ABSTRACT
It has long been recognized that both the Thurstone and Likert methods of scoring attitude scales suffer from certain weaknesses, and several attempts have been made to develop a new method which would combine the advantages and avoid the defects of both. The Thurstone method accurately weights each item, although it does not allow for shades of agreement or disagreement; the Likert method permits a number of possible responses to each item, ranging, for example, from "strongly agree" to "strongly disagree". On balance, Likert's technique appears slightly preferable, both because scores are easier to compute and because it has been shown by Likert, Roslow & Murphy to produce rather higher coefficients of reliability. P. F. C. Castle's scale products are obtained by first computing the deviation of each item's scale value from the neutral point of the scale, and then multiplying this deviation by the Likert weights.
