Mission Statement

This project studies leadership accountability in polities with multi-party elections around the world.  Its core ideas and empirical approach can be summarized in a few propositions.

  1. Electoral democracies, unlike other political regimes, institutionalize the accountability of political rulers to their citizens through elections under conditions of civil as well as political liberties. Electorally successful candidates pass the accountability test. But what are the kinds of activities and information conveyed to voters that make politicians pass that accountability test?  Read More
    Electoral democracies, unlike other political regimes, institutionalize political accountability of rulers to their citizens. Democracies institute the periodic (re)election of political leaders under conditions of universal suffrage with civil and political freedoms enjoyed by all citizens. Citizens hold candidates for political office—and entire political parties as brand associations of electoral office-seeking politicians and their close supporters—accountable through their vote choice and their public deliberations about politicians’ performance. Citizens direct their votes toward those candidates who in their eyes pass the accountability test better than other contestants.Observation of candidates’ activities and informative signals constitutes the grounds for voters to sort political ballot alternatives into those that do or do not pass the accountability test. But what kinds of information and observable activities make voters endorse a candidate or party? What thereby motivates them to establish “linkage” to their favorite party through electoral support? The activities and information politicians convey to voters in order to obtain support vary in their effectiveness across individual voters, aggregate electoral constituencies, political parties, countries and historical time periods. It is therefore important to determine what exactly it is that voters honor and that politicians supply in order to be successful.Moreover, from an aggregate perspective, the overall level and profile of democratic accountability supplied by politicians may vary across countries and time periods. Indicators such as voter turnout, electoral volatility of party support, party system fragmentation and party turnover as well as “soft” indicators of trust in politicians and democratic institutions—may signal differential accountability patterns.

    DALP contributes to the measurement of accountability relations between citizens and politicians in polities with contested multi-party elections. Not all of them, however, provide for a level playing field for all candidates and parties in the electoral contest. The polities included in DALP extend from liberal democracies with firm civil and political freedoms all the way to hybrid regimes with multi-party electoral contests, but opportunities to win skewed in favor of hegemonic ruling parties.

  2. The political science discipline has amassed information about the variety of considerations voters employ to choose among candidates and parties in competitive elections. But scholars know a great deal less about how parties and politicians choose and focus their activities and informational signals to influence that vote choice. DALP data help to fill a supply-side gap to map the variety of linkage efforts that parties may undertake to attract electoral support. Read More
    Legions of surveys have enabled political scientists to develop vote choice functions across most democratic polities around the world. We know that voters may be influenced by a wide range of party/candidate considerations, starting from habitual support for the same known figure or label and party identification, through descriptive trait similarity of voter and candidate, targeted conditional material benefits to individual voters and small groups (“clientelism”), retrospective performance evaluation of political incumbents often based on incumbent valence performance (economic growth, employment, law and order), personal qualities of the candidates (“charisma”), and finally partisan positions on the provision or distribution of local pork, large-scale club and universal collective goods in a polity in light of voters’ own broad ideological preferences (more so than detailed policy knowledge).There is much less systematic information about the efforts political candidates and parties themselves undertake to demonstrate their worthiness to pass the voters’ accountability test. Here DALP provides a “party centered” supplement to the “voter centered” electoral choice and party literature on democratic accountability and linkage relations, with an emphasis on supplying information about clientelistic linkages, but also including a modicum of information about other linkage mechanisms.

    The emphasis of DALP’s expert survey is to collect data on the incidence and organization of targeted, conditional exchanges between politicians and voters typically referred to as “clientelism.” Clientelism comes in single spot-market and durable relational transactions. It may involve an exchange of individual politicians or parties, on the supply side, with individual citizens or small groups of citizens, defined by kinship, residence or social network attributes, on the demand side. Preceding DALP I, there was no systematic cross-national dataset whatsoever benchmarking parties’ clientelistic partisan efforts and the perceived electoral impact of that effort. Until DALP II, no equivalent data effort has ventured to update the earlier study to provide more inter-temporal depth to explore the dynamics of party competition with clientelism as one accountability mechanism.

