One research approach to ascertaining the validity of predictor variables is to have subjects rate their environmental preferences and the related predictor variables. Upon using this type of approach in rating experiments, the authors encountered an inconsistency: the inability to ascertain the predictor variables' relevance to evaluation, even though those predictor variables were in fact the parameters cited by the subjects as justification for their evaluations. Data from rating experiments were used to verify the authors' hypothesis that this inconsistency arises from the difference in direction (i.e., positive versus negative) in the parameters' effects on evaluation. It was also demonstrated that changes in the direction of effect on evaluation arise from differences in a parameter name's connotation. The fact that some parameter names have different connotations in different circumstances while others have a consistent connotation is said to indicate the existence of a hierarchy in the connotations of the words used to represent parameters. Finally, the need for hierarchical analysis is stated.