Abstract
The U.S. Army developed the Global Assessment Tool (GAT) to monitor psychosocial fitness and well-being among soldiers and provide a means to objectively gauge the success of newly implemented resilience training programs. Despite its widespread use (taken over 5.2 million times) and stated utility for program evaluation, there is relatively little published evidence regarding the GAT’s reliability and validity. We used exploratory structural equation modeling (ESEM) with 4 random samples of soldiers (n = 10,000 each) to examine the factorial validity and reliability of the GAT. An 11-factor solution (Self-Management, Positive Affect, Meaning, Work Engagement, Organizational Trust, Loneliness, Negative Cognitions, Hostility, Negative Emotions, Depressive Symptoms, and Emotion-Focused Coping), with 4 additional factors assessing character strengths (Intellect, Warmth, Civic Strengths, and Temperance), fit well and replicated in a second random sample. A higher order, 2-factor model using composites scores and positing positive and negative psychosocial competencies also fit well. Tests of measurement invariance using a third random sample reinforced consistent measurement properties across gender, age, and rank, with the exception of character strengths, which produced different factor structures for males and females. Further validity tests using a fourth random sample underscored a modicum of divergence among the resilience factors and convergence among the character strengths factors. We conclude with recommendations for enhancing and refining the GAT and discuss the GAT’s utility as a reliable, multidimensional psychosocial assessment that can be used to evaluate the efficacy of military resilience training programs.
Notes
1 The terms psychosocial fitness and resilience have been used somewhat interchangeably in the GAT literature to reference the compilation of strengths and assets (i.e., skills, cognitions, affect regulation, and indices of psychological health) assessed on the GAT and targeted through soldier trainings.
2 Three GAT scales were excluded from the present study. Family satisfaction and family support, which were missing data for soldiers who reported no family or romantic relationships, were excluded in an attempt to avoid biasing our sample. Additionally, we chose not to include a measure of friendship, which consists of five dichotomous (yes–no) questions and one ordered categorical question, because of the different response formats.
3 Both Federal and Department of Defense guidelines mandate certain human subjects’ protections, including adherence to strict protocols for deidentification in the presence of personally identifying and protected health information. For more information regarding the procedures implemented to remove subject identifiers in the PDE, see CitationVie, Griffith, Scheier, Lester, and Seligman (2013) and CitationVie and colleagues (2015).