    DALP is not limited to the study of clientelism, but recognizes the multiplicity of linkage mechanisms involved in democratic accountability. It therefore also includes a battery of questions about parties’ appeals on fundamental policy dimensions. In that vein, DALP I offered the geographically most inclusive dataset of parties’ policy positions around the globe in 2008.  Due to the comparatively large number of experts scoring each party (12-80), it allows for a measure of expert uncertainty of parties’ programmatic positions and thereby the construction of a rough index of parties’ programmatic coherence. Programmatic coherence of parties is a concept that has often been invoked, but never operationally measured in previous studies of party competition. DALP I and DALP II thus complement in geographic and analytical terms existing datasets on parties’ programmatic positions and/or salience of such positions, particularly the geographically narrower Chapel Hill Election Surveys (CHES) and the Comparative Manifestos Project (CMP) and its successor (MARPOR). Unlike the latter, however, DALP does not offer a sustained time series, but only two points of observation in 2008 and 2020.

    Just as voters may take a range of other criteria into account than parties’ programmatic and clientelistic efforts, also parties’ and politicians’ supply of activities and information is not limited to these two linkage mechanisms. DALP includes at least one measure on a variety of additional considerations that voters may take into account in their vote choice and that parties and politicians may influence with their own initiatives.

    DALP I provides at least one measure each on the extent to which parties rely on unique attributes of individual political personalities in their linkage efforts (“charisma”). It also includes an assessment of parties’ capacity to project competence to govern and deliver economic and social stability, a sort of incumbency valence measure. Furthermore, it indirectly allows to track the relevance of descriptive representation for parties’ accountability efforts by including a battery in which experts indicate each party’s associational proximity to religious, ethnic, racial, and regional organizations. Finally, the survey invites experts to assess the parties’ efforts to mobilize affective partisan bonding through symbolic and ritual activities, a supply-side production of “party identification.”

    In addition to replicating the measures included in the earlier study, DALP II extends the range of data collected on varieties of linkage mechanisms. It includes a question on unconditional targeted benefits at the individual level (“constitutency service”) and the collective level (“pork”). It also asks experts to assess the extent to which parties and politicians engage in populist rhetoric. 

  3. Thus DALP data make it possible to study not just individual linkage strategies of parties, but the complex “profiles” of linkage efforts that parties deploy in the pursuit of votes and office. Read More
     Much political science research has focused on studying individual linkage mechanisms in isolation. A prime example is the responsible government theory that inspires almost all of normative democratic theory and much empirical democracy research as well. It focuses on the relationship between parties’ and voters’ preferences over positions on broad ideological dimensions that inform politicians’ choices of concrete policies allocating club and collective goods, whether in such areas as social and economic policy, law and order, citizenship, or ecological protection.The theory has come under attack primarily in the behavioral political psychology literature about voters’ information processing, generating evidence that other considerations are often more important for voters’ appraisal of a candidate’s or party’s accountability. Examining the demand side of voter rationales, programmatic party signals are one of several cues citizens may use in choosing a party in the electoral contest. It is critical to understand the mix and interplay of different inputs in voters’ electoral choice process. Among supply side studies of parties’ appeals to voters, however, there is preciously little research that would recognize the multiplicity of linkage mechanisms that politicians attend to. Most investigations focus on one mechanism only—programmatic or clientelistic. Studies rarely take into account complex profiles of partisan efforts to make themselves attractive to voters by supplying specific profiles of the whole gamut of linkage mechanisms. DALP data make possible this broadening of accountability studies and going beyond the simple responsible partisan government model on the supply side of party competition.

    An often postulated, but never tested hypothesis is that there is a zero-sum relationship between the weight parties and voters attribute to programmatic policy considerations (positional or issue-emphasis) and the weight they assign to clientelistic targeting. DALP I data, however, can show that this hypothesis is only contingently correct.  There is a subset of polities and parties that defy the simplifying trade-off thesis and actually engage in broad “linkage diversification” that combines vigorous clientelistic and programmatic efforts under the umbrella of one and the same party and may fold other non-programmatic linkage mechanisms in as well.

    Hence, modeling politicians’ investment in linkage strategies, the appropriate unit of analysis may not be this or that individual linkage type and its consideration by politicians. Given voters’ selective, complex and varied attention to politicians’ activities, politicians have to choose over the whole set of linkage mechanisms, thus often making the “profile” of linkage commitments—rather than the investment in any individual linkage mechanism—the appropriate unit of analysis, when describing and explaining party strategies.

