Defining the playing field

The playing field is a concept often used to describe level of fairness in electoral competition. With Levitsky and Way’s definition of the playing field as a case in point, this paper takes a critical look at existing work on the playing field, arguing that current conceptualizations suffer from lacking conceptual logic, operationalization and measurement. A new and disaggregated framework that can serve as the basis for future research on the playing field is then proposed. This framework is applied to an illustrative case study on the development of the playing field in Zambia under MMD rule, thereby demonstrating that it is able to capture both the changing nature of the playing field and the differing mechanisms at play to a larger degree than the framework put forth by Levitsky and Way. The 2011 elections in Zambia also clearly highlight the importance of conceptually and empirically separating the slope of the playing field from its impact on both the opposition and electoral outcomes.


Introduction
The past 25 years have seen a large increase in the number of regimes that hold multiparty elections but do not conform to international standards when it comes to the conduct of elections, access to political rights, and civil liberties (Carothers 2002;Hadenius and Teorell 2007). Nowhere in the world are these regimes as prevalent as Sub-Saharan Africa: "The reality across the sub-continent is clearly one of 'hybrid 1 3 regimes'" (Lynch and Crawford 2011, p. 281). Despite the "routinisation of elections" (van de Walle 2003, p. 299) on the continent, the majority of these elections take place in contexts where incumbency advantages are extreme due to excessively strong executives (Rakner and van de Walle 2009, p. 112).
This trend has coincided with increasing use of the concept of the playing field to evaluate electoral quality. Often, and especially in less developed countries, election monitoring reports and the media claim that the lack of an even playing field compromised the quality of a given election because it prevented fair competition between the incumbent and the opposition (Levitsky and Way 2010b, p. 57;Merloe 1997;Schedler 2006, p. 1). General agreement exists among both scholars and election observers that a relatively even playing field between parties and candidates competing in an election is essential for fair competition (cf. Bjornlund 2004;Goodwin-Gill 1998). However, left unresolved are what the concept of the playing field entails within the context of electoral competition, which components are part of the concept, which indicators can measure its evenness and how these indicators should be aggregated. The result has been widespread use of a concept (especially in the media and election monitoring reports) without consensus on what this concept entails or how it should be measured and applied. Levitsky and Way (2002, 2010a, 2010b have changed this through their work on competitive authoritarianism. They argue that in addition to free and fair elections and civil liberties, an even playing field-defined as equal access to resources, media and the law-is essential for democracy, making the presence of an uneven playing field a hallmark of a competitive authoritarian regime Way 2010a, p. 7, 2010b, p. 63). Their work has provided a clear purpose and analytical tools for this concept but as this paper will highlight, it falls short of actually measuring the playing field. Building on Levitsky and Way's work, this paper proposes a framework that can be used to anchor an analysis of the evenness of the playing field. After a brief review of existing work on the concept, a critique of Levitsky and Way's conceptualization and measurement is presented. Subsequently, an alternative, disaggregate framework for measuring the playing field is proposed and discussed before it is presented and applied in a study of the playing field under the competitive authoritarian Movement for Multiparty Democracy (MMD) regime that held power in Zambia from 1991 until 2011.
The analysis highlights several advantages of this framework relative to that of Levitsky and Way: (1) The importance of separating the evenness of the playing field from its effects on political actors, as emphasized by the fact that the playing field was more even in the 2006 election when the MMD retained power compared to 2011, when it lost. (2) That a dichotomous measure of the playing field as either even or uneven as proposed by Levitsky and Way hides important variation between elections in Zambia. (3) Although changes in the evenness of the different attributes of the playing field are relatively synchronized, the initial analysis indicates that there are different dynamics driving these changes for each attribute, implying that a disaggregated framework is fruitful. The Zambian case thus illustrates that the framework proposed here leads to different conclusions and different avenues for future research than do existing studies of the playing field.

3 2 Understanding the playing field
The traditional use of the playing field as a concept can be found in discussions on distributive justice, especially concerning the important principle of equality of opportunity (e.g. Roemer 1998). The notion here is that "justice requires levelling the playing field by rendering everyone's opportunities equal in an appropriate sense, and then letting individual choices and their effects dictate further outcomes" (Arneson 2008, p. 16). Equality of opportunity thus focuses on the opportunity of individuals to make informed decisions through (1) eliminating initial inequalities not chosen by the individual, and (2) providing fair conditions for interaction. The focus on fair interaction is key to understanding the link between an even playing field and democratic electoral competition. As Bartolini (1999Bartolini ( , 2000 notes, competition is a central aspect of democracy. However, for competition to be democratic it depends on contestability. Contestability focuses on the fairness of the political contest between political actors. One of the issues Bartolini highlights as imperative for fairness is "the possibility of accessing resources necessary for an electoral race with the other" (Bartolini 1999, p. 457). Thus in Bartolini's conceptualization of democracy, the playing field is a key issue.
Within the context of real life electoral contests, a "level playing field" has since the 1990s been used to denote a situation where no group participating in an election has a better chance at winning as a result of unfair conditions (Elklit and Svensson 1997, p. 36; see also Goodwin-Gill 1998;Merloe 1997;Gould and Jackson 1995). However, early work on the playing field was often focused strictly on a narrow, preelection time period and on formal and legal factors, thus failing to be of significant analytical value.
Levitsky and Way correct many of the mistakes in early work on the playing field by giving the concept a clear purpose. They argue that the central point is whether or not the playing field is even. This, they contend, can be determined by identifying where "the opposition's ability to organize and compete in elections is seriously handicapped" (Levitsky and Way 2010b, p. 58) as a result of incumbency advantage throughout the electoral cycle (Levitsky and Way 2010a, p. 6). The authors then specify the advantages crucial for the playing field: We define an uneven playing field as one in which incumbent abuse of the state generates such disparities in access to resources, media, or state institutions that opposition parties' ability to organize and compete for national office is seriously impaired. (Levitsky and Way 2010b, p. 57) Thus the playing field is fundamentally about incumbents' abuse of state power in order to generate relative differences in access to resources, media and the law 1 between the incumbent and opposition, both during and between elections. The attributes of this concept-access to resources, media and the law-are largely in tune with the wider literature on electoral competition under authoritarianism, which highlights access to resources, media and the law as central factors for understanding the fairness of political competition and subsequently the degree of control enjoyed by the incumbent (cf. Gandhi and Lust-Okar 2009;Morgenbesser 2013;Morse 2012;Schedler 2013). 2 To measure the concept, Levitsky and Way (Levitsky and Way 2010a, p. 368) further operationalize it by creating an indicator per attribute (with subcomponents), adding that a violation of any indicator is sufficient to score the playing field as uneven. In terms of access to resources, an uneven playing field is present if the incumbent makes widespread use of public resources or uses public tools to limit or skew access to private-sector finance. Concerning access to media, the playing field is uneven if state-owned media is the primary source of news and biased towards the incumbent, or if private media is manipulated through various mechanisms. While these operationalizations are clear, the operationalization of the third attribute, access to law, is somewhat problematic. This problem is linked to the first of my three criticisms of Levitsky and Way's conceptualization of the playing field: the partly lacking conceptual logic. The other issues are untested causal relationships in the conceptualization and the dichotomization of a disaggregated concept. I will deal with each in turn below.

