The messy, passionate, ideological world of politics has always been in conflict with the sober reasonableness of good governance. But it is hard to remember a time where evidence-based decision making was in shorter supply than right now. Last issue, we published the proceedings of a symposium on how advocacy by trans service members and scientists led to the lifting of the previous ban on transgender people serving in the US military. At the time of that symposium, in May 2017, it was unclear whether the election of Donald Trump would have an impact on that decision, but today, the US president’s willingness to sacrifice reasonable policy for political gain can no longer be in doubt. For those interested in the arguments and supporting evidence that helped overturn that ban, so that this current ban can be re-fought, please take a look (Ahuja, Ortega, Belkin, & Neira, Citation2019).
In this issue, we kick off a new article series, Perspectives, with an article by Dr. Joseph Wise that argues against this ban from his perspective as a former military psychiatrist. The Perspectives series will consist of pieces that come at issues of LGBTQ mental health from a personal perspective. Please contact us at [email protected] if interested in submitting to this series.
We have a lot of original research to dispense to you in this issue; in fact, no fewer than seven studies on issues ranging from teachers’ views on gender non-conforming children, to the intersectionality of LGBTQ mental health with the experiences of ethnic and racialized groups, as well as multiple articles on anxiety and depression in the LGBTQ community. One discussion we have had within the editorial team has been whether our readership would benefit from some continuing education pieces on statistics and methodology to better be able to interpret what they see as the Journal publishes more and more original research. Associate Editor Dr. Ronald Hellman put together the following primer on the statistical methods used in “Stigma and Suicide Risk among the LGBTQ Population: Are Anxiety and Depression to Blame and Can Connectedness to the LGBTQ Community Help?”, an article from this issue by Kaniuka et al.:
Kaniuka et al. report on research utilizing a mediator/moderator model that explores how the minority stress theory of Meyer may operate. For readers unfamiliar with this methodology, an example should help.
Say you want to explore the effect of a drug in promoting better sleep. The drug is the exposure/independent variable and quality of sleep is the outcome/dependent variable. The drug might have a direct sedative effect on sleep, or an indirect causal effect, perhaps through an analgesic property. Its analgesic effect is indirect because it targets pain, not sleep, but is a mediator in the pathway that improves sleep through pain relief. Mediators are important because they are potential targets for intervention. Would we feel more comfortable giving a patient an aspirin at bedtime or a sleeping pill?
Moderators are other factors that may be associated with a differential outcome regarding sleep quality in the example, but are not causal per se, such as gender or age. Maybe males and females rate the quality of their sleep differently, but that is not because gender is having some causal effect on sleep itself.
For Kaiuka et al., the exposure is “stigma” and the outcome “suicidal behavior.” They explore whether anxiety and depression are mediators in the pathway to suicidal behavior, where stigma would have an indirect effect on suicidal behavior mediated by anxiety or depression, thus the title of the article. They theorize that a connection with the LGBT community may function to moderate the impact of anxiety and depression, but is not causal in suicidal behavior. Such models rely on certain assumptions that derive from the theory, and are subject to unidentified variables that may confound observations, among other concerns. See what you think.
Kaniuka et al. also use a statistical tool called “bootstrapping” to derive more accurate estimates of confidence intervals for indirect mediator and moderator effects. That is because the equations used to determine these effects assume a normal distribution that does not often occur with scant or skewed data. We would like to say that we are 95% confident that the mean of our data falls within a defined range that is set by the confidence limits, and we accept that there is a 5% chance of error that it could fall outside that range. But with skewed data, the true value falls to the left or right of the confidence limits more often than predicted, resulting in less power to detect a true indirect effect.
Bootstrapping gives a more accurate estimate of the confidence intervals by treating the study sample of the population as a proxy population, and then randomly resampling it many times. Averaging the data in each subset produces means, and a lot of them when plotted form a normal distribution. This phenomenon is proven and ensconced in the Central Limit Theorum. When the mean and confidence limits derived from bootstrapping are applied to indirect mediator and moderator effects, this has been proven to enhance accuracy and confidence in the results.
Please let us know if you would like to see more primers on statistics and methodology like this in future issues of JGLMH.
Lastly, in this issue, we are also fortunate to have another contribution to the JGLMH Oral History series with Emeritus Editor Dr. Jack Drescher interviewing Dr. John H. K. Sweet.