This dissertation explains why certain African politicians are able to create ethnic voting

blocs whereas others fail. Challenging current wisdom, I argue that this ability is conditional on the level of "groupness" cultivated by ethnic elites through ethnic associations. These historically rooted associations are involved in defining ethnic boundaries through social and economic activities that cultivate a sentiment that encourages group members to see themselves as sharing not only cultural and linguistic characteristics but also as having life chances that are intertwined with those of co-ethnics (linked fate) and their co-ethnic political leaders. These associations then use these sentiments as the basis upon which the group is mobilized politically and the political elite coordinated under a single banner. The varying success with which these ethnic associations are able to carry out these tasks explains contemporary differences across groups in cohesion, party entry, turnout and party competition.

The argument's causal logic is developed –and its mechanisms highlighted – throughout the

dissertation using evidence from Kenya's politically relevant ethnic groups. In the introductory

chapter, I present the context in which ethnic associations operate in sub-Saharan Africa. This is followed by an exposition of the groupness theory and its composite parts in Chapter 2. I show how ethnic associations, with varying degrees of success, are responsible for marking the ethnic group's boundaries; influencing how individuals think of their position within the group and their relationship to both each other and their co- ethnic leaders; and politically mobilizing the group and its elite based on these perceptions. This theoretical discussion is accompanied by empirical case studies of each of Kenya's politically relevant ethnic groups –Kamba, Kikuyu, Luo and Luhya–, their history with ethnic associations and these associations' involvement in the creation and maintenance of groupness. The process tracing approach used in this chapter addresses the model's potential endogeneity problem. Use of the method allows me to show that despite emerging from and responding to similar historical events – the creation of ethnic associations that provided a social safety net and mediator in response to colonial rule– these associations have had varying levels of success in coordinating their co-ethnics, their political elites, and influencing the type of party competition experienced.

The third chapter then examines the theory's primary observable implication: if ethnic associations have historically exercised this power over ethnic group sentiment and political organization, then there ought to be observable and systematic differences in the level of groupness across groups. Using ordered logit models, the analysis not only reveals systematic differences across groups in the beliefs in groupness but also that these differences correspond to the particular ethnic group's history with ethnic associations.

The dissertation's remaining chapters shift the focus of analysis from the individual to the

group. They explain how district variation in groupness --- an indicator created using multi-level regression and post-stratification (MRP) – explains party entry, turnout and competition at the constituency level. In Chapter 4, I address the question whether of groupness influences the entry decisions made by politicians and their parties. I argue that parties are constrained

by the level of groupness; that is, high groupness limits the political space for entrepreneurial

co-ethnic politicians to emerge. The results of the quantitative analysis support the hypothesis: constituencies embedded in districts with higher levels of groupness tend to see lower candidate entry rates for parliamentary seats.

Chapter 5 investigates the impact of groupness on individual turnout decisions. It argues that

groupness, rather than mobilizing higher turnout, actually leads individuals to engage in vote

approximation, which has a suppressive effect. That is, individuals believe that due to the high level of groupness within the group, political preferences among members will tend to be similar. Such a belief leads them to make the rational calculation, given the high costs associated with turning out to vote in these low information settings, that their co-ethnics will vote in a similar as they would, if they were not to cast a ballot. The results show that indeed places with higher levels of groupness, do experience lower turnout levels compared to areas with lower groupness levels.

The final empirical chapter considers how the concept of groupness and its variation across

groups affects the level of party competition. The argument states that in places where groupness levels are higher, political competition will tend to be lower, as proxied by the vote margin with which the winner wins. This is because voters are and the political elite are coordinated both by their ethnic associations and the groupness sentiment that exists among them. As a result, voters are aware of the association's preferred candidate and party. The quantitative analysis reveals that areas where groupness levels are high, candidates tend to win with higher margins.