Conceptual logic: redundancy 3
While the initial discussion of the attributes of the playing field seems to indicate that access to law refers to the relative neutrality of nominally independent dispute resolution institutions (Levitsky and Way 2010a, pp. 10-12), this is not reflected in Levitsky and Way's operationalization, where the indicator focuses on the politicization of state institutions in general (Levitsky and Way 2010a, p. 368). This creates a discrepancy between conceptualization and operationalization, making it difficult to know what Levitsky and Way really mean by access to the law. At the component/ indicator level, Levitsky and Way state that access is uneven if "state institutions are widely politicised and deployed frequently by the incumbent" (Levitsky and Way 2010a, p. 368). This overlaps with both the indicators for resources and of media, as these are also fundamentally about incumbent abuse of public institutions to garner 2 There are some who would disagree, however. In a recent analysis of the playing field in South Africa and Botswana, de Jager and Meintjes (de Jager and Meintjes 2013, p. 249) argue that Levitsky and Way's categories are not comprehensive enough, as the difference in the playing field between these two countries is fundamentally about the African National Congress's (ANC) liberation credentials in the South African case. What de Jager and Meintjes fail to specify, however, is whether the liberation credentials are an unfair advantage in and of themselves, or if it is the potential consequences (such as increased access to public finances) of the liberation credentials that could create an uneven playing field. This highlights the importance of separating the causes and effects of the playing field from the measurement of the concept, and underscores the importance of it being as minimalist as possible for the concept to have any analytical purpose (for a similar argument on measures of democracy, see Munck and Verkuilen 2002, p. 9). 3 At the level of conceptualization where Levitsky and Way link the playing field to the issue of regime identification, they are guilty of another breach of conceptual logic as they have the playing field as a separate component at the same level as civil liberties, and free and fair elections and as a subcomponent of free and fair elections (l.evitsky and Way 2010a, pp. 366-368). This is an example of conflation (Munck and Verkuilen 2002, p. 13), which is a potentially serious problem not just in a logical but also in an empirical sense. However, since this article is primarily concerned with the playing field as a concept and not how the playing field is linked to the issue of regime, the issue will not be pursued further in this article. 1 3 advantages in terms of funding or attention. It is thus an example of redundancy: Access to law as defined by Levitsky and Way relates to the same issues of the overarching concept as the two other attributes within the concept (Munck and Verkuilen 2002, p. 13). The consequence is that there should be no instances that violate the criteria for access to resources or media but not access to the law. Yet Levitsky and Way identify several such instances, notably Benin, Botswana and Madagascar between 1990, and Botswana in 2008(Levitsky and Way 2010a. As the attribute is operationalized, it does not serve a clear purpose. Had the operationalization been more connected to supposedly neutral arbiters and their conduct, access to law would be clearly distinguishable from the other attributes. 2.2 Untested causal relationships: agency, cause and effect As noted by Goertz (2006, pp. 54-55), the fact that causal relationships are often central components of concepts can potentially be problematic and they should therefore be both explicitly stated, theoretically well-founded and, if possible, empirically tested. Levitsky and Way include two causal relationships in their definition of the playing field. First, they state that the playing field is "uneven" if and only if "incumbent abuse of the state" generates disparities (Levitsky and Way 2010b, p. 57, emphasis added). Here they argue that for the playing field to be uneven, any disparities between incumbent and elite must be a result of incumbent agency: The incumbent must intentionally manipulate access to resources, media or the law. However, must unfair disparities always be the result of incumbent abuse? And how do we separate "abuse" from use? These questions should be addressed in greater detail.
I argue that while intent to abuse might be present in most cases, this is not necessarily so. The issue of patronage highlights this. According to Levitsky and Way's analysis, patronage-based machines were one of the most important aspects of the uneven playing field in many competitive authoritarian regimes in Sub-Saharan Africa (Levitsky and Way 2010a, Chap. 6). However, if the patronage networks were created as a side effect of policies intended to do otherwise-for example policies aimed at boosting employment-they would not qualify as contributing to an uneven playing field according to Levitsky and Way's definition. Botswana is a case in point. While the state-owned Debswana company's monopoly in the diamond industry undoubtedly contributed to an uneven playing field as illustrated by Levitsky and Way (2010a, pp. 255-256), it is hard to prove that the intent of establishing and running such a monopoly was and is to prevent the opposition from effectively mobilizing. The question of agency and intentionality should therefore be analysed and measured to the greatest extent possible, but not necessarily included in the definition and operationalization.
The second causal relationship mentioned and included in the definition is that the playing field is only uneven if the disparities created by incumbent abuse impede the opposition's ability to organize and compete (Levitsky and Way 2010a, p. 368). As it stands, Levitsky and Way's framework is not primarily intended for the purpose of understanding the playing field. Instead it is primarily used for operationalizing a cut-off point between different regime types based on precisely this point. However, while there might be merit to Levitsky and Way's causal claim, this should not in any way rule out a more thorough focus on and measurement of the playing field before investigating its relative impact. If, as Levitsky and Way argue, an uneven playing field is an increasingly important "tool" for autocrats (Levitsky and Way 2010b, p. 57), then it is imperative to investigate the issues involved separately from their origin and especially their impact. Furthermore, it is important to investigate the causal claim and identify where and when the playing field prevents opposition mobilization. It is therefore necessary to separate cause and effect. Some existing studies highlight this. Opalo points out that the 2011 election in Zambia led to a turnover despite "the tilt that he [incumbent President Banda] and his party … had given to the electoral playing field by misusing state resources and vastly outspending the opposition" (Opalo 2012, p. 80). Similarly, while de Jager and Meintjes (2013) find that the playing field is uneven both in Botswana and South Africa, its composition is very different, which in turn has significant effects in terms of its impact on the opposition. Cause and effect should therefore be held as separate as possible.

