Social cohesion is a key concept in development studies. Weak social cohesion is often related to slow economic growth and (violent) conflict. So far few attempts have been made to measure this complex concept in a systematic manner. This paper introduces an innovative method to measure national-level social cohesion based on survey data from 19 African countries. We distinguish three dimensions of social cohesion; i.e. the extent of perceived inequalities, the level of societal trust, and the strength of people’s adherence to their national identity. Importantly, our Social Cohesion Index (SCI) is based on individuals’ perceptions vis-à-vis these three different dimensions of social cohesion rather than certain macro-level ‘objective’ indicators such as GDP/capita or Gini-coefficients. We develop two social cohesion indices: a national average SCI and a Social Cohesion Index Variance-Adjusted (SCIVA); the latter one takes into account the level of variation across different ethnic groups within countries. The SCI and SCIVA are computed for and compared across nineteen African countries for the period 2005–2012 on the basis of Afrobarometer survey rounds 3, 4 and 5. We also investigate quantitatively the relationship between countries’ levels of social cohesion and the occurrence of a range of conflict events. As expected, we find that countries with low levels of social cohesion in a particular year according to our SCI are more likely to experience a range of different violent conflict events in the subsequent year.
This is a preview of subscription content, access via your institution.
Buy single article
Instant access to the full article PDF.
Tax calculation will be finalised during checkout.
Generally, a positive association between what individuals perceive and what is actually observed has, indeed, been found. For example, using the Afrobarometer database we find that having more water, food and cash income is positively associated with perceiving one’s living conditions as more favourable.
Macro-level indicators of enclave living could be used to test relationships with social cohesion. We see this exercise being undertaken at the community level (cities, villages etc.) as enclave living is often limited to certain areas (e.g. major cities) of a country which might make it difficult to be picked up in nationally representative surveys. However, at the individual level there is already an established line of research focusing on contact theory: Does increased contact with other (cultural) groups also increase individual and intergroup trust? As trust is one of our major dimensions, these additional questions—currently not in the Afrobarometer—could be useful: e.g. how much do you engage with people from a different ethnicity, religion etc, and in what way (work, private life …). We could also expect a relationship between contact and the identity dimension.
This is similar to the inequality-adjusted Human Development Index of the UNDP. For more information: http://hdr.undp.org/en/content/inequality-adjusted-human-development-index-ihdi.
This can be measured by the Cronbach coefficient (see e.g. Deafys et al. 2011; Hooghe 2012; Lord and Novick 1968).
It is important to note here that survey samples are sensitive to different types of errors and inaccuracies, including for example measurement error due to poorly designed questions or interviewer effects (see a.o. Biemer et al. 1991). Given that the questions in this paper are generally simple and transparent, it is unlikely that poorly design questions have introduced serious bias into the analysis. On the other hand, interviewer effects could potentially create a bias in the questions regarding identity. As indicated by Berinsky (2004) survey data collection constitutes a social interaction. However, to what extent such interviewer effects may have introduced significant variation is not known. For more information on Afrobarometer surveys see: www.afrobarometer.org.
Designing ideal measurements for social cohesion requires quite elaborate empirical investigation. For example, we could see social cohesion as a second-level construct with inequality, trust and identity as first-level constructs. Each of the social cohesion dimensions could be measured with several items and analyzed via confirmatory factor analysis. The use of several items to measure our three dimensions of social cohesion would allow us to reduce measurement errors possible in some of the questions used here: the trust questions could be refined in terms of trust in whom (people from a different ethnicity, religion etc.) and in what situations (e.g. business transactions, private friendships), whereas nationality items could be designed to pick-up intergroup attitudes more precisely: in our country people should be educated in the dominant language; the existence of multiple cultural practices undermines/strengthens our nation etc. The difficulty in this approach is likely to lie in the necessity for our measurement items to have equivalent meanings in different countries (with different historical and cultural trajectories). Indeed, arguably the most difficult issue in designing cross-culturally comparable social cohesion indices lies in measuring the most politically salient cleavage in society: ethnicity, religion, migrants, caste etc. Furthermore, the use of multiple items might reduce errors in the measurements of our three dimensions, but a composite index of a limited number of items is generally much more straightforward to comprehend and analyze.
In the identity cluster, we restrict answers to individuals who feel a stronger affinity to a national identity than their own group identity. In a robustness test we also include those individuals who feel equally ethnic as national. The ranking of countries is similar to the one reported here. Also a similar variation over time is observed. More generally, we also perform robustness tests changing the cutoff points of all other questions (e.g. to include not just the extreme answer, but to include the two most extremes). Although minor differences can be observed, the overall ranking and variation over time seems to be a robust finding. This supports our belief that our index is a robust description of social cohesion.
