Religious Benefits to Children Unproven, Despite Media Hype
Amanda K. Metskas
On October 13, 2007 the Boston Globe ran a story headlined, "Religious upbringing found to aid children." The story reported on the results of a study by Rajeev Dehejia, Thomas DeLeire, Erzo F.P. Luttmer, and Josh Mitchell entitled, "The Role of Religious and Social Organizations in the Lives of Disadvantaged Youth."
While the article correctly points out some of the conclusions drawn by Dehejia et al. in their paper, it underplays the important caveats and cautionary notes that are included in the paper, leaving the reader of the Globe article, and especially the headline, with the conclusion that the study's findings are more conclusive and far-reaching than they really are.
-
More Parenting Analysis:
- A Wellness Perspective On Scientific Studies by Dr. Don Ardell
- Religion and Child Development by Dr. Jean Mercer
First of all, it is important to note that the main results of the study showed a correlation between religious attendance of disadvantaged parents and two specific outcomes for the children from those families – completing high school, and being a non-smoker. There was not a significant relationship between religious attendance of disadvantaged parents and the subsequent income level, health, or psychological wellbeing of their children.
The Boston Globe article gets this right in the body of the article, but a quick glance at the headline suggests a much broader effect on children's well-being, and also suggests that these effects hold for children across the socio-economic spectrum, instead of applying only to those with a disadvantaged upbringing.
Additionally, the Globe describes the families in the study as "religious" and "nonreligious" when it would be more accurate to describe them as "church attending" and "not church attending." This is important because with the language the Globe article uses it suggests that the study differentiates between people of different beliefs, when really the study differentiates between people of different behaviors. The majority of the people in the study who reported that they did not attend church must have a religious affiliation because the authors report that 25% of respondents reported that they go to church once per year or less, yet only 8% of the respondents in the study identified as having "no religion," (Table 3, and Table 4). This means that at least 68% of the non-attenders reported a religious affiliation.
Clearly a newspaper article must simplify the results of a study, and a headline must simplify even more, but readers should be skeptical of these simplifications because they often make the results of studies appear to be more conclusive and far-reaching than the actually are. Newspaper articles are also unlikely to put the results of a study in the context of other studies on the same topic, and thus may give readers an impression of more scientific consensus about results than truly exists.
With regard to the Dehejia et al. paper, it is important to note that the paper has not yet been published in a peer-reviewed journal. It is a working paper that was presented at a conference, not a published final product. I point this out not to suggest any problems with the quality of the paper which may well pass peer-review and be published in the future, but merely to point out that the Globe article accorded the results of the study a degree of authority that would probably be better reserved for academic papers that have passed a peer-review process. Peer-review is the gold standard in scientific research because it ensures that fellow experts in a field of study evaluate the quality of a paper before it is accepted for publication.
Correlation not Causation
From my reading of the Dehejia et al. paper it seems to be a methodologically sound and well-researched piece, and the authors are careful to point out the limitations of their findings at several points in the text. Two important caveats that the authors mention in their paper are underemphasized or absent in the Globe article. The first is that as the authors point out the results of this study are correlational rather than causal. The authors note:
Because participation in a religious or social organization is a choice that a child's parents actively make, we must be cautious in interpreting the buffering effect of religion as a causal effect of religious participation. For example, the effect of participation could be confounded with other coping strategies that families adopt in response to disadvantage, leading to our estimated buffering effect to capture the combined effect of all of these strategies. (Dehejia et al., 2007, p. 3-4)
The association between increased religious attendance of parents and better outcomes for children could be the result of omitted variables. In families that are socio-economically disadvantaged higher church attendance rates could be related to a number of other important factors that influence the quality of family life. For example, those parents who attend church more in these circumstances might be the same parents who are more focused on providing a good upbringing for their children, and these parents and families may simply have their act more together than those disadvantaged families that attend church less. To assess whether these omitted variables are the cause, it would be important to look at why those who do not attend religious services are not attending – is it because they do not believe, or are they believers who don't go to church?
Dehejia et al. make these points very clearly in their paper, further pointing out that:
In particular, it is possible that parental religious involvement is more strongly associated with omitted characteristics that affect later outcomes for disadvantaged children than it is for non-disadvantaged children. For example, it is possible that parents who participate in religious activities out of concern for their children growing up in a disadvantaged environment might also decide to enroll their children in after-school activities that could mitigate the effects of disadvantage. (Dehejia et al., 2007, p. 15)
The authors go on to say that they would like try to separate out these effects, but they did not have the data necessary to do it for this study.
While the Globe article mentions that the study finds a correlation, and that "religious parents might have other attributes that contributed to the study's findings," the headline and tone of the article strongly suggest that there is underlying causality.
Alternate Explanations
The second problem with the Globe article is potentially more serious. The Globe article does not mention that the Dehejia et al. paper also included measures of disadvantaged parents' involvement in several types of social groups, and they report that "leisure clubs also seem to confer substantial buffering," (p. 24). They define leisure groups as "sports groups, youth groups, hobby or garden clubs, or literary/art groups," (Table 4). The authors don't find a similar benefit for other types of social groups like work-related organizations, political groups, and veterans' groups. That isn't surprising, because the "leisure groups," unlike the other social groups studied, seem like activities that a parent would be involved in with their kids.
Given the data about the role of leisure groups, one could easily argue that disadvantaged families who take the time to be involved in activities with their children like attending church or "leisure groups" have children with better educational outcomes later on. This interpretation draws on the point that the authors themselves make that disadvantaged parents may choose to get involved in activities with their kids to try to help them overcome that background, and attending church is just one of those structured family activities.
Although the Dehejia et al. article presents interesting findings that begin to shed some light on ways to improve some outcomes for disadvantaged children, the authors' conclusion that "religion plays an important role in how households respond to the disadvantages they face," is premature (p.29). This premature conclusion is exacerbated by the Globe article which makes simplifications that lead readers to conclude that the findings are stronger than the authors of the original paper suggest.