Skip Navigation

The Role of Leaders in Democratic Deliberations

by MACARTAN HUMPHREYS, WILLIAM A. MASTERS, and MARTIN E. SANDBU. Reading for Tuesday, 18 March 2008.

STRONGLY participatory approaches to political decision making have been widely promoted in a bid to deepen the practice of democracy.  Supporters of these approaches attribute multiple benefits to participation, on normative, substantive, and instrumental grounds.  However, there is to date little or no evidence that these practices in fact return the benefits attributed to them. This is of particular concern, since there are strong arguments from the political and anthropological literatures that suggest that such processes may be subject to various forms of manipulation.

This article reports on a field experiment that provides an unusual opportunity to examine the extent to which participatory processes of this form are in practice vulnerable to manipulation by political elites.  The occasion for the study was an exercise in deliberative democracy in the small African island state of São Tomé and Príncipe. In 2004 the country held a national forum in which citizens gathered in small groups throughout the country to discuss topics related to economic policy priorities for the country. In many ways the consultations, open to every adult in the country and managed in part by the United Nations Development Program (UNDP), were a model of consultative practices in developing countries.

Unlike other such consultation processes, however, the organizers introduced a random element to the design of their consultation: the leaders who moderated the discussions were assigned randomly to groups throughout the country.

This feature, along with a formal reporting mechanism on outcomes of discussions, affords a unique opportunity to identify the impact of leaders on the preferences expressed by groups. While we cannot identify the difference between outcomes that would obtain with and without leaders, we can exploit random variation in the identity of leaders to establish that leaders matter: we do this by showing how the identities of leaders determine outcomes.  Identifying the impact of discussion leaders is usually rendered difficult by a selection problem: leaders’ characteristics could be linked to discussion outcomes because of the criteria used in choosing leaders, rather than because of any independent influence of the leaders as such. But we are able to overcome this identification problem because of the randomization introduced into the Saotomean forum.

The findings are striking. Even though in many ways the deliberations in São Tomé and Príncipe were held in an ideal communication environment, there is robust evidence that the influence of leaders on the outcome of deliberation is extremely strong, with leadership effects accounting for a large share of the variation in views elicited across the country. If similar dynamics are at work in such consultations elsewhere, then our evidence suggests that these participatory practices may not return the benefits so commonly attributed to them by civil society groups, governments, and international organizations.

Our examination proceeds as follows. In Section II we describe the nature and rationale behind participatory decision processes, as well aswhy we might expect them to be vulnerable to political manipulation that would thwart their aims. In Section III we describe the political context of the deliberations in São Tomé and Príncipe, and in Section IV we discuss the structure of the deliberations in more detail. In Section V we present the main results of our analysis, estimating the degree of leadership bias and determining systematic correlates of bias.  In Section VI we move from identification to explanation and seek to relate the outcomes of the discussions to rival mechanisms that can explain leader influence. Section VII addresses the question of the external validity of our results and Section VIII concludes.

Download File: humphreys.pdf

Center for Comparative and Global Research