學刊論文
Applications of Stochastic Media Theory to 1992, 1996, and 2000 National Election Study Panel Data

中華心理學刊
民96 , 49 卷, 3 期, 225-244


Yung-Fong Hsu(Department of Psychology, National Taiwan University, Taiwan);Michel Regenwetter(Quantitative Division, Department of Psychology; Department of Political Science,University of Illinois at Urbana-Champaign, USA)

 

Abstract

We study a class of stochastic models of persuasion that form an application of media theory developed by Falmagne and others. These models describe the evolution of preferences over time. We consider the case where
personal preferences are represented by (strict) weak orders and semiorders. Over time, these preferences may change under the influence of ‘‘tokens’’ of information arising in the environment. Successful applications of
some weak order implementations of stochastic media theory to 1992 U.S. National Election Study (NES) panel data have been reported by Falmagne and various collaborators. However, past attempts to fit a semiorder model to the same data have failed. We successfully fit four media theoretic models, including two semiorder models based on the ‘‘neighboring’’ response mechanism, to 1992, 1996, and 2000 NES panel data. We compare the fit of these four models, discuss the psychological interpretation of key model parameters and illustrate applications to negative political campaigning. 

 

Keywords: Panel Data, Semiorder, Stochastic Media Theory, Thermometer Scores, Weak Order

登入
會員登入
更新驗證碼