Online Program Home
My Program

Abstract Details

Activity Number: 655 - Applications in the Analysis of Survey Data
Type: Contributed
Date/Time: Thursday, August 1, 2019 : 10:30 AM to 12:20 PM
Sponsor: Government Statistics Section
Abstract #305289 Presentation
Title: Using Vector Generalized Linear Models to Assess Kurdish Democratic Progress Under Erdogan
Author(s): Ole Forsberg*
Companies: Knox College
Keywords: Differential Invalidation; Electoral Forensics; Vector GLM; Democracy; Beta-Binomial Distribution

Recep Tayyip Erdogan has led Turkey since first being elected prime minister in 2003 (and president in 2014). During this period, Erdogan has passed several democratic reforms with a stated goal of bringing democracy to all in Turkey, including the Kurds, who have been systematically discriminated against since the founding of the Republic.

This article examines the effects of Erdogan's reforms on the Kurds in terms of differential invalidation (DI) in the constitutional referenda of 2007, 2010, and 2017. DI is the act of vote-counters having a higher probability of invalidating a ballot based on demographic or electoral reasons. Existence of DI suggest ballots are invalidated due to unethical reasons like whom the ballot is cast for, ethnicity of the voter, etc.

DI is tested using regression because the variables are numeric. The dependent variable is the proportion of votes invalidated and the independent variables are the proportions of votes in favor of the government-backed referendum and of Kurds in the province. Because the dependent variable is an overdispersed count variable, a VGLM in the form of a beta-binomial distribution is used to account for this overdispersion.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program