  4. DALP provides a macro-level dataset on linkage strategies that fills a unique niche of investigations not serviced by more fine-grained, micro-level data on political transactions between voters and politicians, whether in an observational or an experimental mode.
    Read More
    Political science research typically proceeds from macro-level correlations between phenomena based on observational data as a first step. Here aggregate data about institutional rules, resource distributions, and aggregate behavioral patterns will suffice. But research then tends to move into more micro-level observational and experimental studies to identify the causal mechanisms at the level of individual actor cognitions, preferences and strategic moves that make possible the generation of macro-level configurations as aggregate effects.In the supply-side party- and candidate-centered political linkage literature, however, by and large the first step to map parties’ deployment of political linkage mechanisms has never been mapped, because comparative data on linkage mechanisms had never been collected before DALP I. Nor have hypotheses ever been thoroughly tested empirically that are widely accepted throughout the community of scholars dealing with party competition. One example is the presumption that in the mix of parties’ accountability efforts the role of clientelism declines in a linear trajectory with a polity’s rising economic development. At the individual level, clientelism is primarily a linkage mechanism for poor people, but not the middle class of a society. Another widely accepted, but only selectively demonstrated, hypothesis is that candidate-centered electoral systems promote clientelism. A third popular claim is that “young” parties and party systems tend to be more clientelistic. DALP data make possible a closer empirical look at such hypotheses across polities and individual parties. Even a cursory review of the data actually reveals that common wisdom in the scholarly community is often misleading and not corroborated at all, or only in highly conditional terms, by the evidence of the DALP data.By offering another data point for many of the same parties and polities experts scored in 2008, DALP II will be an important step forward to understanding the dynamics of change in parties’ deployment of linkage mechanisms over time. While DALP’s observational data will not provide direct causal leverage, cross-national and intertemporal correlations can nevertheless throw light on the kind of theoretical arguments that are more promising for a causal investigation. Macro-level correlational analysis may identify the kinds of questions that deserve deeper analysis and investigation of micro-level mechanisms that may underlie them. Without guidance by interesting macro-level regularities, micro-level studies may venture into projects with little broad relevance. Moreover, they may get bogged down in a cacophony of contradictory findings, not realizing that the contradictions may be due to scope conditions such that the micro-level mechanisms work differently contingent upon macro-level conditions and constraints.

    As an example, consider investigations of clientelistic practices around the world, the linkage strategy to which the greatest range of variables is devoted in DALP. There has been an explosion of micro-level investigations of clientelism with a focus on the organizational and social network conditions that enable voters and politicians to perform clientelistic transactions in local settings. Moreover, micro-level studies probe into the informational processes that make voters and politicians adopt or abandon clientelistic exchanges. In this quest, a great deal of research has paid attention to the role of intermediaries (brokers) facilitating the clientelistic exchange between citizens and electoral politicians. While many of these studies are meticulously researched and can draw strong inferences about the operation of mechanisms within the settings for which data were collected, as a set, they have generated a proliferation of contradictory findings. Do brokers have superior knowledge of potential clients’ preferences and behavior or do they not? Are brokers instrumental in delivering benefits (or sanctions?) to clients or are they not? What other role may brokers play in clientelistic linkage mechanisms, if not delivery and monitoring of exchanges? Correlational research with coarse, more highly aggregate data comparing linkage efforts across parties, countries and time, with different structural and institutional features, may indicate how and why findings on the same micro-level relations contradict each other, when sampled in different contexts. 

  5. Macro-level data sets such as DALP may be indispensable to answer questions about broad shifts in parties’ and entire party systems’ linkage profiles across time and space. Micro-level studies are uniquely limited in addressing such questions. This applies particularly to the rise or decline in the prominence of specific partisan linkage strategies, such as clientelism, in the linkage portfolios of party systems.
    Read More
    One of the big questions that inspires DALP data collection is the challenge to identify the economic, social, institutional and political conditions under which clientelism and/or programmatic partisan appeals become important in parties’ linkage profiles. Where and when does clientelism prevail over programmatic appeals, if there is such a trade-off, and where/when the latter? Where can politicians jointly deploy clientelism and programmatism in order to build voter linkages (no trade-off)? What are the broad constraints and parameters and settings in which politicians can use micro-level informational and incentives strategies to promote or dissuade voters from opting into clientelistic politics?For at least two reasons micro-level studies have difficulties in addressing these macro-questions in a systematic fashion. One concerns the problem of scaling-up micro-level relations to macro-level aggregate processes. The other has to do with the political “reflexivity” of research on linkage mechanisms itself.First, there may be no one-to-one relationship between micro-level mechanisms and macro-level relations. If a structural setting is conducive to the rise of clientelistic relations, politicians may choose among a whole host of micro-level techniques to achieve that outcome. Diverse local techniques are causally and functionally equivalent. The aggregate pattern will not depend on the viability of any specific informational, network, incentive etc. strategy, for example, to promote or depress the effort parties make in crafting clientelistic linkages to electoral constituencies. Conversely, when the structural configuration is unfavorable, few, if any micro-level techniques will work. Evidence that the presence of specific micro-level techniques enables or thwarts specific linkage strategies, by itself, will always be of rather subordinate relevance, as others may take their place elsewhere with similar structural conditions. Politicians’ failure to deploy a particular mechanism, by itself, will not prevent broad aggregate patterns in the deployment of linkage mechanisms, as long as structural conditions are conducive. In other words, localized practices cannot be “scaled up” to account for the macro-level outcomes.

