Thurstone Scales

The Method of Equal-Appearing Intervals

In creating a scale using the method of equalappearing intervals, the first step is in determining the attitude object that will be evaluated. The next step is to construct a set of statements about the attitude of object that captures an entire range of opinions, from extremely unfavorable (e.g., “Abortion weakens the moral fiber of our society”) to neutral (e.g., “Abortion brings happiness to some, unhappiness to others”) to extremely favorable (e.g., “Abortion should be encouraged for unwanted pregnancies”). Thurstone used 130 statements to develop one of his original scales, although subsequent researchers have suggested that 40 to 50 statements are typically enough to fully capture the continuum of a particular attitude object.

Once the researcher develops a complete set of statements, judges rate favorability values for each statement using an 11-point scale, which may take various forms (see Figure 1). Typically, only the two extreme ends and the midpoint of the scale are labeled. These three forms are alternatives to Thurstone's original procedure of sorting the statements, presented one at a time on cards, into 11 physical piles. Thurstone used 300 judges to obtain favorability values, but other researchers have successfully constructed scales using as few as 10 to 15 judges.

When making these ratings for the pool of statements, judges are instructed to respond objectively on the favorableness or unfavorableness of the statements, not based on their personal agreement or disagreement with the statement. Judges are also asked to treat each interval as being equal. To ensure that judges' ratings are not careless and conform to the instructions, Thurstone recommended a screening procedure to eliminate the data set of judges who placed 30 or more statements (of the 130 total) into just 1 category out of the 11. This is equivalent to placing 23% or more statements into a single category, which may be problematic considering the care taken to develop the pool of statements to capture the entire range of opinions from favorable to unfavorable. After this screening procedure, a measure of central tendency for each statement and a measure of variability are calculated across the remaining judges. Thurstone used the median and interquartile range, but in recent years, means and standard deviations [Page 1003]have more commonly been used. Low variability indicates relative consensus among the judges, whereas high variability indicates little agreement between the judges, signifying that the particular statement may be ambiguous.

...