Paired comparison experiments involving all possible pairs

Abstract

The analysis of paired comparison experiments using the tie and order model due to Davidson & Beaver is extended to account for the inclusion of pairs of identical items in the experiment (design 1). It is shown that in testing for order effects, the asymptotic relative efficiency of design 1 to design 2, which includes the distinct ordered pairs, is always greater than one in non‐trivial cases.

Replicated paired comparisons at the individual level

“…This latter approach was extended by Davidson (1970) and more recently by Bradley (1984), Colonius (1980). Koehler & Ridpath (1982), Sirotnik & Beaver (1984) and Bockenholt (1988). Other aggregate, o r population level, contributions have been given by Cooper and Nakanishi (1983), Davoodzadeh & Beaver (1983) and Takane (1989).…”