Please note that we can only include countries for which we have data for more than one survey round. Hence even though the Afrobarometer Round 5 survey was conducted in 33 countries, in about 13 countries it was the first time that the Afrobarometer survey was conducted.
While the question related to ‘fair treatment’ is also framed around identities, this is not driving the significant correlation. Also a significant correlation exists between Identity and the ‘living conditions’ componenst of the inequality cluster.
This is also confirmed by the robustness results. Changing the way we compute SCI does not change the time-variation we observe. This suggests that variation over time is a key characteristic of SCI, and not a matter of measurement error.
This coefficient of variation, along with alternative measures of variation are reported in “Appendix 1”.
As measured as the one-year period after the last interview date in a specific national survey.
Amuwo, K. (2009). Dynamics of civil society in liberalising francophone Africa: A case study of Benin Republic, 1990–2008. Pretoria, South Africa: Africa Institute of South Afric.
Bareebe, G. & Titeca, K. (2012–2103). Personalisation of power under the Museveni regime in Uganda. In F. Reyntjens, S. Vandeginste & M. Verpoorten (Eds.), L’Afrique des Grand Lacs. Annuaire 2011–2012 (pp. 83–106). Paris: L'Harmattan.
Bates, R. (2006). Ethnicity. In D. A. Clark (Ed.), The elgar companion to development studies (p. 167–173). Cheltenham: Edward Elgar Publishing.
Bécares, L., Stafford, M., Laurence, J., & Nazroo, J. (2011). Social cohesion among different ethnic groups in the UK composition, concentration and deprivation: Exploring their association. Urban Studies, 48(13), 2771–2787.
Berinsky, A. (2004). Can we talk? Self-presentation and the survey response. Political Psychology, 25(4), 643–659.
Beugelsdijk, S., de Groot, H. L. F., & van Schaik, A. B. T. M. (2004). Trust and economic growth: A robustness analysis. Oxford Economic Papers, 56(1), 118–134.
Biemer, P. P., Groves, R. M., Lyberg, L. E., Mathiowetz, N. A., & Sudman, S. (Eds.). (1991). Measurement errors in surveys. New York: Wiley.
Bjørnskov, C. (2007). Determinants of generalized trust: A cross-country comparison. Public Choice, 130(1), 1–21.
Bloor, K. (2010). The definitive guide to political ideologies. Bloomington: Author House.
Bratton, M. (2008). Vote buying and violence in Nigerian election campaigns. Electoral Studies, 27(4), 621–632.
Brown, G. K., & Langer, A. (2010). Conceptualizing and measuring ethnicity. Oxford Development Studies, 38(4), 411–436.
Cederman, L.-E., Weidmann, N. B., & Gleditsch, K. S. (2011). Horizontal inequalities and ethno-nationalist civil war: A global comparison. American Political Science Review, 105(3), 478–495.
Cole, J. S. (1988). Social capital in the creation of human capital. American Journal of Sociology, 94, S95–S120.
de Haan, A., & Webbink, E. (2011). Social cohesion and development: Using cross-country data to understand social cohesion and development. In International conference on social cohesion and development. OECD, Paris.
Easterly, W., Ritzen, J., & Woolcock, M. (2006). Social cohesion, institutions, and growth. Economics and Politics, 18(2), 103–120.
Eifert, Benn, Miguel, Edward, & Posner, Daniel N. (2010). Political competition and ethnic identification in Africa. American Journal of Political Science, 54(2), 494–510.
Europe, C. O. (2007). Report of high-level task force on social cohesion in the 21st century—Towards an active, fair and socially cohesive Europe Brussels, Council of Europe.
Fenton, S. (2009). Malaysia: a very particular multiculturalism. Bristol: University of Bristol.
Foa, R. (2011). The economic rationale for social cohesion—the cross country evidence. In The inernational conference for social cohesion. OECD. Paris, OECD.
Gambetta, D. (1988). Trust: making and breaking cooperative relations. Oxford: Basil Blackwell.
Great Britain. Parliament. House of Commons. International Development Committee. (2014). Recovery and development in Sierra Leone and Liberia: Government response to the Commitee’s sixth report of session 2014–15. London: The Stationery Office.
Hooghe, M. (2012). Social cohesion in comntemporary societies: an update of theoretical appproaches. In M. Hooghe (Ed.), Contemporary theoretical perspectives on the study of social cohesion and social capital. Brussels: Konlinkelijke Vlaamse Academie van Belgie voor Wetenscheppen en Kunsten.
