them must have an intimate knowledge of the political culture of the country in question.

9 Panelists Each country/election panel will consist of four experts who will be identified by the PIs and RAs and listed on the website. There will ideally be three locally based and one international expert on each case. Panel experts will include local non-political academic, legal, and election administration experts. The panel will mix scholars and practitioners and one member will ideally be a commissioner or senior figure within the domestic election commission (EMB). Expert panelists will not be party functionaries. The panelists should be knowledgeable, detached, and diverse, and we believe that the data indicators identified give us the best purchase on the questions we seek to answer. It is possible that country experts may assign different scores within some of the survey questions. This is not necessarily evidence of measure problems, as it can also reflect the framework’s sensitivity to a continuum of indicators.

Outline of project activities

Fall 2007 – The framework, scoring and methodology of assessment will be finalized at a workshop consisting of academic and practitioner experts held at the University of North Carolina, Chapel Hill. The group will subsequently be transformed into an expert steering group for the duration of the project.

Fall 2007 – A guide book (manual) for expert panelists will be written to facilitate the scoring of elections. The manual will be accessible on the web site.

Fall 2007 – A web site designer will be contracted to design an interactive site which receives the data from members of the expert panels, processes that data and then publishes the relevant parts on the website for both practitioner and scholarly use.

Fall 2007 – Graduate student project managers will be employed at both UNC and Aarhus to oversee panel operations and data collection and production.

Jan 2008-Jan 2010 – Panel assessments of each legislative election held in an independent state will collect the relevant data.

2010-2011 – The data collected over the previous two years will be analyzed in light of the hypotheses outlined. Research publications will be submitted.

Hypotheses to be tested on gathered data set

The data gathered will speak to the broad issue of the relationship between elections, electoral administration, the fairness of the election process, political legitimacy, and democratic vibrancy. Control variables of a structural character such as development level, number of elections, electoral system, literacy levels, form of political regime (and former regime support), and previous experiences of conflict will be included. One would

10 also expect the character of any recent regime transformation process to play a role, so the inclusion of agency character control variables aiming at taking into account the importance (or otherwise) of such factors is also foreseen.

H1: High quality elections are correlated with democratic progress and institutional consolidation of non-violent democratic norms

Rationale: We expect higher quality elections to be correlated with government legitimacy and over time the consolidation of democratic norms. The means by which we measure election quality will capture how well the bureaucracy functions, the level of inclusivity obtained, and the participation of voters.

H2: First time elections in embryonic democracies are likely to be of higher quality than subsequent elections in said multi-party democracies

Rationale: One might expect a learning curve where an election administration improves over time with experience, and parties and voters acclimatize to new political rules; in contrast we expect that first time elections after a regime change being of better quality than subsequent national elections (as illustrated several times in the democratization literature). First elections often receive vast international attention, resources, and expertise but international support tends to decline dramatically for subsequent elections.

H3: Elections in pseudo-democratic or authoritarian states will be of systematically lower quality

Rationale: The raison d’être of non-democratic states is to restrain free choice and impose political will. Thus elections in non-democratic states will almost by definition be substantially flawed and the level of election quality and level of democracy will – expectedly – be correlated. The inclusion of this hypothesis is therefore not primarily to gain new and surprising insight, but to test the quality and calibration of the scoring tool. It is also included in order to ensure enough variation among cases.

H4: Low quality elections in established democracies will not entail unstable outcomes to the same degree as may low quality elections in inchoate democracies

Rationale: We expect election failures in established democracies to not cause civil unrest, political violence, or civil/ethnic strife to anywhere near the degree that flawed elections can trigger state breakdown in embryonic democracies. One candidate losing the popular vote but winning office in Nigeria, for example, will lead to greater instability than the same scenario occurring in, say, the United States.