Dichotomization: oversimplified measurement of a complex concept
Given that Levitsky and Way are primarily interested in the impact of the playing field for categorical purposes, a dichotomous measure makes sense as it provides a clear cut-off point. However, there are important arguments for a more differentiated measurement of the evenness of the playing field. While it is true that something is either even or uneven, there are clearly different degrees of unevenness, and with events involving people there is almost never perfect evenness. Levitsky and Way themselves acknowledge this as they discuss the "slope" 4 of the playing field (Levitsky and Way 2010b, p. 64). They furthermore describe a total dominance of the playing field in the early 1990s in Zimbabwe, but after the economic downturn in the late 1990s the playing field became much more even (Levitsky and Way 2010a, p. 241). Nevertheless, the playing field is coded as static throughout the entire period, thus hiding temporal variation. De Jager and Meintjes (2013) find the playing field uneven in both South Africa and Botswana, but describe significant variation in this unevenness. A more discriminating and disaggregated measure would therefore allow for more variation across both time and space.
A disaggregated concept furthermore makes it possible to look at the different aspects of the playing field separately or together, depending on the purpose. This does however necessitate an open approach regarding which data is used, who does the coding and what the coding decision is based on. Thus the coding process must be transparent and open, especially as coding decisions involved in measuring an abstract concept such as the playing field will entail some sort of subjective judgement. While such judgements are not inherently unwanted and are often necessary and even productive, they necessitate a very open approach in terms of how judgements are made (Schedler 2012). While Levitsky and Way provide a list of indicators (Levitsky and Way 2010a, p. 368), they are extremely hard to replicate, as they are very broad. They do not provide an explicit list of coders or sources of data. Instead, they base their decisions largely on secondary literature and expert opinions, without applying any standardized framework. 5 We are thus left to wonder why Zambia violates the criteria of access to media in 2008 whereas Benin and Mali do not, despite VonDoepp and Young (2013) identifying relatively free private media that all experienced harassment in the period in all three countries.

A new framework studying the playing field
Following the discussion above, the playing field is defined as the balance between incumbent and opposition in access to resources, media and the law. The proposed concept is organised as a four-level framework (Goertz 2006;Munck and Verkuilen 2002), with each level (concept-attribute-component-indicator) 6 organised in a hierarchical relationship as illustrated in Table 1. The framework roughly corresponds to Levitsky and Way's framework in two ways: (1) the overall attributes (column 1 in Table 1) discussed above, and (2) which actors that are relevant to study. Given that the playing field in this instance is related to electoral politics, it makes sense that what is relevant is the difference in access between the incumbent and opposition competing in elections. 7 While it might be slightly misleading to lump all opposition parties together, it is analytically useful as it provides a better understanding of the opportunity structure they face.
It is nevertheless important to do something that Levitsky and Way disregard as a result of their focus on the impact of the playing field: specify the different components (column 2 in Table 1) of the attributes of the playing field and clarify the respective indicators (column 3 in Table 1). The point of departure of the components of the framework is that access to resources, media or the state can come from different sources. This means that each component in the framework represents a different possible avenue of access. This focus on different sources allows for an understanding of which parts of the playing field are "closed" to the opposition, and which are open. Previous studies have highlighted that precisely the diversity of the "menu of manipulation" (Schedler 2002) and the interplay between different sources of resources is important to understand the electoral playing field between incumbent and opposition (e.g. Albaugh 2011; Arriola 2013; Lynch and Crawford 2011; Rakner and van de Walle 2009).
The indicators that the coders should use to assess the evenness in access to the different components are posed as a set of questions. This is a common way of measuring elements related to complex, composite concepts with regard to electoral quality (e.g. Bland et al. 2013;Elklit and Reynolds 2005;Norris et al. 2013). The coder should code each indicator on a scale from -4 to + 4, where -4 is a situation where the indicator totally favours the opposition, 0 symbolizes a relatively even opportunity for both incumbent and opposition, and + 4 is a situation where access to the indicator totally favours the incumbent. The intermediate scores of -3, -2, -1, 1, 2, and 3 refers to a high, moderate and low advantage for the opposition and incumbent respectively. 8 The indicators are then averaged at the component level. For the temporal scope of each unit, the coder should focus on the situation during a full electoral cycle. 9 This means that a score should be assigned based on the situation from the point in time when a national election ends (the result is official and all complaints are settled) and until the next national electoral cycle is complete (the results of a new national election is official and all complaints settled). 10 However, the coder will not only have to assess the score of the component, but also its importance for the playing field. As highlighted by Bland et al., a common problem with measures that rely on a list of issues that should be assessed is that they often "do not provide a means of weighing the relative importance of each item on the list" (Bland et al. 2013, p. 360). Goertz (2006, p. 46) argues that the issue of weighting is especially crucial for concepts where the different attributes and components are potentially substitutable and vary across space and time, both of which I argue are true for the playing field. While adding an importance weight might seem similar to the impact criteria of Levitsky and Way, it is different because importance refers to whether and what importance the component has in the overall picture of the playing field rather than the impact it has on the competition between the incumbent and opposition. Each coder must therefore assign an "importance score" of 0 (no importance), 0.5 (low importance), 1 (medium importance), or 1.5 (high importance) based on how important the component is deemed to be for the playing field. This is a version of what Goertz (2006, p. 46) calls "weighting of necessary conditions". The component score will then be multiplied with the importance score of that component, thus eliminating the value of any component with little or no importance and increasing the value of any component with high importance. The importanceadjusted component-scores should then be added together and standardized at the component level. 11 In the following, the components of each attribute are explained and discussed. The discussion of the framework is based on studies of electoral competition in Sub-Saharan Africa. However, I would argue that the attributes of the framework are universal in nature and can therefore serve as a point of departure for studies of the playing field from other locations as well.

Access to resources
Access to resources is extremely important in the Sub-Saharan African electoral game. Large geographic distances and poor infrastructure combined with a campaign culture of visiting as many people as possible and organising rallies demands large quantities of both material and non-material resources (Saffu 2003, pp. 21-23). On average, incumbents in Africa tend to be well funded, especially in countries where the state plays an important economic role (Butler 2010;Saffu 2003;van de Walle 2003). And with a weak private sector and relatively few instances of institutionalized public funding, opposition parties tend on average to be underresourced (Arriola 2013; Rakner and van de Walle 2009; . Five broad categories of funding sources are particularly important for political parties in the Sub-Saharan African context: internal funding, private funding, public funding, illicit public funding, and foreign funding (Bryan and Baer 2005;Butler 2010;Helle 2011;Saffu 2003).
In terms of internal and private funding, the indicators selected focus on fundraising through small-scale membership contributions as well as wealthy individuals and businesses, all of which have an effect on the Sub-Saharan African context (Bryan and Baer 2005;Butler 2010;Saffu 2003). Despite this, there is still no denying that on average the most important source of revenue for political parties in Africa is the state (Saffu 2003, Bryan andBaer 2005). If there is legal public funding of parties or party campaigns, it is still not necessarily based on fair criteria and implementation (Bryan and Baer 2005;Helle 2011). The indicators on public funding therefore focus on this. In terms of illicit public funding, either abuse of public resources such as staff and infrastructure or direct money flows form public coffers to the ruling party or elite can contribute to an uneven playing field (Helle 2011;Prempeh 2008). Finally, the diaspora, foreign governments and business interests have been found to play a significant role in funding political parties in Sub-Saharan Africa (Bryan and

