經濟投票與總統選舉：效度與內生問題的分析

透過分析1996年至2008年四次總統大選的面訪資料，本文檢視面訪中的回溯與前瞻經濟評估題組，是否適合用以推論經濟投票的存在。研究指出，回溯性與前瞻性社會經濟評估兩道題目，設定了一個詢問的時段，若在選後新政府上台後進行面訪，在概念上會產生效度的問題，特別是回溯性社會經濟評估，變成是對前後任政府施政的比較，問題較為嚴重。另外在實質的回答上，因為訪問時已經知道當選者是誰，選民的回答可能會受到政黨偏好或候選人偏好的影響，產生了內生的問題，產生因果方向的不確定性。與此相對，候選人經濟處理能力這一題，並不像回溯或前瞻社會經濟評估在題目上設定一個詢問的時段，所以不會出現回溯社會經濟評因為選後面訪，產生的效度問題。不過這個變項還是可能會內生的問題，在沒有更多資訊的情況下，文中建議至少要在模型中控制住政黨認同與候選人形象，這樣做可以控制這兩個變項對投票選擇與經濟評估的影響，以釐清經濟評估對投票的影響。

關鍵字

總統選舉；經濟投票；效度；內生經濟投票

並列摘要

We assess the retrospective and prospective economic survey questions by empirically examining the 1996 to 2008 TEDS presidential election surveys. We discuss which question is more appropriate in determining economic voting. Because the social economic assessment questions specify a time frame, they generate the validity problem when asked after the elected president has taken office. This problem is particularly severe for retrospective economic assessment. It induces respondents to compare the economic conditions of two consecutive presidential terms rather than the economic conditions one year before the election. In addition, because respondents already know those who have been elected, their answers are more likely to be influenced by partisan or candidate preferences, resulting in the problem of endogeneity. By contrast, the questions regarding the predecessor's and the candidates' handling of the economy do not specify a time frame. The validity problems are less severe. In actual fact, we demonstrate that the endogeneity problem goes with each set of economic assessment questions. To better examine the effect of economic assessment on vote choice, it is necessary to address the endogeneity problem. Without further information, researchers should at least control party identification and candidate image.