There's a major bias in this study. Many of the people most negatively affected by combat have exactly 0 interest in discussing it, even with those close to them or other veterans. This study observed this bias but seems to have made no effort to go beyond acknowledging it:<p>---<p>Combat Study: ... non-responders had more long-term sick leave and social benefits than the survey responders (p < .001). Accordingly, there was a response bias in terms of study participants being in better health and having less need of government assistance than the non-responders. Moreover, there were significant differences in biological sex and age between responders and non-responders (p < .001) such that women and older veterans had higher response rates.<p>---<p>Noncombat Study: The responders were significantly older and had significantly lower frequencies of short- and long-term sick leaves, long-term welfare benefits, and sick leave due to mental illness (p < .001). Thus, similar to the trend in the combat-oriented sample, these results show a response bias in terms of responders being in better health and having less need of government assistance than the non-responders.<p>---<p>It dismissed the need to consider these biases with: "Although the non-responder analysis revealed some response bias, the overall response rate was high in both studies, and the effect size estimates associated with the observed response bias were small." But that seems questionable. The response rate was less than 60% in both studies, and giving an effect size estimate on an unmeasured and "significantly" demographically different population seems to be a textbook example of begging the question.