Access to media
In the last 10 years, online media and mobile technology have gradually expanded the variety of sources and availability of media in Sub-Saharan Africa (Wasserman 2011, p. 4). However, access is contingent on social status and place of residence, creating vast differences in the importance of various media sources (Hyden and Okigbo 2002;VonDoepp and Young 2013). Three broad categories of media are important: private, public, and PSC 12 media.
Private media varies significantly in importance and outreach, but has still played a central role in electoral struggles on the continent. Like private funding, access to private media can be uneven, either as a result of ownership structures (Andriantsoa et al. 2005) or through government pressure or harassment that might induce selfcensorship and biased reporting (VonDoepp and Young 2013). In terms of public media, demands should be stricter. Public media should in theory represent the interests of the citizens of the state, and provide relatively equitable coverage for all political parties competing in an election. For the playing field to be coded as even in terms of public media, the opportunities of access should be equal both in terms of quantity and quality of coverage in all relevant media platforms. With other types of media, a combination of new technological innovations and public frustration with existing 1 3 media has fuelled an increasingly vibrant popular and communal media environment in many African countries that uses both new and old technology to reach out to people (Wassermann 2011). Opportunities to access and use alternative media such as communal radio stations, SMS campaigns and mobile services, as well as online and social media should therefore also be equal where relevant.

Access to law
As mentioned above, this framework focuses on the narrower notion of access to law as access to nominally independent arbiters that affect the playing field. The two prominent institutions here are electoral management bodies (EMBs) and courts. In the case of the former, these can be uneven in their direct or indirect dependence on government, or through skewed access (Bland et al. 2013;Elklit and Reynolds 2005;Lopez-Pintor 2000;Mozaffar 2002). EMBs must thus be neutral in composition, funding, and access; act on all legitimate input despite where the input originates from; and act as a neutral arbiter if conflicts arise.
In cases where the EMB is not the designed arbiter for complaints related to conduct of elections and thus the playing field, the court system usually is (Schedler and Mozaffar 2002, p. 16). Given that courts can be controlled through direct and indirect means (Ginsburg and Moustafa 2008, pp. 14-20; Gloppen et al. 2010), it is essential to develop indicators that investigate the fairness of access to the courts. All relevant parties should have equal possibilities to forward, advocate for and win their cases, and the courts should uphold their role in ensuring that the legal aspects of the playing field remain equal.

A small note on data
Given the specific nature of the indicators above, the demand on data quality is high. The framework is therefore best suited as a guide for primary data collection. A nonexhaustive list of data collection techniques and sources of data include interviews with key actors, stakeholders and experts; expert and media surveys; media publication data; party accounts; legal evaluations; and public documents and reports. Thorough primary data collection would allow for a triangulation of data to evaluate the composition of the playing field. This could potentially make the framework a valuable tool for election observation missions, political risk analysis and political economy analysis.
Using the framework to analyse the playing field by investigating secondary sources is more difficult, especially across time and space. This is because the framework asks for very specific information that may not be easily accessible. If secondary sources are used, it requires a very rigorous analysis of the variety of sources available. The coding is much more difficult because secondary sources invariably involve some form of subjective evaluation, and the differences found in access between the indicators might be as much about differences of opinion between the researchers as actual differences in access to different sources of the playing field. A triangulation exercise is therefore preferred because it is easier to code cases where plenty of information exists, and where this information is recent. Another option would be to find indicators that function as proxies for the questions posed in the framework. For example, instead of making an overall assessment of access to private media, a researcher might use social network analysis to calculate the relative closeness of media owners to the incumbent or opposition.
Although there are clear challenges in using secondary data, it is a worthwhile effort, especially as a point of reference given that Levitsky and Way base their analysis on secondary sources. Thus in the following section, the framework is applied to one of Levitsky and Way's cases: Zambia after the MMD came to power in 1991.

