Political science applications of expert-coded data (Семинар по современным вопросам политической науки)

31 октября на факультете социальных наук НИУ ВШЭ состоится очередное заседание семинара по современным вопросам политической науки.

Докладчик: Кайл Маркварт, доцент департамента политики и управления НИУ ВШЭ

Тема: Political science applications of expert-coded data

Prominent social science datasets such as the Electoral Integrity Project and the Chapel Hill Expert Survey use expert coders to gather data on concepts that are important but difficult to directly observe. However, experts are subject to a variety of well-known biases that could negatively affect the quality of data that are based on their coding. In this lecture, Kyle L. Marquardt discusses the advantages and disadvantages of using expert-coded data for political science applications. Focusing on data from the V-Dem Project — a cross-national data gathering project that includes over 150 expert-coded variables on concepts ranging from sovereignty to electoral violence — Marquardt illustrates how issues such as variation in expert scale perception and reliability can affect estimates that use expert-coded data. He then describes how a Bayesian item response theory model can account for these common sources of measurement error, as well as contexts in which this model encounters difficulties. Marquardt concludes with a discussion of future directions for modeling expert-coded data.

Рабочие языки: английский и русский.

Адрес: Кривоколенный пер., 3, ауд. 3-204.

Начало: 31 октября, 16:40.

Если вам необходим пропуск, зарегистрируйтесь, пожалуйста, по адресу: apetrushenkova@hse.ru (не позднее 30 октября, 17:00)