Development and Validation of a Brief Self-Report Measure of Agitation: The Brief Agitation Measure

Abstract

Emerging evidence indicates that agitation is an ominous precursor to imminent death by suicide, yet measures of it are few, and to our knowledge, no self-report measure of agitation exists. To fill this gap, we have developed the Brief Agitation Measure (BAM), which is designed as a brief measure to assess agitation. In this article, we provide preliminary evidence from 2 studies examining the reliability and validity of the BAM in an undergraduate sample as well as a clinical sample. We close with a discussion of the limitations of the studies and implications of our findings.

Notes

Analyses were also conducted using weighted least squares as the method of estimation, given that some of the BAM items evidenced skewed distributions. Results were similar to those found using ML. In the student sample, item loadings were as follows: BAM 1 = 0.76, BAM 2 = 0.90, AGI 3 = 0.87. In the clinical sample, loadings are BAM 1 = 0.77, BAM 2 = 0.95, BAM 3 = 0.92. All estimates remained highly significant (i.e., p < .001).

We recognize there might be some question concerning the incremental value of a CFA in a just-identified model above and beyond the information obtained from other internal consistency analyses, given the interpretation of model fit statistics in a just-identified model does not add any substantive information. Despite this, we argue that a CFA is still useful in that loading strength of each scale item can be evaluated when forced to load on the same latent factor. Accordingly, if an item does not measure the underlying latent construct in a similar manner as other items, then it would result in a diminished factor loading relative to the other scale items. Furthermore, unlike exploratory factor analysis, CFA allows for significance testing for each item loading on the latent construct, and if any item is not a strong contributor to the latent construct then it is unlikely to have a significant factor loading.