The Handbook of Religion and Health positioned an examination of the association, or lack thereof, between religion and health. Some contemporary medical, psychological, and public health researchers and practitioners have positioned religiosity and theism (e.g., prayer or church attendance) as a protective factor with regard to health and well-being. The resulting conclusion that religiosity improves one’s health has a salient and meaningful impact on not only people who are religious and/or theistic but also on those who are neither religious nor theistic. By supporting that church attendance improves health, researchers and practitioners tacitly or declaratively exclude nonreligious and/or atheist people from the possibility of maximal health and well-being. Additionally, perspectives on the association between religion and health remain biased. As one example, the author typed “is religion good for your health” (in quotations) into Google and received 3,860 hits– one of which was text by Harold Koenig titled Is Religion Good for Your Health: The Effects of Religion on Physical and Mental Health?. Comparatively, the search term of “is religion bad for your health” resulted in only five hits and no books. Additionally, when the author conducted a review of articles available through the Ovid database on the topic of health education and religion, the results showed similar biases. In the field of health education, religion or theism related articles accounted for 8.03% (1,436/17,887) of published health education articles, whereas non-religion or atheism accounted for 0.13% of articles (24/17,887). In other words, the number of religious/theistic articles was about 60 times greater than non-religious/atheistic articles. The number of religion or theism related health education articles increased from 236 between the years of 1995 and 1998 to 553 between the years of 2003 and 2006. During the same time periods, the number of non-religion or atheism related health education articles increased from two to nine. Furthermore, in 2002, the United States Center for Complementary and Alternative Medicine reported that prayer was by far the most common form of complementary and alternative medicine with over 55% of adults in the United States using prayer for health reasons. Additionally, in the United States, faith has increased its impact on policy and politics in recent years (e.g., abstinence until marriage sexuality programs in public schools).
I decided to run some quick analyses regarding the association between religiosity and health. The results were expected to support one of three groups: 1) fundamentalist religionists, 2) fundamentalist atheists/secularists, or 3) liberal religionists who follow the social gospel and secularists/atheists who focus on social determinants of health. The fundamentalist religionist would position that in and of itself religion would have a positive influence on public health, regardless of social determinants of health. The fundamentalist atheist/secularist would position that in and of itself religion would have a negative influence on public health, regardless of social determinants of health. The liberal religionist and secularist/atheist would presume that religion may be associated with social determinants of health and public health, but social determinants of health would attenuate the associate between religion and public health. Four research questions guided the study.
1) Is religiosity associated with health?
2) Is religiosity associated with murder?
3) Is religiosity associated with robbery or burglary?
4) Is religiosity associated with social exclusion?
Methods and Justification
Various data sets and data sources exist to address these research questions. For the sake of simplicity and considering the less research orientated reader, the author chose to use statemaster.com as the data source. Statemaster.com offers various data sets from several data sources on dozens of topics. By using statemaster.com, the reader could quickly access data sets in one location and examine other relationships. As with any research, the use of statemaster.com trades rigor for simplicity, which the author accepts as a limitation for the purpose of this article.
Seven variables were included in the analyses used to address the four research questions; all of the variables were measured at the state level within the United States. Religiosity was measured as the percentage of people within each state who categorized themselves as nonreligious during the American Religious Identification Survey. Health was defined by the state ratings of health as part of the Morgan Quinto Press’ Health Index. Murder was defined in two ways: 1) murder was measured as the per capita rate of homicide by state according to the United States Department of Justice, and 2) murder was defined as the historical per capita rate of completed capital punishments (or carried out death penalties) by each state. The per capita rate of robbery per state, as measured by the United States Department of Justice, defined the robbery variable, and the per capita rate of burglary per state defined the burglary variable. Lastly, social exclusion was defined as states that voted for the defense of marriage act in 2004.
The following research results are based on linear correlations. The analyses were run at the state level, not at the level of individuals within states. There are five types of correlations. First, one variable increases as another variable increases (e.g., as ages increase from birth to 18 years, height increases). Second, one variable decreases while another variable decreases (as caloric intake decreases, weight decreases). Third, one variable decreases while another variable increases (e.g., as number of minutes of aerobic physical activity decreases, weight increases). Fourth, one variable increases while another variable decreases (e.g., as number of minutes of aerobic physical activity increases, weight decreases). Fifth, there is no linear relationship between the two variables (e.g., as SAT scores increase, the college grade point average of Harvard University scholarship students increases). One and two are called positive correlations; three and four are called negative correlations; and five means there is no correlation. A variable is defined as a group of data that has more than one level. Age in years is a variable people have various ages. The number of suns in our solar system is constantly one and therefore not a variable in our solar system. Due to concerns regarding small sample size (i.e., a maximum of 50 states in any given analysis) and interest in increasing power (the ability to find an effect when one is truly present; also known as a hit), the type 1 error rate (the probability of finding an effect when one is truly not present) was set to .10 instead of the more traditional .05. As a result, false alarm would occur in one out ten analyses instead of one out of twenty analyses. Contemporary justifications have also supported using a more liberal type 1 error rate in social science and public health research. Increasing the type 1 error rate decreases the likelihood of missing an effect when an effect is truly present (type 2 error).
Results
1) At the state level, do higher levels of religiosity predict better health?
Higher levels of religiosity do not predict better health at the state level. In fact, the exact opposite association emerges. Higher levels of nonreligiosity were associated with higher levels of health (n=47, correlation=.291, p<.10). In other words, more religious states displayed lower levels of health (correlation=-.291, p<.10). The magnitude of the correlation was +.291 with p value of less than .10 meaning that there was a statistically significant association between religiosity and health at the state level. This correlation would be considered medium or large in magnitude according to Jacob Cohen’s effect size rules of thumb for correlations.
