(PS2-B41) A Psychometric Examination of Interpersonal Risk Factors for Suicidal Thoughts and Behaviors Using Confirmatory Item Factor Analysis

Leading theories of suicidal thoughts and behaviors (STBs; Joiner, 2005 & Klonsky & May, 2015) propose that interpersonal risk factors, such as perceived burdensomeness (PB) and thwarted belongingness (TB), in combination with hopelessness impact the development and maintenance of STBs. Recent research proposes the use of computerized adaptive testing to assess STB risk (King et al., 2021; Gibbons, et al., 2017); however, existing approaches fail to directly assess interpersonal risk factors or hopelessness. Since no adaptive testing item banks for interpersonal risk factors for STB exist, this study examines the plausibility of pooling existing scale items for a preliminary item bank. Hopelessness scales commonly used in suicide research were subjected to confirmatory item factor analysis to determine if they reflect a common latent construct. A similar analysis was conducted for social connectedness scales. The two factor structure of the commonly Interpersonal Needs Questionnaire (INQ; Van Orden et al., 2012) to assess TB and PB is also examined.

            Participants (N=1013, 55.6% male) were adults recruited from Amazon Mechanical Turk as part of a study examining measurement characteristics of self-report STB risk factor scales. Respondents were required to pass 80% of attention checks to be retained in final sample. 12.1% of participants reported the presence of past month suicidal ideation. The INQ was used to assess PB and TB. Hopelessness was assessed via the Beck Hopelessness Scale (Beck et al., 1974), Brief Hope scale (Everson et al., 1996), Adult Hope Scale (Snyder et al., 1991) and trait items of the State Trait Hopelessness Scale (Dunn et al., 2014). Social connectedness was assessed using the Social Connectedness Scale (Lee et al., 2001), the Multidimensional Scale of Perceived Social Support (Zimet et al., 1988), and the UCLA Loneliness Scale (Russel, et al., 1978). Individual scales demonstrated good to excellent reliability (ω = 0.80 – 0.95). Based on exploratory factor loading patterns, only the friend MSPSS items were retained.

            The two factor model for the INQ-15 failed to demonstrate adequate fit (CFI = 0.78, RMSEA = 0.116, SRMR = 0.117); modification indices suggested cross-loadings for items 9, 11, and 12. A single factor model for the hopelessness scales demonstrated adequate fit (CFI = 0.727, RMSEA = 0.066, SRMR = 0.073), suggesting the plausibility of a single latent hopelessness factor. The single factor model for the connectedness scales failed to demonstrate adequate fit (CFI = 0.622, RMSEA = 0.079, SRMR=0.105); follow-up analyses suggest the UCLA loneliness scales constitutes a separate factor relative to the social connectedness items (CFI = 0.76, RMSEA = 0.063, SRMR = 0.072). Loneliness was also distinct from TB.

            Results suggest that the hopelessness scales examined may reflect a common latent factor; however, lack of social connectedness and loneliness are psychometrically distinct constructs. These findings will inform subsequent development of item bank development for computerized adaptive testing of STB risk factors which have great potential for suicide risk assessments and intervention. Future research should seek to calibrate interpersonal risk factor items for item response models and adaptive testing.