考量規範性社會影響之意見動態模擬

Abstract

集體意見之動態是社會科學中一項重要的議題。二十世紀晚期，以多代理人為基礎之模擬逐漸被使用在研究意見動態上，企圖找出像投票、革命、流行風潮等社會現象背後淺在的法則。 本研究基於社會心理學的實驗以及理論來擴充過去的意見動態模型，在此新模型中的代理人和他人進行意見交換時會視情況考慮週遭的輿論，也就是所謂的「規範性社會影響」，而不像過去那樣不顧他人的看法任意表態。為了區分社會心理學中所謂的「表面順從」和「私下接納」這兩種效應，本研究將過去只有「意見」單一層的代理人擴充成含有「意見」和「態度」兩個層次，以期能更廣泛且真實地模擬社會上的意見互動情形。 經由模擬實驗，發現除了模擬基本的意見交換過程之外，我們的確能夠探討更複雜的議題，像是人眾無知、革命、少數者影響、新進者影響、思想控制、媒體等。相較於過去的模型，本研究模型率先嘗試以社會心理學之理論和實驗數據為依據來制定模型，以提高模型的參考價值，為社會科學領域的研究者提供一個觀察和實驗平台。

Dynamics of collective opinion is an important issue in social science. In the late 20th century, agent-based simulation was gradually adopted in researching opinion dynamics. By simulation, researchers attempted to find the rules behind some social phenomenon such as voting, revolution, fashion etc. In this thesis, I expand previous opinion dynamics models base on experiments and theories of social psychology. In my model, when agents exchange their opinions, they consider not only his own opinion but also his neighbor's reception, which is the so-called "normative social influence". In the previous models, agent is described by a single variable "opinion". In order to separate the effect of "public compliance" from the effect of "private acceptance", I propose the second variable attitude. In addition to basic opinion exchanging process, my model can simulate more complex social phenomenon such as pluralistic ignorance, revolution, minority influence, newcomer influence, mind control, media etc. Compared to previous models, this is the first thesis that based on experiments and theories to increase factuality and generality, which provide the social scientists a platform to observe and experiment.