A PAIRED-COMPARISON METHOD FOR INTERVAL SCALING

A new method of analyzing paired-comparison data is presented: the lin ear paired-comparison (LPC) method. The LPC solves some deficiencies i n the implementation of the eigenvector and geometric mean ratio hiera rchy paired-comparison methods proposed by T. L. Saaty. In addition to formats typically used to co paired-comparison data, an alternative f ormat for collecting data is presented Levels of significance with reg ard to rater consistency were evaluated for bo geometric mean and LPC methods, and it was found that levels of significance were not an adeq uate predictor of the proper ordinal ranking of the scale values eithe r method. A constant value for a consistency criterion is more appropr iate. Saaty's criterion of an acceptable consistency index (CI) to jud ge a rater's consistency of CI less than or equal to 0.10 is defended. In addition, an appropriate value for the geometric mean method's rat er consistency criterion, the residual mean square (S-2), was found to be S-2 less than or equal to 0.27. For the LPC method, an acceptable criterion for the standard deviation from consistency (sigma(c)) was f ound to be sigma(c) less than or equal to 2.25.