    Second, there is the challenge of political “reflexivity” in research on micro-level mechanisms. Reflexivity means that a research finding itself affects the actions of research subjects such as to confirm (self-fulfilling prophecy) or disconfirm (self-destroying prophecy) its further operation. Or powerful actors will simply make a mechanism cease operation, if its widespread application were to threaten the political status quo. Consider informational studies about techniques to undercut clientelism, e.g. by offering information about the incidence of clientelism, engaging candidates in non-clientelistic campaigns, or by attempting to persuade voters through public broadcasts and townhall meetings that clientelism is detrimental to democratic politics and collective goods provision. A localized manipulation of voter information (e.g. about bribery accusations of incumbent politicians) that undercuts the operation of a clientelistic linkage mechanism favorable to the dominant political powers of a polity may itself not register with the powers that be, even if that investigation leads to the conclusion that manipulating voters’ informational environment indeed will change their linkage habits. As soon as someone, however, tried to utilize the same type of manipulation of voter information to undercut clientelism of powerful parties and politicians on a grand scale, she would run into political resistance. Politicians themselves would perceive the research as a political intervention dedicated to changing the power structure.

    Deploying an informational technique to alter political transactions and linkage strategies, therefore, may demonstrably work in a micro-setting. But this finding will have little implication for macro-level aggregate processes: Researchers would have to identify the macro-conditions that prevent and incapacitate status-quo oriented political incumbents from opposing the activation of such micro-level linkage techniques intended to undercut their power positions.

