The heat-illusion test and the structure of suggestibility

Abstract

2 similar forms of the heat-illusion test are shown to correlate to a smaller degree than would be expected if they measure the same attribute. The 2 versions also differ in the way in which they correlate with other suggestibility tests. It is shown that linear regression techniques are not appropriate for analyzing the data concerned. The interaction of various nonlinear relationships with a difference in “difficulty,” as between the 2 forms of the illusion, seems to provide an adequate explanation for the results. It is suggested that these nonlinear relationships may indicate the existence of an attribute which prevents some Ss from responding to any suggestibility test, irrespective of what the specific mechanisms of response may be.