指數或量表？以TEDS的政治知識測量為例

Abstract

公民的政治知識被認為是探討民主政治品質的重要面向，當公民對政治事務具備基本認識，將有助於民主政治的發展。雖然已有不少學者從理論觀點探索政治知識的來源或解釋因素，在經驗層次也發展出若干測量政治知識的方式，但迄今仍缺乏強健的證據證明其測量方式的有效性。換言之，從研究方法觀點析論政治知識測量之信效度的研究仍不多見，更遑論採取複合性測量方式所設計的題目，其難易度是否足以區辨不同政治知識水準的民眾？及該題組究竟是屬於指數或量表的型態？本研究以2012年TEDS的政治知識題組為例進行效度分析，並檢證這些題目是否符合Guttman量表結構。本研究的發現如下：首先，這七個題目涵蓋了政治知識的不同面向，但有關政治人物的測量偏多。其次，影響公民之政治知識高低的個人背景變數與過去研究幾乎一致，顯示測量效度不錯。第三，題目的難易度雖符合應介於0.1和0.9之間的經驗法則，但題目的難易區隔仍有改善空間。第四，此一題組也符合Guttman量表的邏輯結構，具有累加性質。最後，從量表建構的效度與精簡性觀點，此一測量題組仍有精簡空間，未來可設計一些有關政府本質、施政作為或政治環境的題目，從中選擇難易度適中的題目納入量表組合，以提高量表的鑑別力。

Citizens` political knowledge has always been believed as an important dimension of the quality of democracy. A citizenry that possesses basic knowledge of political affairs is helpful for the development of democracy. Although many political scientists have theoretically explored the origins and determinants of political knowledge, as well as empirically developed a number of measurement techniques to gauge the level of citizens’ political knowledge and in turn analyze its relationship with other concepts, at present the evidence for the effectiveness of these techniques is still lacking. That is, from a methodological perspective, discussion about the reliability and validity of political knowledge measures is rare, let alone an analysis of whether the degree of difficulty of questionnaire items that were created by using a composite measurement method is sufficient in distinguishing citizens with different levels of political knowledge. In fact, whether in the end these composite measurement items should be considered as an index or a scale currently draws little scholarly interest. The study`s objective is to examine the validity of political knowledge items found in the 2012 TEDS, and to verify whether or not the questions, in addition of possessing the form of an index, also fit the special structure of a Guttman scale. The study has obtained the following research findings. First, TEDS has seven questions that cover different aspects of political knowledge, but there seems to be too many measures focusing on political figures. Second, level of political knowledge is found to have a statistically significant and consistent relationship with individual background variables which founded in past research, thus indicating that the measurement validity is satisfactory. Third, although the seven questions’ degrees of difficulty are within the range (between 0.1 and 0.9) set by convention, there remains room for improvement in the difficulty levels between questions. Fourth, the survey questionnaire items measuring political knowledge fit the logical structure of a Guttman scale, and are cumulative in nature. Last, from standpoint of scale construction, items measuring political knowledge can be simplified further, although future surveys may also consider designing - and incorporating - questions of appropriate difficulty level that are related to the role of government, policy accomplishments, or political environment, thereby increasing the item discrimination power of the political knowledge scale.