“…It is well established in psychometrics that these are ordinal, not interval, measurement scales (Coombs, 1964(Coombs, , 1976Siegel, 1956;Stevens, 1946Stevens, , 1951. However, with some exceptions (Benjamin, et al, 1998 andGolob andHensher, 1998), attitudinal variables are generally assumed to be interval-scaled continuous variables in attitude-behavior models (e.g., as in Bernardino and Ben-Akiva (1996) and Lyon, 1981a,b), and this must lead to biases in hypothesis testing. The method used in the present research attempts to avoid these biases.…”