信度、Alpha係數與相關議題之探究

A Research for Reliability, Coefficient Alpha and Related Issues

在心理與社會科學研究中，呈現α係數作為測量工具之內部一致性指數似乎是件例行工作，然而使用者對內部一致性與同質性或單一向度須加以區分。內部一致性指量表中之試題間相關的程度，而同質性和單一向度則指試題中包含一個潛在因素。一般而言，內部一致性是同質性或單一向度的必要而非充分條件，除非量表內之所有試題間均有關連，否則該量表不具同質性，但是，量表內可包括許多彼此相關之題目卻不是單一向度。α係數不是內部一致性良好的指數，因為它受到測驗題數與試題間相關係數平均數的影響，研究者提出建議，當試題超過40題時，α係數不具功能，試題間相關係數之平均數的功能反而優於α係數，雖然試題間相關係數之平均數低和具多向度，該量表可能仍有個高α值。

Presentation of coefficient alpha as an index of the internal consistency or reliability of psychological measures has become routine practice in all psychological and social science research. It is necessary to distinguish between internal consistency and homogeneity or unidimensionality. Internal consistency refers to the overall degree to which the items that make up a scale are intercorrelated， whereas homogeneity and unidimensionality indicate whether the scale items contain a single latent factor. Generally speaking， internal consistency is a necessary but not sufficient condition for homogeneity or unidimensionality. A scale cannot be homogenerous unless all of its items are interrelated， a scale can contain many interrelated items and still not be unidimensional. Coefficient alpha is ambiguous and imperfect indicator of internal consistency because it essentially is a function of the number of test items and the average intercorrelation among the items. Researcher suggested that coefficient alpha is useless as an index of internal consistency for scales containing 40 or more items. The average interiterm correlation is a much more useful index than coefficient alpha. Alpha can be high in spite of low item intercorrelations and multidimensionality.