We reviewed Mrdjenovich’s (2018) article and found issues with both the qualitative study component, and in the critical review of the atheism–health literature. Our major criticisms of Mrdjenovich’s work are that: (a) the qualitative piece was questionably executed, (b) there was an ostensible misunderstanding of regression, (c) there are confusions about the mechanisms behind the religion/spirituality (R/S) and health relationship, and (d) the description of atheism–health research as “adversarial” is unjustified. The present commentary draws attention to these misunderstandings and suggests several ways forward.

Mrdjenovich’s article used a literature synthesis approach to identify print media that investigated why atheists engaged in R/S activities. However, only four key terms related to atheism (and a handful of related phrases) were used to find articles (omissions include “secular,” “humanist,” “Sunday Assembly”). The consequences of this search promptly manifest, as Mrdjenovich only identifies 22 media articles, of which a mere nine were written by atheists or were from an atheist POV. Mrdjenovich claims data saturation was achieved, which is a believable, albeit unremarkable claim. Mrdjenovich notes that collecting data from atheists is difficult, but then in a stroke of irony, dismisses the utility of survey data despite it showing promise in addressing that very issue (e.g., Farias et al. 2018; Preston and Shin 2017; Price and Launay 2018; Silver et al. 2014). Rather than completing the admittedly more difficult task of collecting primary data or using researchers’ existing data on the subject, Mrdjenovich is attempting to answer a large and important question by fishing in a puddle instead of a sea.

Later in the article, Mrdjenovich is critical of research examining R/S identity as influencing the relationship between R/S and health. Mrdjenovich specifically references Speed and Fowler (2015) and Speed (2017) who used moderation analyses to determine that atheists reported a negative relationship between religiosity and health, while theists reported a neutral relationship between religiosity and health. Mrdjenovich states that this research was not novel, and the findings were obvious. However, Mrdjenovich proceeds to state that what research really needed was to disaggregate the R/S–health relationship across groups, which is the very thing Speed and Fowler had done in their work! We suspect that Mrdjenovich does not completely grasp the implications of moderation analyses within regression, which is additionally supported by his further claim that secular minorities are adequately included within research. Mrdjenovich correctly notes that secular minorities are sampled within the occasional study, and concludes from this that the R/S-health research field is not monolithic. However, without moderation terms to explore the unique relationships that secular minorities report between R/S and health, these minorities’ data are being “overpowered” by the effect of the larger group. Moreover, while atheists are fewer in numbers than religious majorities, they are extremely diverse in attitudes, opinions, and behaviors, suggesting a concerted effort (as opposed to an incidental one) would be required to investigate the group (Silver et al. 2014).

At a later point, Mrdjenovich recognizes that belief in God has not itself been found to improve health outcomes, but then seemingly concludes that belief content must still play a role in health outcomes regardless (“recognizing that belief in God is something so fundamental and paramount to religious traditions”). Besides sweeping aside non-theistic religions, Mrdjenovich offers an assertive argument in lieu of an empirical one. He then turns the scientific process on its head by stating that, “Rather than implying that belief is not healthy for believers, the implication is simply that non-belief is not unhealthy for non-believers.” As it stands currently though, theistic belief has no health implications be they positive or negative. By using a parallel example of, “Rather than implying that homeopathy is not health-promoting, we could instead imply that homeopathy is not illness-causing,” the speciousness of Mrdjenovich’s reasoning becomes apparent.

Mrdjenovich refers to several studies to bolster the claim that religiously based social support and coping methods explain variance in health outcomes beyond variance from secular sources. However, the research cited (e.g., Pargament 1997) does not indicate that religious sources are superior to secular sources for non-believers. Indeed, as pointed out by Galen (2018), work such as Pargament’s (1997) that focused on religious coping typically included samples with overwhelming proportions of religious individuals. Although these studies may be equipped to assess the efficacy of religious methods among the religious, they are not able to compare the relative superiority of religious versus secular appraisals in general.

Astonishingly, Mrdjenovich’s view on future research does not include attempts to determine if mechanisms of health necessarily operate in an R/S context or would be equally applicable in a secular context. Fortunately, recent studies have begun to address this very question: Price and Launay (2018) have found that participation in “Sunday Assembly” positively influenced attendees’ wellbeing. Their research supports the notion that group participation need not be religious to be salutary, but also highlights the question of what, if any, role does belief in God have in promoting health. Future research would best proceed by sampling relatively equal numbers of religious and non-religious participants and comparing them on levels of engagement in religious and secular activities, or in the case of intervention comparisons, assigning religious and non- (not low) religious patients to either R/S or secular treatments.

Is there an “adversarial stance” to the atheism–health research, as claimed by Mrdjenovich (p. 23)? The hypothesis that there is “nothing inherently healthy” about R/S constructs is not an “adversarial stance” or “blanket statement,” but rather a null hypothesis based on the standard scientific principles of parsimony and incremental validity. The burden of proof is on the advocates of “R/S uniqueness,” using proper controls, to demonstrate that there is in fact an inherently healthy R/S mechanism. The authors of the current paper will prove to be easy converts to the “R/S = health” idea when evidence for such a relationship is provided, but until then, we remain in doubt and disbelief.