H5: Failures in certain elements of the election process will lead to highly unstable and perhaps violent aftermaths while failures in other elements will not cause such consequences

11 Rationale: We expect that election failures in areas such as counting, dispute resolution, or voter education, for example, to have varying degrees of impact on outcomes depending upon the case involved. But the salience of a specific element may vary across time and cases.

Intellectual Merit

It is only recently that the academic literature on democratization – and in particular the literature on the impact of institutions on democratization, political legitimacy, regime support etc. – has begun to acknowledge that governance issues must encompass issues related to the conduct of elections, in both consolidated and emerging democracies. Those analyses generally agree on conceptualizing electoral governance as a set of closely linked activities, sometimes categorised under the headings of rule making, rule application, and rule adjudication (Mozaffar and Schedler, 2000; Kjær, 2004: 157-171; Elklit & Reynolds, 2002; López-Pintor, 2000; Elklit, 1999; Norris, 2004).On the importance of EMB independence from government, see Mozaffar (2002). Mozaffar and Schedler (2000: 5) claim that because elections in established democracies tend to be routine events, usually producing results within a narrow, but fully acceptable margin of error, systematic analysis of electoral governance has not attracted much scholarly interest. There will always be some margin of error as it is difficult to envisage any large-scale operation such as a national election not being occasionally infected by defective ballots, incomplete voter registers, inaccuracies in counting, impersonation, etc. Human mistakes happen, but if these errors are random and do not accumulate to determining the outcome of the election, electoral credibility survives, which is exactly why these credible routines themselves tend to obscure how important electoral governance is. It is a general rule that electoral governance issues only attract critical attention when something goes seriously wrong, or when an electoral procedural or management issue is taken up as part of a more general election related controversy (Mozaffar and Schedler, 2002: 6; Schedler, 2002). It seems self apparent that good electoral governance contributes to the democratic legitimacy of competitive elections, but it is not an easy task to determine exactly how electoral governance per se affects political democratisation and the development of democratic legitimacy, which is the substantive topic we want to address in this project as it has not at all – in spite of its potential importance – been addressed in previous studies concerned with the development and importance of political legitimacy (such as Turner and Martz, 1998; Rose et al., 1998; see, however, Moehler, 2005, for an interesting addition to this literature). Furthermore, we also intend to collect a completely new data set on elections and election management, which will be made available in the public domain. However, Hartlyn, McCoy, and Mustillo (forthcoming) have pushed the boundaries of knowledge further in this regard in their study of Latin American cases since 1980. They note the positive role professional and independent election commissions have had in Latin America, which is similar to what we have seen in relation to the (post-apartheid) Independent Electoral Commission in South Africa (Padmanabhan, 2002; Kamemba, 2005). The claim that electoral and electoral management quality has a bearing on political legitimacy matters is intuitive, but it is more difficult to offer

12 supportive convincing theoretical arguments and empirical evidence. Indeed, previous attempts at conceptualising electoral manipulation have aimed at measuring violations of democratic norms during the electoral process and thus have focused on electoral manipulation as an indicator of illegitimacy, not a cause (Elklit and Svensson, 1997; Goodwin-Gill 1994/ 2006).

Results and publications from pilot research

Foundational research and pilot project implementation of the framework has already resulted in four peer reviewed articles. The project has its seeds in research which began while the co-PIs were in residence at Notre Dame in 1999. This led to a Kellogg Institute for International Studies working paper (Elklit and Reynolds, 2000) which was further developed into an article published in 2002 (Elklit and Reynolds, 2002), which applied the methodology to eight sub-Saharan African multi-party electoral states. The method was further developed during a workshop held with senior members of the Australian Electoral Commission in Canberra in June 20044 which again led to an article in Democratization (Elklit and Reynolds, 2005). The Canberra meeting focused on six cases (Australia 2001, Denmark 2001, East Timor 2001, South Africa 1994 and 2004, and Zimbabwe 2002) as the room contained academics and practitioners intimately involved with the workings of these elections (as Chief Electoral Officers for the domestic election commission or United Nations election support mission, or academic experts on the cases). The weighted and standardized scores (out of 100, cf. above) were:

1. Denmark 2001: 93 2. Australia 2001: 89 3. East Timor 2001: 83 4. South Africa 2004: 77 5. South Africa 1994: 72 6. Zimbabwe 2002: 41 (Elklit and Reynolds, 2005: 159).