(Re)coding Zambia under MMD rule
Levitsky and Way (Levitsky and Way 2010a, pp. 370-371) code the playing field in Zambia as static. It is uneven in terms of access to resources, media and the law both during the 1990-1995 period and in 2008. While acknowledging in their empirical analysis that changes did occur in access to the different aspects of the playing field (Levitsky and Way 2010a, pp. 288-291), this is not reflected in their measurement. By applying the framework presented above, a somewhat different scenario emerges. Tables 2-6 in Appendix 2 present the information, the coding decisions and the results of the coding exercise. The coding is based on a review of published literature, including Levitsky and Way's own sources that focus on political competition in Zambia. Empirical information from country reports on Zambia from Freedom House, the US Department of State's Bureau of Democracy, Human Rights and Labour, and election-monitoring reports have also been consulted. 13  The changes are described in Fig. 1. A score of 0 constitutes an even playing field, while scores above 0 can be seen as higher access for the incumbent relative to the opposition. A score of 3 would indicate that access largely favours the incumbent.
In Zambia the playing field has moved from providing the incumbent with a clear advantage in the 1996 election, to a much more even (though still unfair) contest in the post-Chiluba period from 2001 onwards. At the aggregate level the changes between electoral cycles have been relatively synchronized. However, while access to media and access to law have become increasingly even over time, access to resources has consistently favoured the incumbent to a large extent. The analysis of the Zambian case thus shows how the framework presented here offers a much more dynamic picture of the playing field than that of Levitsky and Way. It shows not only that the playing field is changing, but also which elements of the playing field are changing and at what level. While the playing field has indeed remained uneven and tilted towards the incumbent in all elections under MMD rule in Zambia, the degree of the tilt has varied considerably.
While it is not within the scope of this article to thoroughly investigate what is driving these changes, scores at the component and indicator level provide at least 1 3 some clues that can form the basis for further research. Regarding access to media, it becomes clear that what changed from the 1996 election cycle onwards was not that access to public media became significantly more even, but that there were simply other alternative media channels for the opposition to access and that these became gradually more important. First the presence of private newspapers and community radio stations, and later private broadcasters and new media, made the continued bias of public broadcasters less significant. Thus, while Moyo (2010) might be right in labelling the MMD government "reluctant liberalizers", the fact that several private media outlets with significant outreach did emerge out of the liberalization process in Zambia (Willems 2013, p. 225), has contributed to a more even playing field. While some private outlets, notably The Post newspaper after 2006, were biased towards the opposition, most provided relatively balanced political coverage, demonstrating the positive role of neutral media outlets in balancing the playing field. Access to law slowed over time, as the supposedly "neutral arbiters" (the EMB and the courts) became more substantively neutral. There are several potential reasons for this: increasing professionalization and capacity, a more benevolent incumbent, or that the level of electoral malpractice has decreased over time. These explanations nevertheless support Lindberg's (2009) hypothesis of the democratizing effect of elections over time, as institutions and actors gradually adopt more democratic practices through repetitious processes. 14 As Fig. 1 shows, access to resources is the attribute of the playing field that remained tilted against the opposition throughout MMD rule. Even in the 2011 elections won by the opposition, the MMD had a massive resource advantage, largely as a result of its creative use of state resources (EUEOM 2011, p. 14). As Pitcher (Pitcher 2012, p. 118) highlights, the MMD abused state resources both for in-campaign and between-campaign purposes throughout its tenure. The lack of access to state funding for other parties meant that opposition parties had to rely on wealthy private indi-14 However, as noted by Bogaards (2013), the evidence regarding the overall effect of elections is far more varied than Lindberg's theory predicts. 1 3 viduals in order to compete with the financial muscle of the MMD. While available private sector funding never balanced the playing field, the presence of at least some independent private funding, both as a result of liberalization processes and wealthy individuals defecting from the ruling party in the 1990s, meant that the MMD did not succeed in closing access to resources completely (Pitcher 2012, pp. 116-125). This contrasts with Mozambique or Uganda, where the ruling regimes have been able to successfully control funding from private businesses as well as private media to a much larger extent (Helle 2011;. Again, this highlights the importance of understanding different sources of access to different aspects of the playing field.
These discussions show the merit of disaggregating a complex concept such as the playing field to understand not only what drives changes in the playing field at the aggregate level but also what drives the changes in its respective attributes. It is especially important given that the underlying causes of changes in the playing field might vary significantly between its different components.
The point about access to resources also highlights why it is so important to separate the playing field as a concept from the impact it has on the opposition. As Hess and Aidoo (Hess and Aidoo 2013, p. 138) have recently argued, in the 2011 elections the Patriotic Front (PF) actually managed to turn the resource advantage of the MMD into an electoral advantage, as the opposition could claim that the government was bought by foreign (mostly Chinese) interests and therefore did not have the best interests of Zambians in mind. Party Leader Michael Sata and the PF also actively encouraged voters to accept bribes offered by the MMD but to vote for the PF instead, as it would put more money in voters' pockets after the elections instead of offering government hand-outs in advance (Helle and Rakner 2012, pp. 10-11). While the abuse of state money clearly made access to resources uneven, the impact it had on the opposition was at the same time partly positive. This was, however, contingent on an intermediate factor: the tactics chosen by the opposition.
This leads to a final note regarding the mapping of the playing field in Zambia: The MMD did not lose power at the point in time when the playing field was most even. Though the differences were not large, the 2006 elections were in many ways fairer than the 2011 elections. However, while Levy Mwanawasa and the MMD won the election in 2006, Rupiah Banda lost in 2011. This corroborates the finding above that the impact of the playing field on not just the opposition but also the outcome of elections is contingent on or subservient to a number of other factors. Opposition parties can and do win when the playing field is uneven, as long as it is able to adapt to the circumstances. As highlighted by Schedler (Schedler 2013, pp. 369-370), the response of the opposition to the uneven playing field might be just as critical for understanding eventual regime change as underlying structural variables and the state of the playing field. It is therefore essential to see the nature of the playing field as simply one of many variables that affect the opposition's ability to mobilize and compete.

3 5 Studying the playing field: a way forward
The case of Zambia thus highlights three important issues regarding the concept of the playing field. First, the playing field should not be seen as static, as the level of unevenness shifts significantly between electoral cycles. Second, a disaggregated approach to measuring the playing field highlights that there may be different underlying drivers and mechanisms of change involved in the different attributes of the playing field. Third, the slope of the playing field is not necessarily a suitable indicator of the impact it has on the opposition or election outcomes.
The framework put forth and discussed in this paper thus presents a very different reality to the static presentation of the playing field in Zambia made by Levitsky and Way. It highlights the deficiencies of Levitsky and Way's definition and measurement, particularly with respect to the static nature of their measure and their failure to separate the playing field from its impact on the opposition. While the creation of a dichotomous measurement might have been necessary for Levitsky and Way's purpose, it might not serve as the best tool for understanding the playing field. Studies of the playing field or its attributes should thus use caution when applying Levitsky and Way's framework.
If the primary purpose of future studies is to understand the playing field itself, the framework presented here can provide a point of departure. It opens several new interesting avenues of research. First, efforts should be made to use this framework to investigate the causal claims inherent in Levitsky and Way's definition. How important is agency for the emergence or preservation of an uneven playing field? When does an uneven playing field prevent the opposition from mobilizing effectively? Second, further effort needs to focus on how different attributes of the concept act separately and collaboratively. Are the same underlying conditions driving changes in all attributes? Are some combinations of an uneven playing field more unfavourable for fair competition than others? Finally, efforts need to be focused on solving the "big" question: What drives changes in the overall playing field? How can we achieve true "contestability" (Bartolini 1999, p. 457) in a political system? And what level of fairness is fair enough? These are all questions that future applications of this framework should aim to address.