International and Ibero-American Foundation for Administration and Public Policies, F. (2011). Strategies for integrating social cohesion in public policies. Retrieved April 3rd 2013.
Knack, S., & Keefer, P. (1997). Does social capital have an economic payoff? A cross-country investigation. Quarterly Journal of Economics, 112(4), 1251–1288.
Kymlika, W. (1995). Multicultural citizenship: A liberal theory of minority rights. Oxford: Oxford University Press.
Langer, A., & Smedts, K. (2013). Seeing is not believing: perceptions of horizontal inequalities in Africa, CRPD Working Paper 16. Leuven: Centre for Research on Peace and Development (CRPD).
Marc, A., Willman, A., Aslam, G., Rebosio, M., & Balasuriya, K. (2013). Societal dynamics and fragility: Engaging societies in responding to fragile situations. Washington, DC: World Bank.
Maxwell, J. (1996). Social dimensions of economic growth, Eric John Hanson Memorial Lecture Series (Vol. VIII). Edmonton, Alberta: University of Alberta Press.
Mustapha, A. R. (2006). Ethnic structure, inequality and governance of the public sector in Nigeria. In Y. Bangura (Ed.), Ethnic inequalities and public sector governance. London: Palgrave.
OECD. (2008). Handbook on constructing composite indicators: Methodology and user guide. Paris: OECD.
Putnam, R. D. (2000). Bowling alone: The collapse and revival of american community. New York: Simon & Schuster.
Raleigh, C., Linke, A., Hegre, H. & Karlsen, J. (2010). Introducing ACLED-Armed conflict location and event data. Journal of Peace Research, 47(5) 1–10.
Rothstein, B., & Uslaner, E. M. (2005). All for all: Equality, corruption, and social trust. World Politics, 58(01), 41–72.
Rustad, S. A. (2015). Socioeconomic inequalities and attitudes toward violence: A test with new survey data in the Niger delta. International Interactions. doi:10.1080/03050629.2015.1048856.
Schmeets, H. (2012). Social cohesion: An integrated empirical approach. In M. Hooghe (Ed.), Contemporary theoretical perspectives on the study of social cohesion and social capital. Brussels: Konlinkelijke Vlaamse Academie van Belgie voor Wetenscheppen en Kunsten.
Stewart, F. (Ed.). (2008). Horizontal inequalities and conflict: Understanding group violence in multiethnic societies. London: Palgrave.
Tripp, A. M. (2010). Museveni’s Uganda paradoxes of power in a hybrid regime. Boulder, CO: Lynne Rienner.
Uslaner, E. M. (2008). Corruption, inequality, and the rule of law: The bulging pocket makes the easy life. Cambridge, New York: Cambridge University Press.
Warren, M. E. (Ed.). (1999). Democracy and trust. Cambridge: Cambridge University Press (CUP).
Zak, P. J., & Knack, S. (2001). Trust and growth. The Economic Journal, 111(470), 295–321.
We are grateful to Hiroyuki Hino for support and ideas, as well as to participants at seminars at the Kenya Institute for Public Policy Research and Analysis (KIPPRA) Nairobi and the University of Cape Town for their comments.
See Table 7.
Appendix 2: Definitions of ACLED Conflict Variables
The following event types are based on the ACLED dataset. Some event types are directly taken from the dataset, other types used in our analysis were constructed by limiting events to certain ‘actor types’.
ACLED defines a battle as “a violent interaction between two politically organized armed groups at a particular time and location.” Typically these interactions occur between government militaries/militias and rebel groups/factions within the context of a civil war. However, these interactions also include militia violence, rebel on rebel violence and military on military violence. There is no causality minimum necessary for inclusion.
The specific elements of that definition are as follows:
(1) A violent interaction is the use of armed force, including guns or military hardware, machetes, knives or any tool to inflict harm upon the opposing side.
(2) Organized armed groups including but not limited to rebel and government groups. (codebook 3, 2014, p. 9).
Depending on the outcome of a battle, ACLED distinguishes ‘Battle-No change of territory’, ‘Battle-Non-state actors overtake territory’, and ‘Battle-Government regains territory’. For the ‘Battle’ events used in our analyses, these types have been taken together.
This constructed event is a Battle which involves a government actor (Code 1 in the INTER1 or INTER2 variables).
This constructed event is a Battle which does not involve a government actor (Codes 2 (rebel force), 3 (political militia), 4 (ethnic militia) in the INTER1 and INTER2 variables).
