本文以實驗方法，探究資訊對Palfrey-Rosenthal關鍵投票者模型（簡稱PR模型）中的投票行爲的影響。作者比較完全資訊情境和部分資訊情境:前者爲受試者知道每一政黨的基本盤大小，後者爲受試者知道某一政黨基本盤大小，而另一政黨基本盤大小，受試者僅知其機率。兩發現如下：第一，部分資訊情境下，當知悉某政黨爲小基本盤時，受試者相信自己是關鍵投票者的機率會高於理論值，使其投票率增加，不低於其在完全資訊情境的投票率。第二，完全資訊情境下，當受試者支持的政黨爲大基本盤時，受試者投票率會增加，顯著高於其相信自己爲關鍵投票者的機率下的最適反應;PR模型結合厭惡失望效果可解釋此行爲。

This paper experimentally examines the effects of information on voter turnout in the Palfrey-Rosenthal pivotal voter (PR) model. In particular, I compare two different degrees of information revelation: one is that voters know each party’s support base (complete information scenario), and the other is that voters know one party’s support base but are limited to knowing the probability of the support base of the other party (partial information scenario). There are two main findings. First, in the partial information scenario with a revelation of a weak support base, subjects tend to have a higher belief in being pivotal than theory predicts, which causes their turnout rate to be not lower than those in the corresponding complete information scenario. Second, in the complete information scenario, turnout of the subjects of the frontrunner party is significantly higher than the subjects’ best response to their pivotal belief, which can be explained by a PR model incorporating the generalized disappointment aversion effect.