Item request has been placed! ×
Item request cannot be made. ×
loading  Processing Request

다차원 이상점 추론(ideal point estimation)을 통한 투표 분석 ; L1-norm Based Multidimensional Ideal Point Estimation: With Application to Roll Call Voting Data ; 유엔총회 사례를 중심으로

Item request has been placed! ×
Item request cannot be made. ×
loading   Processing Request
  • Additional Information
    • Contributors:
      박종희; Sooahn Shin; 사회과학대학 외교학과
    • Publication Information:
      서울대학교 대학원
    • Publication Date:
      2019
    • Collection:
      Seoul National University: S-Space
    • Subject Terms:
      327
    • Abstract:
      학위논문(석사)--서울대학교 대학원 :사회과학대학 외교학과,2019. 8. 박종희. ; It has been known that conventional ideal point estimation methods in social sciences (e.g., the item response theory (IRT) model or the NOMINATE-based methods) fail to recover multidimensional ideal points because of the rotational invariance of multidimensional ideal point estimates. The use of anchoring legislators in each dimension to avoid the rotation turns out to be hopeless as the number of dimension increases. Also, misplacement of any of anchoring legislators would be detrimental to inferential outcomes. On the other hand, assuming the orthogonality of each dimension along the principal axes makes it highly difficult to estimate correlated issue positions by legislators, parties, coalitions, or countries. In this paper, we present a new Bayesian method to fix the rotational invariance problem in multidimensional ideal point estimation. Our method makes two revisions: (1) we substitute the L2 norm (Euclidean distance) of existing ideal point estimation models with the L1 norm (Manhattan distance) and (2) we employ a multivariate slice sampling method to jointly sample multidimensional ideal points (Neal, 2003). Our simulation studies show that the proposed method successfully recovers multidimensional ideal points with a varying degree of correlation. We apply the proposed method to the analysis of the United Nations General Assembly (UNGA) roll-call voting data in which voting coalitions are highly flexible along issue dimensions and no clear dimensional information is known to researchers. The results of our analysis show that the post-Cold War UNGA voting has been highly multidimensional and there is strong evidence to support the U.S. as a lonely superpower. ; 투표 자료를 통한 이상점 추론 방법은 정치 행위자의 선호와 사회의 균열을 규명할 수 있다는 점에서 매우 유용한 연구 방법이다. 그러나 기존의 이상점 추론 방법들은 기술적인 한계로 인하여 다차원의 이념 공간에 대한 분석이 어려웠던 바이다. 본 연구는 선행연구들이 지니는 문제점을 분석하고 이를 대체할 새로운 방법으로 맨하탄 거리와 슬라이스 샘플링을 활용한 MSS 방법을 제안한다. 기존의 다차원 이상점 추론 방법의 경우, 정치 행위자가 이념 공간상에서 자신의 이상점과 안건에 찬성 혹은 반대했을 때의 결과에 대응하는 이상점 사이의 ...
    • ISBN:
      978-0-00-000000-2
      0-00-000000-0
    • Relation:
      000000157691; https://hdl.handle.net/10371/161441; http://dcollection.snu.ac.kr/common/orgView/000000157691; I804:11032-000000157691; 000000000040▲000000000041▲000000157691▲
    • Online Access:
      https://hdl.handle.net/10371/161441
      http://dcollection.snu.ac.kr/common/orgView/000000157691
    • Accession Number:
      edsbas.6A17477E