Methodological and conceptual issues in the use of the elevated plus-maze as a psychological measurement instrument of animal anxiety-like behavior.

There has been some suggestion that 'risk assessment' defensive behaviors in rodents might resemble some of the behavioral/somatic symptoms of generalized anxiety in humans. Although the inclusion of some risk assessment behaviors enhanced the sensitivity of the elevated plus-maze to detect subtle changes in anxiety-like behavior, there is little evidence to support the inclusion of 15 or 20 indicator variables in an analysis. Several methodological, conceptual, complexity and interpretation problems associated with the factorial validity of recently published ethologically-derived large-scale principal components analyses of elevated plus-maze behavior are examined in this review. The utility of confirmatory factor analytic work currently being conducted in our laboratory to test structural hypotheses of anxiety-relevant elevated plus-maze behavior is then discussed with a view to address some of these issues. Finally, we propose that the growing number of measured behavioral indices in the elevated plus-maze test battery be reduced, and suggest that some of the underlying constructs thought to drive behavior in the apparatus are in need of re-evaluation.