This dissertation is comprised of three papers, which aim to improve our knowledge of how democracy and legislature size matter for ethnic representation among political leaders. The first paper introduces the Global Leadership Project (GLP), which provides the first dataset to offer biographical information on an array of leaders (i.e. members of the executive, the legislature, the judiciary and other elites) around the world. Personal characteristics and identities of political leaders influence their behavior in elective office, and thus carry implications for the course of politics and policy. The GLP encompasses 145 nation-states and 38,085 leaders, each of whom is coded along 31 parameters, producing approximately 1.1 million data points in a cross-sectional format centered on 2010-13. This data allows comparison of the demographic characteristics (i.e. gender balance, age, ethnicity, education, languages spoken, education, and tenure) of leaders within countries, across countries, and across regions. The second paper examines the causal mechanism of how democratization increases ethnic descriptive representation through a longitudinal case study of Kurdish representation in the Turkish parliament from 1920 until 2011. It argues that ethnic descriptive representation increases in a competitive democracy because out-parties collaborate with ethnic groups to gain electoral advantage over their rivals. In Turkey, the collaboration of emerging out-party actors with the Kurds explains the rise of Kurdish descriptive representation. Through process tracing, this paper examines this collaboration, explaining the precarious increase in Kurdish descriptive representation in Turkish political history. The third paper argues that larger legislatures foster greater ethnic descriptive representation regardless of regime type. Theoretically, larger legislatures provide more "room" to pay off key elites, improve the inclusion of disadvantaged groups by diminishing the value of a seat, and are less subject to stochastic features that might upset descriptive representation. The argument is tested with a series of cross-national research designs drawing on the GLP database. A new disproportionality index of Ethnic Representation is created with the aggregated data at the national level. The argument is also tested with Two-Stage Least Squares analysis where the variation in population size is taken as an instrument of legislative size. / 2031-01-01T00:00:00Z