equal appearing intervals

equal appearing intervals This is often referred to, after its principal author L. L. Thurstone, as the Thurstone technique of attitude measurement. On the assumption that any attitude involves a continuum ranging from the strongest possible appreciation of a value at one extreme to the strongest possible depreciation of that value at the other, Thurstone devised a scaling procedure which attempted to divide any given attitude into a number of equal appearing intervals, and so give respondents a scale-value according to their responses to the statements in a particular attitude or value-set.

Scaling attitudes in this way involves a number of steps. In the first, a group of judges sort a number of attitude statements into categories, according to the relative positions of these statements along an attitude continuum which has already been divided into categories having positions equidistant from those immediately above and below them. The allocation of statements to categories by the judges is then used to construct a scale, securing an individual scale-value for each item, assigning it the median position given by the judges as a whole. If a large number of statements has been used at the outset, then it is necessary to select from these in such a way as to exclude all ambiguous items, and to represent the full range of the attitude in question. Finally, respondents are invited to accept or reject the items in the scale, each respondent being given a score equal to the average of all the statements he or she accepts. The technique has also been used to construct cognitive and behavioural scales.

There are numerous well-documented dangers to be guarded against in using the technique. The criteria for eliminating the irrelevant items are various and contentious. The scale is intended to provide interval measurement of the attitudes of the respondents, but this presupposes that the items in the scale are measured at the interval level (‘equal appearing’), a questionable assumption given the way in which the final scale-value of each item is derived. Because the scale-values depend on the judges, it is important that as large a number as possible of well-informed judges are used, and that they are not systematically biased with regard to the attitudes being measured.