Validation of the Suicide Resilience Inventory–25 (SRI–25) in Adolescent Psychiatric Inpatient Samples

Abstract

Resilience has been associated with a markedly decreased chance for risky behaviors following a trauma or other negative life event. This study examined the factor structure and psychometric properties of a self-report measure of resilience, the Suicide Resilience Inventory–25 (SRI–25; Osman et al., 2004), among psychiatric inpatient adolescents. In Study 1, we conducted confirmatory factor analysis to provide additional empirical support for the structure and invariance of the 3-factor model of the SRI–25 in youth samples, ages 14 to 17 years (N = 152 boys, 220 girls). Scale reliability analyses provided good evidence for internal consistency reliability of scores on the SRI–25 total and scales. In Study 2 (N = 30 boys, 40 girls), we presented data in support for the concurrent validity (i.e., known groups) of scores on the SRI–25. Additionally, we identified potential correlates for the SRI–25 total scale scores.

Notes

We did not include several measures with limited information regarding structure and psychometric properties (see Smith et al., 2000). As an example, the Brief Resilient Coping Scale (Sinclair & Wallston, 2004) includes only four items, and was validated to assess the process of resilient coping with physical pain-related stressors such as rheumatoid arthritis. Because of its length and low estimates of internal consistency for the subscales of the Resilience scale (Jew, Green, & Kroger, 1999), this measure was also not considered further. The Resilience Appraisals scale (Johnston, Gooding, Wood, & Tarrier, 2010) assesses resilience as a global (unidimensional) construct, and therefore is not comparable to the SRI–25. The Connor–Davidson Resilience (Connor & Davidson, 2003) lacks psychometric data for adolescent samples.

We thank Elizabeth O’Neil for coordinating the data collection procedures.

It is important to emphasize that the focus of this study was on the adolescent psychiatric inpatient sample. Thus, the descriptive analysis, estimates of reliability, and CFA were conducted for the full clinical sample before we undertook the invariance analyses.

As noted by Brown (2006), factor intercorrelations ≥ .85 often suggest poor discriminant validity estimates. The factor intercorrelations in this study were below the proposed cutoff value, suggesting adequate evidence for discriminant validity for the SRI–25 factors.

A major assumption in forming these groups is that youth admitted to a long-term inpatient psychiatric hospital have a history of exposure to a range of negative life events that predispose them to self-harm behaviors. However, as shown in the obtained responses, many were admitted for reasons other than suicidal behavior.