Minorities at Risk Project

Research project

The Minorities at Risk (MAR) Project is an independent, university-based research project that aims to monitor and analyse the status and conflicts of so-called communal groups (i.e., cultural and religious minorities) around the world (Davenport 2003: 5). The project was initiated by Ted Robert Gurr in 1986 and has been based at the Center for International Development and Conflict Management, University of Maryland, since 1988. The dataset is widely used in studies of minority groups, e.g. Fearon and Laitin (1999), Gurr (1993) and Saideman et al. (2002).


MAR Data

MAR Data


Minorities at Risk Project






120 countries

Last reviewed


Data types and sources

Some of the variables in the database contain data from official statistics or similar sources, such as estimates of group size. The most important variables are, however, mainly standards-based data, coded by trained students.

Data download

MAR Quantitative Data


The MAR project tracks 282 politically active ethnic groups on political, economic, and cultural dimensions. The dataset contains a large number of variables that describe the status and activity of communal groups. The variables are organised in five main groups: (1) group characteristics and status, (2) group discrimination, (3) group organisation, (4) group collective interests, and (5) group conflict behaviour. See Davenport (2003) for a more detailed description of the variables. In September 2008 the MAR project introduced a subsidiary dataset called the Minorities at Risk Organizational Behavior (MAROB). Focusing initially on the Middle East and North Africa, the MAROB project provides information on the characteristics of ethnopolitical organizations, and seeks to identify those factors that motivate some members of ethnic minorities to become radicalized, to form activist organizations, and to move from conventional means of politics and protest into violence and terrorism.

Geographical coverage

The basic unit in the MAR database is not countries, but communal groups. The current version of the MAR monitors 282 politically active communal groups considered to be “minorities at risk” in some 120 countries throughout the world. Originally the concept referred to "an ethnopolitical group (non-state communal group) that: collectively suffers, or benefits from, systematic discriminatory treatment vis-à-vis other groups in a society; and/or collectively mobilises in defense or promotion of its self-defined interests” (Davenport 2003: 5). In 2004 the concept of a "minority at risk" was re-developed as apart of an effort to address issues of selection bias.In 2006, MAR decided to adopt far more inclusive criteria to identify communal groups around the world. These new criteria, which were used in the new 2004-2006 MAR release, move MAR away from its traditional criteria of discrimination and/or mobilisation and aim to capture all ethnic/communal groups in a country that meet a minimal population threshold. MAR has also decided to reduce the number of variables in future updates to include only a core of approximately 40 variables. This set of variables will also be back-coded for all new groups (MAR website, accessed 16 January 2007). See online available codebook the for the current definition.

The MAROB dataset covers 118 organizations representing 22 ethnic groups in 26 countries, providing information on 176 variables on an annual basis from 1980-2004. Variables included cover organizational characteristics, state-organization relations, external support, and organizational behavior (nonviolent, violent and criminal).

Time coverage and updates

MAR: 1940-2003 and 2004-2006. The latest addition was released in February 2009.
MAROB: 1980-2004.


The dataset is described in a fair amount of detail in the online available codebook. The MAR project has not published any overview of the sources used to construct the indicators, but individual researchers can gain access to some of the documentation by arrangement with the project coordinator.

Access conditions and cost

Available free of charge. Registration required.

Access procedures

The 2004-2006 MAR release can be downloaded in CSV, STATA or SPSS. The 1940-2003 MAR dataset can be downloaded as an ASCII file. However, the full dataset may be cumbersome and difficult to manipulate. The best way to access the data is therefore through the MARGene program, which has been designed specifically to allow easy access to data in the MAR dataset. Through MARGene, scholars can access variables, create subsets of the data, interpolate across unobserved data points and set up the data for merging with other datasets. The program also allows users to specify subsets of the data based on time and space, and specify various options concerning data creation. The program then creates a new, customised dataset that can be loaded into statistical analysis software such as Stata, SPSS or Limdep for further manipulation and/or analysis (Bennett and Davenport 2007: 1). The MARGene program can be downloaded after registering on the MAR website. Though the program is designed to make it easier to access the data, users must be prepared to spend some time to understand how it works.

The MAROB dataset can be downloaded in in ASCII, SPSS and STATA formats.

Data formats

MAR 1940-2003: MARGene software or ASCII.
MAR 2004-2006: DTA, CSV, STATA or SPSS.

Comparability and data quality

One of the main strengths of the MAR project lies in its focus. According to Gurr (1993: 162), “No international body certifies, counts, or records statistics on communal groups.” MAR thus fills a gap in social science statistics and is, in the words of Landman and Häusermann (2003: 24), “a unique effort to study the state of minority groups around the globe”. Another strength of MAR is its fairly long time scope, with some time-series stretching back to 1940. The significant changes in data collection methods in 2004 should be noted. However, project staff are now reviewing past coding systematically in order to release a single, integrated dataset coded on an annual basis from 1980 through 2006 (MAR homepage 2009).

Earlier the criteria for including a group in the dataset were quite restrictive: only groups that collectively suffered or benefitted from discrimination and/or are collectively mobilised, were included in the MAR project. The dataset therefore suffered from a selection bias, since many possibly relevant groups – e.g., ethnic groups that are not discriminated against and that are not mobilised politically – were not included in it (Hug 2003: 256). In order to adress this, the criteria were redeveloped for the 2004-2006 update. See the MAR codebook for details. The problem of selection bias is thereby reduced, even though any criteria used to identify communal groups will involve some bias.

A second problem concerns the reliability of the data. The sources used to code the variables are not made easily accessible to independent researchers, and it is therefore difficult to replicate the results. In addition, the project has not yet assessed inter-coder reliability. And finally, as all standards-based human-rights indicators, the data may suffer from validity problems. The quality of the standardised scales depends in large measure on the quality of the primary data. Accordingly, if there are any deficiencies in the primary sources, these will be reflected in the standardised scales.


Bennett, D. Scott, and Christian Davenport. 2007. MARGene: Minorities at Risk Data Generation and Management Program. V1.05, Documentation. University of Maryland.

Davenport, Christian. 2003. Minorities at Risk. Dataset Users Manual, 030703. University of Maryland.

Fearon, James D., and David D. Laitin. 1999. “Weak states, rough terrain, and large-scale ethnic violence since 1945”. Prepared for delivery at the 1999 Annual Meetings for the American Political Science Association, 2-5 September 1999, Atlanta, Georgia.

Gurr, Ted Robert. 1993. “Why minorities rebel: a global analysis of communal mobilization and conflict since 1945”. International Political Science Review 14 (April): 161-201.

Hug, Simon. 2003. “Selection bias in comparative research: the case of incomplete data sets”. Political Analysis 11 (Summer): 255-274.

Landman, Todd, and Julia Häusermann. 2003. Map-making and Analysis of the Main International Initiatives on Developing Indicators on Democracy and Good Governance. Report prepared for Eurostat. Colchester, UK: Human Rights Centre, University of Essex.

Saideman, Stephen M., David J. Lanoue, Michael Campenni and Samuel Stanton. 2002. “Democratization, political institutions and ethnic conflict: a pooled time-series analysis, 1985-1998”. Comparative Political Studies 35 (February): 103-129.