Non-State Battle Between Communal Groups
This constructed event is a Battle which involves ethnic militias (Code 4 in the INTER1 and INTER2 variables).
Violence Against Civilians
Violence against civilians is defined as deliberate violent acts perpetrated by an organized political group such as a rebel, militia or government force against an unarmed non-combatant. These acts are political and harm or kill civilians, and are the sole act in which civilians are an actor. There is no minimum number of victims needed to qualify as an ACLED event.
Although the victims can be combatants in a different context, here they are UNARMED and NOT ABLE to defend themselves. One-sided violence also includes inflicting significant harm (e.g. bombing, shooting, torture, rape, mutilation etc.) or accosting victims (e.g. kidnapping and disappearances). It does not include incidents in which people are not physically harmed, (e.g. looting or burning, destruction of sacred spaces, and forced displacement.) (codebook 3, 2014, pp. 11–12).
Violence against civilians perpetrated by a government actor (Code 1 in the INTER1 or INTER2 variables).
A riot is defined as “a violent disturbance of the public peace by three or more persons assembled for a common purpose.” ACLED records reported information on both spontaneous and organized rioting. Organized riots can be planned by a previously recognized political group. The rioting group is not necessarily an inherently violent organization. A political party can riot (i.e. ZANU-PF in Zimbabwe). If the protesters or rioters are representing a group, the name of this group is recorded in the “ally” section. Spontaneous riots primarily involve civilians, without direct reference to an organized political group. Protests are nonviolent spontaneous organizations of civilians for a political purpose. Protesters do not engage in violence, and if violence occurs during a protest as a result of protesters “actions, this event is coded solely as a riot. If violence is done to protesters in the event of a protest, the event is coded solely as an act of ‘violence against civilians’”. (codebook 3, 2014, p. 11).
The Riots/Protests event type in ACLED is limited to Riots by only counting cases in which actors were defined as ‘Rioters’ (Code 5 in the INTER1 or INTER2 variables). Events are limited to the Riots/Protest event type to avoid double-counting as Rioters can also perpetrate ‘Violence against civilians’ in the dataset.
The Riots/Protests event type in ACLED is limited to Protest by only counting cases in which actors were defined as ‘Protesters’ (Code 6 in the INTER1 or INTER2 variables). Events are limited to the Riots/Protest event type to avoid double-counting as Protesters can also be the victim of ‘Violence against civilians’ in the dataset (e.g. in the case of government repression, see above).
Distinction Between External and Internal Events
All events were limited to events perpetrated by internal actors to ensure compatibility with the Afrobarometer surveys. The following procedures were followed:
For the event type “battle—no change of territory” data was browsed for 2 opposing national armies and/or police forces via the variables ACTOR1 and ACTOR2. These cases are dropped. This also includes national military forces battling mutinous forces of a foreign national army.
For all 3 battle categories, we filter out cases in which armed groups operate across the border of the country to which the group belongs. Government troops operating on foreign territory are considered as international cases, e.g. Mauritanese military in Mali; Congolese army in Uganda. However, events in which international actors (INTER1 or INTER2 = 8) assist in fighting an internal rebel group is considered as internal conflict (whether or not the government they are assisting is identified in the ALLY categories), e.g. the French military in Mali. If an organized rebel group (i.e. organization name such as Lord Resistance Army) operates outside of the country of origin, this is seen as international (e.g. Sudan’s liberation army operating in Uganda). Often we find ethnic militias or unidentified armed groups with a certain nationality between brackets. If the nationality of one of the armed groups differs from the country in which the event took place we decide that this is an international case.
For the category Riots/Protests we browse ACLED for events identifying one of the actors as “international” or with a nationality foreign to the country. This is based on the ACTOR variables or the NOTES variable. If foreign nationals protest or riot in a particular country, these cases are dropped. For the category “Violence against civilians”, we also browse ACLED for events identifying one of the actors as “‘international” or as having a nationality foreign to the country. If foreign civilians are victims of the violence, this is generally regarded as internal conflict (e.g. journalists, oil workers). However, if the perpetrators are not from the country where the event took place, the cases are dropped. Finally, as pirating can be regarded as an international crime, and as it is difficult to pinpoint the nationality of the pirates, all pirating cases are dropped.
About this article
Cite this article
Langer, A., Stewart, F., Smedts, K. et al. Conceptualising and Measuring Social Cohesion in Africa: Towards a Perceptions-Based Index. Soc Indic Res 131, 321–343 (2017). https://doi.org/10.1007/s11205-016-1250-4
- Social cohesion
- Measurement of multidimensional concept
- Perceptions-based index