r J 
responses. Difference scores were also calculated for each 
criterion by subtracting a subject's mean judgement for 
things not owned from the mean judgement for things owned. 
An examination of the histograms of the 48 
distributions (12 criteria for each of the four judgement 
schemes (i.e. things owned, things not owned, the difference 
scores and the explicit judgements of general merit) 
indicated that many of them were markedly non-normal, with 
scores piling up at one end of the distribution. Further 
examination of the distributions was done using the skewness 
measure in SPSS sub-program FREQUENCIES (Nie et al., 1975), 
in which departure from zero indicates a skewed 
distribution. Absolute skewness greater than one occured in 
21 of the 48 distributions. The extreme skewness value was 
3.13. Although transformations exist for normalizing 
distributions of any shape (Langhorne & Loney, 1978), there 
would have been difficulty interpreting results with such 
transformed data and there would have been concern as to 
whether or not replications would have to use the same 
transformation. Hence, it was decided to perform subsequent 
analyses with non-parametric statistical procedures. 
For each of the four judgement conditions, the 12 
criteria of ownership were rank ordered on the basis of mean 
ranks. The criteria in each rank ordering were compared in