Multidimensional scaling : evaluation of confounds in the method of complete triads

In many instances researchers are hesitant to employ "too many" objects or stimuli in a multidimensional scaling study using the method of complete triads. One of the major reasons for this is that such use is postulated to lead to experimental confounds. In particular, researchers have postulated subject fatigue/boredom effect or practice effect as a major source of subjects' unreliability when many triads are presented to them. It is interesting to note, however, that there has not as yet been any mutual agreement among researchers as to when a subject's responses should be rejected as being "too inconsistent"; nor has there been any empirical or theoretical studies illustrating the presence of postulated confounds and their effects on the solution structure resulting from a multidimensional scaling analysis. In the present project an experimental design was constructed in an attempt (a) to demonstrate subject fatigue/boredom effect or practice effect if it exists, and (b) to assess the influence of these effects on a resulting multidimensional scaling solution structure if any The results showed that: (a) the postulated confounds were not present, and (b) if the confounds did exist they had no effect on the multidimensional scaling solutions. In cases where the confounds are present, the paper suggests using balanced and partially balanced incomplete block designs as a possible alternative to complete balanced design to reduce and/or eliminate the experimental confounds.