2) At the state level, do higher levels of religiosity predict less murder?
Higher levels of religiosity do not predict lower levels of homicide at the level of the state. Again, the opposite trend is found. Higher levels of nonreligiosity were associated with lower levels of homicide (correlation=-.248), and higher levels of religiosity were associated with higher levels of homicide (correlation=.248), which is medium in magnitude. Additionally, the association between capital punishment (total number of carried out death penalties per capita) and percentage of religiosity was tested. Again, states that have historically reported higher numbers of capital punishment per capita were associated with higher levels of religiosity (correlation=.520), and states with historically lower numbers of capital punishment per capita displayed higher levels of nonreligiosity (correlation=-.520). According to Jacob Cohen’s rules of thumb regarding the effect size of correlation supports that this relationship shows a large effect.
3) At the state level, do higher levels of religiosity predict less robbery and/or less burglary?
With regard to statistical significance, religiosity was not associated with robbery or burglary. However, examination of the trends showed small effect sizes of .103 and .093, respectively. The results again ran in the opposite direction of what would be expected from the religious perspective. Higher levels of religiosity trended towards higher levels of robbery and burglary, and higher levels of nonreligiosity trended toward lower levels of burglary and robbery.
4) At the state level, do higher levels of religiosity predict higher levels of social exclusion?
Using states that voted for the defense of marriage act in 2004 as a proxy for social exclusion, a statistically significant association was found between religiosity and social exclusion. States that displayed higher levels of religiosity were more likely to support social exclusion meaning that states reporting higher religiosity were more likely to support the defense of marriage act (correlation=.317; p<.10), which is a medium to large effect. States reporting less religiosity were more likely to not support social exclusion (i.e., the defense of marriage act). Additional evidence of social exclusion of atheists can be found in the research of the American Mosaic Project. The American Mosaic Project found that atheists are more distrusted and less accepted than other marginalized groups such as people categorized as homosexual, African-American, Muslim, or immigrants. Intolerance toward atheists increased with higher levels of religiosity.
Given the faith-based perspective of some religionists and atheists in the United States, many covariates that a grounded public health researcher would typically include would be neglected. The analyses up to this point simply examined religiosity’s associations with public health outcomes from a fundamentalist perspective that neglected well established impactful variables of the natural world. Moreover, to anyone who has knowledge of public health in the natural world, it would be obvious to include many other variables (e.g., poverty). With the exclusion of obviously important variables such as poverty, religiosity at the state level seems to support the fundamentalist atheist position. The research presented in this article should not be used as evidence of causation but could be used as evidence of associations between religiosity and public health outcomes at the level of the state. Furthermore, in order to avoid the ecological fallacy, data at the state level, as presented in this article, should be interpreted at that level and not at the level of the individual.
Public health researchers and practitioners who inform federal policy based on state level data should continue to focus on the natural world and other social determinants of health to avoid misleading fundamentalist positions. Through examination of the association between per capita poverty rates and religiosity at the state level, one finds that states with higher levels of per capita poverty display higher levels of religiosity (correlation=.271, p<.10). When exploring per capita child poverty and religiosity at the state level, a similar significant trend was found (correlation=.277, p<.10). With the exception of capital punishment, including state level child poverty as a confounding variable in analyses of religiosity and health outcomes significantly attenuates the association between religiosity and health outcomes. For example, the unique variance in the Health Index accounted for by child poverty is 51.7% (semipartial correlation=-.719), whereas the unique variance accounted for by religiosity is only 0.6% (semipartial correlation=.08). Capital punishment is, however, an exception in this research because in the presence of one another both religiosity and child poverty are associated with a significant proportion of the variance in capital punishment. In this case, child poverty explains 25.8% (semipartial correlation=.508) of the variance in capital punishment, and religiosity explains 9.9% (correlation=.315) of the variance in capital punishment. However with the edition of other social determinants of health variables, this association would attenuate as well. These results support the position of the liberal religionist and liberal atheists/secularists.
Public health researchers and practitioners should remain critically objective in addressing the questions of religion and health. David Seedhouse has stated a rather convincing case that health promotion activities are inherently prejudiced, and health promoters’ prejudices often blindly lead them to decisions based more on values than objective evidence. In the case of religion, values and faith guide thought and behavior far more than evidence and reason for fundamentalists. As a result, health promoters should be more not less weary of studies positioning the fundamentalist perspective of the impact of religion. Health promoters must remember that supporting religion as a method of health improvement is only justified if the evidence supports positive health impacts of religion. It is not justified to purport religions health value simply because many people are religious and have faith that religion positively impacts health. To correct for these prejudices, health promoters must undergo a process of self-examination as well as an examination of the field in general.
In closing, in the United States, the message of religion improving health is a more common and more socially compatible position in popular culture than examining negative impacts of religion on health or social determinants of health’s attenuating the effects of religion. As a result of the often one sided position regarding religion and health, atheists and the non-religious suffer further social exclusion and marginalization. Additionally, some of the research findings within this research have also be shown to exist at the level of nations through the research of Phil Zuckerman, which was published in The Cambridge Companion to Atheism. To paraphrase and adapt Maurice Ogden’s poem The Hangman, to enable and reify immoral acts such as implicitly or explicitly supporting harmful and unbeneficial public health programs because it helps you for the moment, does not necessarily save you from a broader net of social exclusion in the future.