Directions for how to assign scores on indicators
-4 = Indicator completely favours the opposition. Opposition has total control over access to this source. -3 = Indicator largely favours the opposition. Opposition has good control over access to this source, but the incumbent also enjoys negligible access. -2 = Indicator favours the opposition to a medium degree. Opposition has a clear edge over the incumbent in access to this source, but the incumbent also has some access.
-1 = Indicator slightly favours the opposition. Opposition has a slight edge over the incumbent with regards to access to this source, but the difference is small. 0 = An even playing field. Either both the opposition and the incumbent have little to no access, or access is equal. 1 = Indicator slightly favours the incumbent. Incumbent has a slight edge over the opposition in access to this source, but the difference is small. 2 = Indicator favours the incumbent to a medium degree. Incumbent has a clear edge over the opposition in access to this source, but the opposition also has some access. 3 = Indicator largely favours the incumbent. Incumbent has good control over access to this source, but the opposition also enjoys negligible access. 4 = Indicator completely favours the incumbent. Incumbent has total control over access to this source.
Directions for how to assign importance scores on components 0 = No importance for the playing field. Access to component is non-existent, either as a result of the issue not existing or none of the actors using it. 0.5 = Low importance for the playing field. Access to component is of limited importance for electoral competition. 1 = Medium importance for the playing field. Access to component is important, but is not imperative for electoral competition. 1.5 = High importance for the playing field. Access to the component is critical for electoral competition.
Appendix 2: Using the framework to for measuring the playing field in Zambia under MMD-rule, 1991-2008 Table 2 Indicators for the playing field in Zambia, 1991Zambia, -1996 Indicator Score Basis of coding Importance RINTPAR 0 No reports found on denial of right to party membership, or abuse based on party membership. MMD only party competing with significant membership base (Lodge et al. 2002, p. 393). UNIP enjoyed small income from some party businesses established during oneparty era, but most of this had been confiscated by state (Kabemba 2004, p. 15) 0 RPRIWEL 1 Wealthy businessmen most important sources of independent funding in Zambia (Lodge et al. 2002, p. 393). Associates of MMD government consistently favoured and business-community firmly behind MMD at the beginning of period. Privatization process favoured associates of ruling party, especially circle surrounding Chiluba who became very rich (van Donge 2008). Some fall-out as several high profile business allies of MMD and ministers left party and government to contest for elections for new parties. New parties largely dependent on personal wealth of party elite or independent businessmen (Ihonvbere 1995, pp. 13-15) 1.5 Some abuse of public office to procure funds through corruption, especially at minister level, used in campaigns (Ihonvbere 1995, pp. 13-15;van Donge 2009). State funds used to conduct campaign and party activities (Le Bas 2011, p. 224;Pitcher 2012, p. 118) 1.5

RILLRES 2
Cabinet ministers and president used official transportation during campaigns (Lodge et al. 2002, p. 393;Rakner 2003, p. 106) RILLAPP 2 MMD increased size of government to accommodate supporters (Rakner 2003, p. 104). Patronage prevalent but corresponded with cuts in reach of state (Rakner 2003, pp. 184-185). Offices handed out based on party membership, not merit (Pitcher 2012, p. 118) RILLPOL 2 Relief food and heavily subsidized housing provided just before or during campaign (Lodge et al. 2002, p. 393). Fertilizer, maize and development fund handed out in rural areas (Rakner 2003, p Relatively few privately owned newspapers -three newspapers that where to some extent independent of MMD control. Had a relatively balanced cover, but favoured opposition candidates (Banda 1997, pp. 38-58) No significant private broadcasting (Chirwa 1997, p. 42). Legal framework contained threat of President closing media (Banda 1997, p. 11), while the constitution at the same time recognizes press freedom (Banda 1997, p. 17). Legal framework also compromises rights for journalists (Banda 1997, p. 17). MMD sources frequently consulted in reports (Banda 1997, pp. 38-58). An issue of The Post banned by the President in early 1996, and several editors arrested. Editors also threatened physically, and journalists harassed (Chirwa 1997, pp. 30-31;Rakner 2003, p. 110) 0.5

MPOPCOP 0
Although the opening up of the airwaves in the early 1990s allowed for communal radio, only limited, church-based radio stations emerged in this period (Kasoma 2002, pp. 21-22) 0 MPOPSOC -Not applicable in this period LEMBNEU 2 EMB appointed and controlled by ruling President, seen as illegitimate (Baylies and Szeftel 1997;Lodge et al. 2002, p. 382). Characterized by both low competence and some bias (Baylies and Szeftel 1997;Banda 1997, p. 2) 1 President could dismiss judges based on incompetence (VonDoepp 2006, pp. 396-397), however, seldom effective, and records of judges appointed show little difference from long-serving judges (Gloppen 2003;VonDoepp 2005). Court much less likely to rule against state if important government actors such as President and Ministers were involved (VonDoepp 2006, p. 395), and generally ruled in favour of government (Gloppen 2003). However, courts did rule against government on important issues such as access to public space and media freedom (Gloppen 2003, pp. 119-120). Bowed to executive pressure on petitions on Presidential elections (Gloppen 2003, pp. 120-121). Ruled for press freedom in several instances, but did not declare controversial law unconstitutional (Chirwa 1997, pp. 34-35;Phiri 1999, p. 56) 1

Access to law = (2 + 1)/2 = 1.5
The playing field score = (2.43 + 2.75 + 1.5)/3 = 2.23 Table 3 Indicators for the playing field in Zambia, 1996-2001 Indicator Score Why Importance RINTPAR 0 Most parties competing in elections were relatively new. Exception MMD and UNIP, but membership fees constituted a small part of funding, and even these parties had relatively weak organizations (Rakner and Svåsand 2004, p. 59;Le Bas 2011, p. 227). Party cards cost more to produce than they cost for buyers to buy, and party business was negligible (Kabemba 2004, p. 15). Members and voters typically expected goods in return rather than to contribute (Pitcher 2012, p. 123). UNIP sold off party businesses to fund campaign (Momba 2004, p. 29) 0 RPRIWEL 0 Several opposition leaders were wealthy businessmen (Masoka, Mwila) who had contributed to funding MMD or former MMDstalwarts (Tembo, Sata, Miyanda) who all brought significant funds to their parties (Burnell 2003, pp. 391-392;Erdmann and Simutanyi 2003, p. 36;Pitcher 2012, p. 123;Rakner 2003). MMD also supported by state-dependent businesses as well as internal party elites who had massively benefitted from privatization processes (Pitcher 2012, pp. 124-135;van Donge 2008). Government threatened those who made efforts to gather support and funding for opposition parties (BDHRL 2001, p. 6) 1.5

RPUBPAR -
No public funding of political parties -RPUBCAM -No campaign funding for political parties 1 3  : 2002, p. 11;Carter Centre 2002, p. 28). Party and campaign activities also funded by state funds (Kabemba 2004, p. 16). Massive abuse of state resources (Gould 2006). President had own discretionary fund that could be used (lawfully outside campaign period) for purposes he wanted (Rakner and Svåsand 2004, p. 61) 1.5 RILLRES 4 MMD made use of civil servants, including lower level officials and newly created district officers (Burnell 2003, p. 394;Carter Centre 2002, pp. 27-28;Rakner 2003, pp. 113-115). Ministers, president and vice-president made use of state vehicles for transportation (Carter Centre 2002, p. 28). Party also used departmental staff and telecommunication facilities (Kabemba 2004, p. 27) RILLAPP 3 MMD had established civil servant positions that were given to party cadres who used them for party work (Burnell 2003, p. 394;Carter Centre 2002, p. 28). MMD used government appointments to build inclusive patronage alliance (Lindemann 2011) RILLPOL 1 MMD created favourable policies for those who supported them financially (Pitcher 2012, p. 124)

