Suicide and Self-Injury
Developing a Novel Task to Detect Suicide-Specific Interpretation Bias
Kayla Wagler, B.S.
Clinical Psychology Doctoral Student
Oklahoma State University
Stillwater, Oklahoma
Logan M. Smith, M.S.
Clinical Psychology Doctoral Student
Oklahoma State University
Stillwater, Oklahoma
Emma Unruh-Dawes, M.S.
Clinical Psychology Doctoral Student
Oklahoma State University
Stillwater, Oklahoma
Tony T. Wells, Ph.D.
Associate Professor
Oklahoma State University
Stillwater, Oklahoma
Cognitive models of suicide posit that suicide-specific cognitive processes contribute to suicidal thinking and crises (Wenzel & Beck, 2008). Empirical data has supported this, as individuals with a suicide attempt (SA) show an attentional bias toward suicide-related words (Cha et al., 2011) and demonstrate an implicit association with death (Nock et al., 2010). While previous work has shown that suicidal ideation (SI) is associated with a general, negative interpretation bias (Beard et al., 2017), no study has yet explored suicide-specific interpretation biases. The present study is an initial investigation of the relationship between SA history and SI on suicide-specific interpretation bias using a novel suicide-specific word-sentence association paradigm (WSAP).
Data were collected from two samples. The first was an online survey given to undergraduates (n = 439). The second was recruited using Amazon’s Mechanical Turk (MTurk) service (n = 283) and oversampled for individuals with SI. A suicide-specific WSAP was developed to assess suicide-related interpretation bias. In the task, a sentence is presented (e.g., “You take out a new razor blade”), followed by a word. Participants then rate how related the word is to the sentence on a 7-point scale. 19 sentences were each presented twice: once followed by a neutral word (“shaving”) and once by a suicide-related word (“wrists”). The order of sentence presentation was pseudo-random with the same sentence never presented consecutively. Total suicide bias score was calculated. Participants also reported presence or absence of current SI and history of SA and completed a self-report measure of depressive symptoms. In the undergraduate sample, mean suicide bias score was higher among those with SI (M = 48.1, SD = 13.9 ) than those without (M = 31.8, SD = 10.9); t(61.23) = -8.20, p < .001, d = 1.44; and those with SA (M = 41.6, SD = 13.5) than those without (M = 32.8, SD = 12.1); t(437) = -4.70, p < .001, d = .72. When added to a linear regression with depression symptoms predicting SI, suicide bias score explains an additional 10.0% of variance in SI beyond depression symptoms; Fchange(1, 437) = 87.74, p < .001. In the MTurk sample, again mean suicide bias score was higher among those with SI (M = 58.7, SD = 13.6) than those without (M = 39.3, SD = 15.3); t(281) = -11.36, p < .001, d = 1.34; and those with SA (M = 59.3, SD = 13.3) than those without (M = 38.9, SD = 14.9); t(281.64) = -12.22, p < .001, d = 1.44. When added to a linear regression with depression symptoms predicting SI, suicide bias score explains an additional 1.8% of variance in SI beyond depression symotoms; Fchange(1, 273) = 243.32, p < .001. Individuals with current SI or a history of SA demonstrate increased suicide-specific interpretation bias compared to individuals without SI or history of SA. Interpretation bias explains additional variance in SI above and beyond depression symptoms. This is the first to investigate a suicide-specific interpretation bias and provides initial evidence for the potential utility of such a measure. Future work is needed to determine if it provides clinical utility in identifying those at risk for future suicidal thoughts and behaviors (STB). If so, this task could hold promise as an easily administered measure of implicit risk for STB.