Statistical analysis of measures of arousal in ovariectomized female mice

In preparation for evaluating hormone effects and hormone receptor gene influences on fundamental processes of behavioral arousal, we subjected 48 ovariectomized female mice to a rigid protocol of several tests bearing on arousal concepts. The central hypothesis was that results would organize themselves according to capacities for sensory alertness, motor activity, and emotional reactivity. The large table of across-mouse correlation was subjected to factor analysis and cluster analysis. Results provided evidence for a general arousal (one-factor solution) which, however, accounted for only 29.72% of the variance. Four-factor (72.03%) matched four-cluster and six-factor (84.38%) matched six-cluster lineups of behavior components quite well, but did not conform to the main hypothesis; e.g., home cage exploration frequency and duration variables did not covary, and locomotor wheel activity grouped with fear indices. Besides providing a baseline for ongoing estrogen/thyroid and related genetic experiments, this statistical approach should be useful for a variety of hormonal studies of complex behaviors in mice.