RFORGOV -
No reports about significance of contributions from foreign sources 0

RFORDIA -
No reports about significance of contributions from the diaspora MPRIMED -1 Private newspapers, TV stations and radio stations owned by individuals who were not in association with ruling party, many supported opposition (Carter Centre 2002, pp. 29-30). Private media covered all parties, but were markedly more critical of MMD (BDHRL: 2002, p. 8; Mwalongo 2002). Journalists from The Post arrested several times when publishing critical reports about regime (Rakner 2003, pp. 199-201). Independent newspapers, radio and TV stations that where consistently targeted and harassed throughout the period and in the build up to the elections (BDHRL 2001, pp. 8-9;Carter Centre 2001, pp. 29-30;FH 1999FH , 2001 1 MPUBMED 3 A clear overrepresentation of MMD (Mwalongo 2002). Coverage of MMD was presented as news, whereas coverage of opposition was presented as adverts (Burnell 2003, p. 394;Carter Centre 2001, p. 29;Kabemba 2004, pp. 27-28). Public media prevented from critical reporting of ruling party-practices self-censorship (BDHRL 2002, p. 8;Phiri 1999, p. 58;Kabemba 2004, p. 28). Government media proclaimed MMD-victory while voting was still ongoing (Burnell 2003, p. 394). Presidential debate on day before voting cancelled by ZBC in favour of MMD-related program (van Donge 2008, p. 307). Government dominates broadcasting, and tightened control in run-up to election in 2001 (FH 2002) 1.5

MPOPCOP -1
By early 2000s the three first independent communal radios were up and running. They were seen as opposition friendly, and communal radio stations that where critical of government were harassed and targeted by the government (Carter Centre 2002, p. 29;Kasoma 2002, p. 289) 0.5

MPOPSOC -
The Post had established a small following of its online paper. State-owned newspapers also had an online presence, but the impact of politics was negligible (BDHRL 2001(BDHRL , p. 9, 2002  Funding and appointment of EMB controlled by government. (Carter Centre 2002, pp. 23-24). Delay of funds led to mismanagement and mistrust-clearly lacked independence. Poor performance in registration exercise (Tordoff and Young 2005, p. 416). It did however strive for more opposition access to media and other smaller issues. Widespread accusations of favouritism towards MMD-also confirmed by courts (BDHRL 2006, p. 7;Carter Centre 2002). Main problem was weakness of EMB, which benefitted ruling party (Kabemba 2004, pp. 33-35) 1 LCOUNEU 1 Executive and legislative had budgetary control of parliament (Gloppen 2003, p. 127). Some evidence of patronage in judiciary (VonDoepp 2005, pp. 292-294). President could dismiss judges based on incompetence and choose new candidates with aid of MMD-controlled parliament (Gloppen 2003, p. 125;VonDoepp 2006, pp. 396-397), however, given that it was end of his term it was less likely to be of consequence (VonDoepp 2005, p. 277). Furthermore, process is very difficult and has multiple veto points (both institutional and external pressure) (Von Doepp 2005, p. 288). Small pool of candidates to select from makes possibilities of appointing "favourable judges" limited (VonDoepp 2005, p. 289). Successful defeat of constitutional reforms that would curb judicial review function of judiciary (VonDoepp 2005, p. 287). Smear campaign by ruling party against the chief justice may have led to him becoming more lenient towards the president (Gloppen 2003, p. 121). Court much less likely to rule against state if important government actors such as the president and ministers are involved (VonDoepp 2006, p. 395). While courts generally favoured the ruling party, the supreme court ruled against the president in key electoral decisions regarding term limit issues, as well as on eligibility for citizenship for former President Kaunda, (VonDoepp 2005, p. 277, 283) and abuse of public resources for electoral purposes (Carter Centre 2002, p. 28). Courts often ruled in favour of opposition with regards to electoral petitions in the 2001 election, especially in parliamentary races, but while critical of performance of EMB as well as abuse of state resources and media, it did not find that irregularities in presidential election were critical for results (BDHRL 2006, p. 7;Gloppen 2003, pp. 120-121;FH 2006) 1.5

Access to law = (2+1.5)/2 = 1.75
The playing field score = (2.25+1+1.75)/3 = 1.66 1 3  (Momba 2005, p. 28) Party cards cost more to produce than they cost for buyers to buy, and party business was negligible (Kabemba 2004, p. 15). Members and voters typically expected goods in return rather than to contribute (Pitcher 2012, p. 123). Most parties demand contributions from MPs salaries for campaigns of party, which is significant source of income for parties (Momba 2004, p. 29). MMD rebuilt some party structures in this period (Le Bas 2011, p. 229) 0.5 RPRIWEL 0 MMD continued to be supported by state-dependent businesses as well as internal party elites who had massively benefitted from privatization processes, as well as business sector dependent on government contracts. Also prevented opposition-linked businesses from competing on equal terms for access to public contracts and parastatals (Pitcher 2012, pp. 139-142;van Donge 2008). Parties largely dependent on candidates contributing funds to own campaigns, especially in opposition. Opposition linked to business community and former MMD stalwarts (Momba 2005, p. 28) 1.5

RILLRES 3
Continuous reports about use of government resources, particularly transportation and infrastructure, in elections (BDHRL 2005(BDHRL , p. 11, 2006EUEOM 2006, p. 19). Some attempts to limit abuse at minister level (Gould 2006) RILLAPP 2 MMD used government appointments to build inclusive patronage alliance (Lindemann 2013). Reports of state officials threatening to fire public employees who votes for opposition and hiring based on party membership (BDHRL 2007, p. 14;EUEOM 2006, pp. 18-19). Extensive use of parastatal system for political purposes (Pitcher 2012, pp. 139-140) RILLPOL 3 MMD created favourable policies for those who supported them financially and used fertilizer subsidies in the countryside (Pitcher 2012, pp. 138-142). Districts that voted for MMD got disproportional access to subsidized fertilizer (Mason and Ricker-Gilbert 2013, p. 89). MMD extensively used local development projects, health initiatives, housing and fertilizer policy selective for campaign purposes, particularly in countryside (BDHRL 2006, pp. 13-14;COG 2006, p. 23;EUEOM 2006, pp. 18-19;Larmer and Fraser 2007, p. 633). However, some of these initiatives must also be seen as general policy successes rather than simply election policies (Larmer and Fraser 2007, p. 634)  Fundraising from abroad legal. UPND raised some funds from unspecified sources abroad, but hard to determine significance (Momba 2005, p. 29). Some general donor funded programs, and some programs with individual parties, but impact is considered small (Svåsand and Rakner 2011Rakner , pp. 1260Rakner -1264 0

