ABSTRACT
Death by suicide is the leading cause of death in American jails. Closely linked with suicides are self-injurious thoughts and behaviors. Even with this knowledge, the risk factors for self-harm ideation and self-harm behaviors have received little empirical work within a jail population. This study couples an integrated importation and deprivation model with an ideation-to-action approach to examine environmental and psychological correlates of self-harm ideation and self-harm behaviors among a sample of 231 individuals incarcerated in jail. Results from the ordinary least squares regression models demonstrate that both deprivation (institutional misconduct) and importation (hopelessness) factors are important when identifying self-harm ideation and self-harm behaviors. In addition, the results of this test of the integrated model offer a proposed new direction for integrated theory testing. Finally, a discussion of policy implications and actionable programs are suggested.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. To make a proper adjustment for adaptations since Sykes defined this in 1958, the current study uses “the deprivation of relationships,” removing the sexuality defined constraint.
2. Information was not collected, or accessible, on how many individuals were in the two ineligible units on the day of collection. These units were excluded due to regulations set by the facility. A request was made to survey the two units, but was not allowed.
3. One of the main limitations of this study comes from a missing data problem. In the final analytic sample, it is apparent that the sample size drops from 353 to 231. Multiple imputation was not used after running a Little’s MCAR (missing completely at random) test (Jakobsen et al., Citation2017; Nussbaum, Citation2015). The data was found to not be MCAR, and thus multiple imputation is not the appropriate method to analyze the data (Nussbaum, Citation2015).
4. Per a suggestion from a reviewer, models were also run after imputation, the results of all models are substantively the same, with the same variables and direction replicated. Therefore, the original models are presented throughout that do not utilize multiple imputation, due to the results of the Little’s test.
5. Only the research team handed out and collected surveys, to reduce any potential coercion that could have arisen if the jail staff had assisted with administration.
6. It is important to note that this self-harm ideation is based on a measure that asks about ideation for planning purposes, not just ideation of self-harm generally. While a limitation of this study, and something future work should address in a more nuanced manner, this item is the best proxy for self-harm ideation in jail from this data and thus is used with caution. A full discussion of this caveat is had in the limitation section.
7. Single-item reliability can be calculated using the correction for attenuation formula (Nunnally & Bernstein, Citation1994). While this should be interpreted with caution, there is support for the reliability of a single measure to be assess in this manner in an exploratory manner (see Abderhalden & Moule, under review, Ginns & Barrie, Citation2004; Nunnally, Citation1978; Wanous & Reichers, Citation1996).
8. Race was dichotomized due to the final sample size of 231.
9. This was dichotomized due to very low response for Transgendered and Other, which did not hold enough power to be run as separate categories. Results are reported as “male” and “not male.”
10. Models were also run with bootstrapped standard errors to see if there were any changes to the estimates, given the small sample and the quantity of covariates. In running the bootstrapped models, there were no substantive changes to any of the findings. However, these models are available upon request.