  6. The DALP macro-level data set creates a bridge to political economy research about the conditions and consequences of affluence, growth, and distributive arrangements. Widely held, but untested beliefs about the relationship between democratic linkage mechanisms and political economy may turn out to be inaccurate, or at least only conditionally corroborated by the evidence provided by DALP data.Read More
    It is a fairly widespread presumptions that democracies in which politicians rely a great deal on clientelism deliver bad political outputs and outcomes. They underinvest in large club and collective goods that cannot be targeted to finely divided constituencies. As a consequence, they suffer from weak economic growth, low state capacity, high household inequality and all sorts of undesirable quality of life outcomes in terms of child mortality, life expectancy, female empowerment, literacy and educational competence, as well as “soft” cultural outcome variables, such as interpersonal trust, confidence in democratic institutions, endorsement of fair process, and allegiance to universalistic democratic tolerance for disagreement and individualism.Because of the absence of macro-level comparative data, preciously few studies have been able to address these questions, let alone provide convincing answers. At least at a correlational level, DALP data make it possible to probe into the empirical plausibility that such relations exist and pinpoint some plausible causal channels that might underlie them and then await further micro-level investigation.
  7. DALP data are based on expert surveys of 15-50 jury members per country. While expert surveys have obvious downsides in terms of reliability and validity, they also offer unique advantages in gathering information about hard-to-observe and difficult-to-operationalize important political phenomena. Read More
    The research instrument employed in DALP is an expert survey that covers the extent to which democratic accountability proceeds through exchanges based on the provision of broad club and collective goods and on goods targeted to individuals and small groups in detail. It also more peripherally touches upon a whole host of other more affective linkage strategies. Unlike party manifestos for programmatic party appeals, there is no readily available documentary source based on which the political use of many non-programmatic linkage mechanisms could be observed and measured. The use of country experts is thus a first step.Experts who complete the survey are scholars who know about their country’s parties, campaigns, and elections because of their academic university training. Most of them also teach the subject in higher education, and some conduct their own research on phenomena covered by the DALP survey. Most of experts professionally reside in political science departments of their country’s universities, but they may also be affiliated with law faculties, sociology, history or public policy departments or work in independent research institutes or non-profit agencies and interest associations. A few experts may have spent years of training and research in the country they are asked to cover in the DALP survey, but now live elsewhere.In each country, between 12 and 30 experts completed the survey. Special arrangements with more numerous expert juries apply to very large compound republics (India, Nigeria). Under DALP auspices, no identifying individual information about socio-demographics or professional affiliation was collected from the experts. But they were asked to indicate their level of confidence in scoring the variables for each survey section. Moreover, experts were asked to indicate their degree of sympathy—for whatever reason—with the political parties they scored in the course of the survey.
  8. DALP strives for a comprehensive coverage of countries, but requires a modicum of competitiveness for a polity to be included.Read More
    In DALP I 2008-09, data collection included all polities of at least two million inhabitants with a minimum recent experience of two rounds of national electoral competition among two or more parties under at least semi-democratic, partly free conditions. The latter were identified in terms of average Freedom House civil and political rights scores of at least 4.0. Beyond this core set of countries, several slightly smaller democracies were included (e.g. Estonia, Latvia, Mauritius). More importantly, a few prominent additional countries with lop-sided multi-party electoral politics that did not quite meet the civil and political rights threshold were also included (e.g. Angola, Bangladesh, Egypt, Morocco, Russia), as well as a few democracies that fell slightly short of reaching the two million inhabitants threshold. In 2008, DALP covered 21 countries in the Americas, 12 countries in the Asia-Pacific region, 19 countries in post-communist Eastern and Central Europe, 16 countries in Western Europe, and 20 countries in Africa and the Middle East for a total of 88 countries.In DALP II, for 2020-21, we have selected countries by the same criteria and in total, we are aiming to conduct the study in 97 countries. 20 countries in the Americas (-1 from DALP I: Venezuela out), 15 in Asia-Pacific (+3: Sri Lanka, Singapore, Nepal), 19 in post-communist Eastern and Central Europe (no change), 17 countries in Western Europe (+1: Bosnia-Hercegovina), and 26 countries in Africa and the Middle East (+6; Burkina Faso, Malawi, Cote d’Ivoire, Lesotho, Liberia, Sierra Leone).
  9. The project has been designed and is implemented by a group of researchers in Duke University’s political science program under the direction of Herbert Kitschelt, George V. Allen Professor of International Relations. For further details of researchers who have worked in this project over the years, visit Duke Team Profiles.Read More
    In 2008, for the Latin American countries, the project proceeded in collaboration with the Institute of Political Science, Catholic University of Chile, Santiago de Chile, as module of a more encompassing investigation into party competition in the region, entitled “Cristalización Programática de los Sistemas de Partidos Latinoamericanos: Congruencias, Desafecciones y Calidad de la Democracia,” designed and implemented by Professors David Altman and Juan Pablo Luna.In 2020, the project benefits from the participation of two post-doctoral fellows at Duke University, Dr. Kerem Yildirim (PhD 2016, Koç University, Istanbul) and Dr. Jeremy Spater (PhD 2019, Duke University).
  10. Collection of the data on citizen-politician linkages and accountability relations has been funded by different sources. DALP I data gathered by this project is already made publicly available on this web site. DALP II data will be also available on this web site.
     In DALP I and DALP II Duke University supports the design of the project, the data collection and the data analysis through Herbert Kitschelt’s research funds.A substantial group of graduate students in comparative politics at Duke University contributed to the design and implementation of the DALP I project. All operational tasks in the process of data collection outside Latin America were handled through Duke University. Second, for DALP I, under grant No 1060749, the Chilean FONDO NACIONAL DE DESARROLLO CIENTIFICO Y TECNOLOGICO (FONDECYT) supported the Latin American project directed by Professors David Altman and Juan Pablo Luna of the Political Science Institute at the Catholic University of Chile to implement an expert survey, a module of which is that fielded in other regions of the world by the Duke research group. Third, the World Bank, through a grant extended to its Development Research Group, led by Phillip Keefer, joined this effort in fall of 2007 and supported the implementation of the expert survey with a data collection grant.