憑據猜答或盲目猜答？以政治知識測量為例

Informed Guessing or Blind Guessing? The Case of the Measurement of Political Knowledge

本文主要在探究受訪者回答選擇題型的政治知識題目時的猜答行為，以東吳大學學生為研究對象，其資料性質屬於調查實驗資料。由於選擇題題目提供選項讓受訪者選擇，所以一般會認為不知道答案的受訪者能夠藉由盲目猜答的方式選到正確答案，因而造成政治知識程度被高估的問題。然而，本文從理論上分析受訪者的政治知識程度與其在開放題和選擇題的回應模式，認為當受訪者不確定答案時，會根據其擁有的部份知識或錯誤知識進行憑據猜答。分析結果顯示受訪者在選擇題進行猜答時，有很高的機率會選擇正確答案，而非等機率的選擇選項，意謂受訪者是憑據猜答而不是盲目猜答。此結果另一個意涵為選擇題比開放題有較高的答對率，是來自於內在知識的反映而非隨機猜答。 This article examines respondents' guessing behavior in multiple-choice items for the measurement of political knowledge. Some researchers suspect that using multiple-choice items to measure political knowledge would overestimate observed knowledge because respondents could correctly answer some questions just by blind guessing. We argue, however, that informed guessing is more common than blind guessing. This is because successful guessing requires certain levels of knowledge, which implies that respondents' observed knowledge is not seriously higher than their actual knowledge. Analyzing survey experimental data, we show that, when the respondents answer political knowledge questions by guessing, they are more likely to select the correct answer. This result implies that respondents tend to guess the answer based on their partial knowledge rather than by chance.