Controlling the effect of stimulus context change on attitude statements using Michell's binary tree procedure

Context effects abound in research on attitudes, making tight stimulus cont rol notoriously difficult. In particular, changing the stimuli in which a s et of statements are embedded appears to affect how these statements are ju dged (Sherman, 1994; Sutcliffe & Bristow, 1966). This challenges an assumpt ion of Thurstone scaling: that constructed scales remain constant for subse quent judgements. However, the binary tree procedure developed by Michell ( 1994, 1998) enables highly structured sets of stimuli to be constructed tha t should be immune to stimulus context effects. To test this hypothesis, tw o sets of attitude statements were used: one set constructed using a binary tree procedure and the other a slightly modified version of Sherman's (199 2) Set 1 statements, which were not constructed using a binary tree. The st atements constructed using a binary tree procedure did not show any context effects, while the Sherman statements were ordered differently depending o n the context.