两阶段审议下的方向性修正：显示偏好分析
Directional Revision under Two-Horizon Deliberation: A Revealed-Preference Analysis
浏览论文内容
中文总结 AI 辅助
本文通过显示偏好分析，刻画了两阶段审议下的确定性选择，提出了可容许的排序及相关检验，适用于强调某一结果维度的决策场景。
中文摘要 AI 辅助
我们研究确定性选择，其中同一决策者在审议干预前后被观测，该干预使两个固定序数结果维度中的一个更显著。两种选择模式均为路径独立。普通选择符合二维优势，而更充分审议下的任何二元反转必须偏向在被强调维度上严格更优的选项。我们精确刻画了可容许的普通及全阶段排序。该刻画产生了受保护偏序、明确的成对及菜单层面识别、菜单层面选择变化的符号限制，以及针对不完整观测的无环性检验。在审议模式的有序序列下，来自任意固定菜单的选择构成一条单调权衡路径：每次切换都会提升被强调维度、恶化另一维度，且被放弃的选项无法重新出现。当被强调的结果序是严格的时，可容许的排序具有等价的Kemeny-Kendall表示。该框架是序数的：它既不需要基数权衡也不需要词典优先，且适用于审议对两个固定结果维度之一给予更大强调的任何情形。
英文摘要
We study deterministic choice when the same decision maker is observed before and after a deliberative intervention that makes one of two fixed ordinal consequence dimensions more salient. Both choice modes are path independent. Ordinary choice respects two-dimensional dominance, while any binary reversal under fuller deliberation must favor the alternative that is strictly better on the emphasized dimension. We characterize exactly the admissible ordinary and full-horizon rankings. The characterization yields a protected partial order, sharp pairwise and menu-level identification, a signed restriction on menu-level choice changes, and an acyclicity test for incomplete observations. With an ordered sequence of deliberative modes, choices from any fixed menu form a monotone tradeoff path: every switch improves the emphasized dimension, worsens the other dimension, and an abandoned alternative cannot reappear. When the emphasized consequence order is strict, the admissible rankings have an equivalent Kemeny-Kendall representation. The framework is ordinal: it requires neither a cardinal tradeoff nor lexicographic priority, and it applies whenever deliberation gives greater emphasis to one of two fixed consequence dimensions.
