ERIC Number: ED399296
Record Type: Non-Journal
Publication Date: 1993-Apr
Reference Count: N/A
Paired Comparison Intransitivity: Useful Information or Nuisance?
Johanson, George A.; Gips, Crystal J.
The decision to use a forced-choice test item format versus an item format where choice is not forced (e.g., a Likert scale) might best be determined by the nature of the information sought since the difficult decisions required for forced-choice formats may result in a different scaling than an unforced method. If a forced choice is desired, there remains the question of whether to have subjects rank order their preferences or to present the items as paired comparisons. One of the main differences between ranking and paired comparisons is the potential in the latter format for what is known as "circular triads" or "intransivities." A subject may choose object A over B, then B over C, and then, intransitively, C over A. Are these intransivities an unnecessary complication or a useful source of information? An example involving the hiring preferences of principals using a national sample of 203 principals with 36 paired comparisons illustrates, through computation of the total number of circular triads, that knowledge of intransitivities adds to knowledge about the real choices of respondents. By excluding the possibility of intransitivities, a ranking format in the hiring preferences example would have been likely to yield similar scale values but much less insight into the respondents' processes. (Contains 13 references.) (SLD)
Publication Type: Reports - Research; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A