ABSTRACT
Attitudes occupy an interdisciplinary position, and in assessing those sociologists have long utilized scaling techniques - Thurstone, Likert, and Guttman - designed primarily by psychologists. The problem is how to achieve both economy and representativeness in attitude measurement without forfeiting the requirements of reliability and homogeneity inherent in scaling. Reliability and homogeneity are essential preconditions for establishing the construct validity of a scale: without reliability there would be no correlation with any external variable; without homogeneity. 'Alpha' is the label given by L. J. Cronbach to a particular type of coefficient which measures the reliability of a test, or item battery, in the special sense of its internal consistency. The entire process of elaborating and verifying the meaning of scales, after their construction, in terms of a whole network of relationships with other variables in the survey, is akin to what Cronbach has described as 'construct validation'.
