Erratum to “Relationship of perceived neighborhood danger with depression and PTSD among veterans: The moderating role of social support and neighborhood cohesion”

Abstract

Galovski, T. E., Rossi, F. S., Fox, A. B., Vogt, D., Duke, C. C., & Nillni, Y. I. (2023). Relationship of perceived neighborhood danger with depression and PTSD among veterans: The moderating role of social support and neighborhood cohesion. American Journal of Community Psychology, 71, 395–409. https://doi.org/10.1002/ajcp.12655

The authors discovered that the independent contracted survey vendor employed to field this survey had inadvertently included duplicate surveys in the deidentified data set that was provided to the authors. This error resulted in 107 duplicated cases being included in the analysis for this paper. We have identified and removed the duplicates from the data sets and the analyses have been conducted with the corrected data set. The manuscript (text, tables, and figures) has been revised to reflect the correct sample size and values accordingly and the changes in the tables and figures are also included here. The text in the Method section that describes the sampling strategy and the new sample sizes for the model is corrected as follows:

“After accounting for nondeliverable and duplicate addresses (n = 10,822, 39%), 17,178 veterans were invited to participate (67.6% high crime, 32.4% not high crime), and 3544 veterans enrolled in the study (21% response rate overall; 20% response rate high crime; 22% response rate not high crime). An analysis of the zip codes used by respondents found the FBI crime index was 99 for the not high crime group, indicating average crime levels matched the national crime average. The high crime group had a mean FBI crime index of 323, indicating crime levels 3.23 times higher than average. Participants with missing data on any of the study variables were excluded from analyses (513 participants). Thus, the total sample consisted of 3031 participants, of which 2090 (69.0%) were trauma-exposed and 1517 (51.0%) were women.”

“Additionally, all depression models included trauma exposure, and the interactions of trauma exposure with any variables included in interaction terms as predictors of mental health outcomes. Trauma exposure was not examined in the PTSD moderation models because they were conducted on only those veterans (n = 2090) who reported experiencing a DSM-5 Criterion A stressor to which they anchored their PCL. Depression moderation models included everyone who completed the PHQ-9 (n = 2942), with N = 3031 across both PTSD and depression models.”

This error did not influence the study's main findings. We did find that the three-way interaction between neighborhood cohesion, perceived neighborhood danger, and trauma exposure approached significance after the application of the Bonferroni test. This change is noted in the Results section as follows: Tables 1–4.

“RQ3a: Does neighborhood cohesion mitigate the effects of perceived neighborhood danger on Veterans' depressive symptoms? See Table 3, Model 2 for all main and interactive effects. The three-way interaction between neighborhood cohesion, perceived neighborhood danger, and trauma exposure was significant at the p < .05 level (p = .018), but approached significance after the application of the Bonferroni test (adjusted critical p < .017).” Figure 1-3

We apologize for these errors.

Thank you,

Tara Galovski

Email: [email protected]