Pilot projects were also conducted with in-country experts applying the method to the post-conflict parliamentary elections in Afghanistan in 2005, where the score turned out to be 64 (Elklit, 2005: 51-52). The Elklit-Reynolds framework was discussed in detail at the Inter-Parliamentary Union round table on election standards, Geneva, November 2004, where it aroused considerable interest and positive reactions. The framework was outlined again in the resulting publication Revisiting Free and Fair Elections (Elklit and Reynolds, 2005b; duplicated in Elklit and Reynolds, 2005c).

Along side the Inter Parliamentary Union a number of international organizations are considering using the framework as part of their electoral observation and support activities, including the United Nations Development Program-Mexico who are considering to implement the method in Latin America, the Electoral Institute of Southern Africa (EISA) who are very interested in applying the framework in sub-

4 Rod Medew, Director Research AEC, Michael Maley, Director International AEC, Andy Becker,AEC Commissioner, Ian McAllister, ANU, Nigel Roberts, Victoria University at Wellington.

13 Saharan Africa, and regional sections of the International Foundation for Election Systems (IFES), National Democratic Institute (NDI), and International Republican Institute (IRI).

Broader Impact of Project

To International Organizations assisting democracy. A robust and rigorous method for assessing election quality is needed not merely for the purposes of scholarship but also to assist international organizations in their assessments of where help is needed, where commissions are failing (or succeeding) and how democratic consolidation can most effectively be promoted. The United Nations has a particular interest in rigorous framework as was highlighted at a private meeting that the co-PIs attended to discuss the future of UN electoral assistance in New York on December 7th, 2006. We strongly hope that the model will be further adopted for use as a practical tool for both non- governmental and governmental election observation missions and as a research tool for academics and practitioners to better understand the issues which determine election and election management quality and its impact on political legitimacy.

To domestic election commissions. The model will also allow election managers and administrators to assess the quality of their own work on a comparative basis and subsequently implement qualitatively better procedures where needed.

To collaboration both within academia and beyond. The project will directly promote collaboration between the University of Aarhus in Denmark and UNC Chapel Hill, but we also expect – based on the reactions we have had so far – considerable cooperation with scholars from many different academic institutions. We further expect extensive collaboration between academia and practitioner organizations such as the UN Electoral Assistance Division (UNEAD), UN Development Program Governance (UNDP), Inter- national Foundation for Election Systems (IFES), the Inter-Parliamentary Union (IPU), Carter Center, and domestic electoral commissions, especially in transitional and post- conflict countries.

To the scholarly community. The publication of collected data will be of use to scholars studying institutional design, democratization, and election management. The detailed cross-country data can be used for a variety of research projects while the overall election quality figure can be used similarly to the ways in which the Freedom in the World Political Rights/Civil Liberties scores, Polity IV, Transparency International’s Corruption and The Minority Rights Group International ‘Peoples under Threat’ indexes are used, i.e. both in large-scale analyses aiming at testing more general propositions and in small- scale, more in-depth regional or country-specific studies, studying a specific over-time development. But, again, we want to stress that the framework suggested here is different form those mentioned above as it provides a more in-depth assessment of election and election management quality than anything suggested so far, at the same time as it is so detailed about the election legislation and the election management process that it also

14 will allow election management bodies to use it to assess their internal performance and improve it.

General newsworthiness. The global and regularly updated ‘league table’ of election and election management quality will be of value to journalists wishing to assess democracy, democracy promotion and electoral trends. This we also consider important as it put less well-performing electoral authorities under pressure to perform better next time around.