MPOPCOP -1
At least three community-based radio stations broadcasted throughout the period, and they and other smaller channels increased outreach significantly. Although nominally independent and critical of government, they faced pressure from MMD officials and where often not allowed to publish political material (Banda 2006;BDHRL 2004, p. 9) Opposition benefited significantly from access to these and private radio stations to spread their appeal (Larmer and Fraser 2007, p. 627) 1 MPOPSOC 0 Government did not restrict access to the internet. Both government-owned media and privately owned media had presence (BDHRL 2004(BDHRL , p. 10, 2006) LEMBNEU 1 EMB received praise for being less pro-government (FH 2007), better organized elections and increased transparency (EUEOM 2006, p. 12;COG 2006, p. 38;Larmer and Fraser 2007, p. 620), but given a lack of legal framework, capacity and bias it failed to act on resource abuse of MMD (BDHRL 2007, p. 14). It also failed to resolve election disputes in a satisfactory manner (EUEOM 2006, p. 9) 1 3 Executive and legislative had budgetary control of judiciary, and president could dismiss judges based on incompetence and choose new candidates with aid of MMD-controlled parliament (Gloppen 2003, pp. 125-127). Some evidence of patronage and corruption in judiciary (BDHRL 2003, p. 7;VonDoepp 2005, pp. 292-294).
In general the courts did not seem averse to ruling against the government on issues related to the playing field such as abuse of funds, and access to media (BDHRL 2003(BDHRL , p. 7, 2004(BDHRL , p. 7, 2005(BDHRL , p. 6, 2007. However, courts were much less likely to rule against state if important government actors such as president were involved and in cases where there were clear and decisive interests for ruling party (BDHRL 2004(BDHRL , p. 13, 2005VonDoepp 2006, p. 395). Courts protected opposition candidates on several instances during campaigns (BDHRL 2007, pp. 7-8) 1

Access to law = (1 + 1)/2 = 1
The playing field score = (1.25 + 0.5 +1)/3 = 0.9  Reports about clear abuse of government funds, especially in presidential election (BDHRL 2008(BDHRL , p. 12, 2009FH 2009;FH 2010). Use of government resources in campaigns (Cheeseman and Hinfelaar 2010, p. 70 Reports about threats of selective use of government programs (BDHRL 2008, p. 12). Big increase in fertilizer subsidies in period before and after election (Mason et al. 2013 MPUBMED 3 Self-censorship and direct government influence in public media, which were still the most important media outlets in print and broadcast (BDHRL 2008(BDHRL , pp. 8-9, 2009FH 2008). Explicit threats if they did not publish positive content on Bandah during MMD succession (BDHRL 2009, p. 8). Limited access for opposition to print and broadcast during campaigns (BDHRL 2009, p. 8;EISA 2010, p. 15) 1.5 MPOPCOP -1 Community radio stations broadcasting as in previous period, but most of them mandated not to broadcast political content (Willems 2013, p. 225). Nevertheless several talk shows with political content and popular participation, and opened up space for opposition (Murthy and Muzzamil 2010, p. 31; 1 MPOPSOC 0 Few restrictions on social media and internet use, but very limited outreach (BDHRL 2008(BDHRL , p. 9, 2009(BDHRL , p. 9, 2010) LEMBNEU 1 Elections were relatively well run, but lack of updated voter registry probably favoured the MMD (BDHRL 2009, p. 11;Cheeseman and Hinfelaar 2009, p. 70;EISA 2010, p. 13). Generally favourable review of EC in monitoring report (EISA 2010). Little or no attempt to sort out unfairness in playing field and resources (EISA 2010) 1 3 Some abuse of government positions to circumvent judicial procedures, but also several cases where court acted independently of government (BDHRL 2008(BDHRL , p. 7, 2009FH 2008). General increase in judicial independence and competence (FH 2009 given the relatively short timeframe between from the 2006 elections and the lack of parliamentary races, many aspects of the playing field are more likely to remain the same than during a normal electoral cycle. Some issues were of less relevance and importance because of either the short space for preparing for the contest or for the presidential nature of it. This must be taken into consideration when interpreting the results

Access to law = (1 + 1)/2 = 1
The playing field score = (2.8 +1 + 1)/3 = 1.6 Table 6 Indicators for the playing field in Zambia, 2008-2011 Indicator Score Why Importance RINTPAR 0 Parties report that internal contributions and membership fees are a source of income, but opposition complains that it is not sufficient to maintain party organization (COG 2011, p. 17). Parties generally free to operate and fundraise without restrictions (BDHRL 2011, p. 13). Top-heavy parties and little membership participation (Simutanyi 2013, p. 17). PF control over local councils contributed to increased evenness in quality and size of party organizations (Simutanyi 2013, p. 22) 0.5

RPRIWEL 1
Top leadership in MMD, UPND and PF were at this time either wealthy businessmen or former leaders in government, contributing heavily to the party (Simutanyi 2013, p. 17). Private businessmen typically supported both incumbent and PF, as the election was expected to be close, but typically the MMD was supported to a larger degree (Media reports and interview) 1.5

RPUBPAR -
No public funding of political parties 0 RPUBCAM -No campaign funding for political parties Misuse of government resources in by-elections throughout period (BDHRL 2010(BDHRL , p. 14, 2011. Advantage of incumbency exploited by president and MMD in election (BDHRL 2012, p. 13;COG 2011, p. 16;EUEOM 2011, p. 14). Contributed to the common perception that the MMD had the most financial leverage in electoral history in Zambia at the time (COG 2011, p. 17) 1.5

RILLRES 3
Use of state infrastructure and resources, including different forms of transportation for candidates and for supporters to attend rallies, both in by-elections throughout period and in national elections (BDHRL 2010, p. 14;BDHRL 2011, p. 18;COG 2011, p. 17;EUEOM 2011, p. 14) RILLAPP 3 Partisan behaviour of state employees. Provincial secretaries and district representatives campaigned on behalf of ruling party, showing their partisan loyalty (EUEOM 2011, p. 14). Growing sense of partisan and ethnic employment (Simutanyi 2013, p. 18) RILLPOL 3 President timed inauguration of new state projects to coincide with campaign period and used them for campaign purposes (EUEOM 2011, p. 14). Maize relief program used for partisan and campaign purposes (EUEOM 2011, p. 14) RFORGOV 1 China accused of supporting MMD, and Banda used inauguration of Chinese infrastructure projects for campaign purposes (Hess and Aidoo 2013, pp. 137-138). Sources in political parties also claimed Chinese state and non-state actors were contributing funds, both to incumbent and opposition (Rakner 2012, p. 10) 0

MPOPSOC 0
In 2008, only about 5.5 % of all country inhabitants used internet (BDHRL 2010, p. 11). While not restricting access, the government monitored critical internet outlets and sometimes acted on this information (BDHRL 2010, p. 11). Social media gradually increased in importance with regard to communication between media and population (Willems 2013, p. 226). Editor of online newspapers investigated and charged (BDHRL 2011, p. 13).