A Longitudinal Examination of Secondary Traumatic Stress among Law Enforcement

Abstract

Current research on secondary traumatic stress (STS) has been drawn from cross-sectional data. To determine how STS manifests over time, we conducted a three-year, longitudinal panel survey of investigators at a federal law enforcement agency. We measured STS scores, coping styles, perceptions of the work environment, and subject demographics. We found that STS scores are fairly stable over time. A variety of both positive and negative coping mechanisms, as well as characteristics about the work environment, were found to impact STS. In particular, coping with denial more often in the previous year was related to higher STS scores a year later, while higher scores indicating supervisory support were related to lower STS one year later. Deputy demographic variables were unrelated to STS.

Keywords:

• secondary traumatic stress
• panel data
• coping
• law enforcement

Notes

1. There is some possibility of endogeneity here. However, it is important to note that the question merely asks “how psychologically or emotionally difficult” the material was to view. It is possible that deputies may rank material as highly psychologically or emotionally difficult but have no signs of STS. STS is a set of symptoms stemming from knowledge of another’s trauma. We cannot assume it is the same as a person’s self-report of the difficulty of interacting with disturbing media, and therefore we have included this as an independent variable.

2. There were some slight differences in the demographics between those that dropped out and those respondents that did not. Those who only took the survey at wave 2 were less likely to have kids and more likely to be female than those who did not drop out (p < .01). Those who dropped out after wave 2 were less likely to be SOICs (p < .01). Those who dropped out after wave 2 also reported significantly less difficulty with disturbing media at wave 2 than the rest of the sample (p < .01).

3. As the dependent variable was slightly skewed, we tested a variety of transformations. The square root transformation was the best fit the dependent variable. We ran the final model against the square root of the dependent variable—all of the independent variables maintained their direction, magnitude, and significance. Therefore, for ease of interpretation, we present the dependent variable in its original, untransformed state.