Health and Christianity: Controlling for Omitted Variable Bias by Using the Data of Twins and Siblings

Abstract

To account for potential bias from heterogeneity in hereditary factors or family background, I use the within-twin and within-siblings differences to estimate the effects of Christian faith on three health outcome variables by applying fixed effect model to the data of twins and siblings from the first wave of the National Survey of Midlife in the United States (1995). Both this model and other statistical tests and model without controlling for omitted variable bias confirm significant positive health effects of religiosity of Christianity. The results also support the three explanatory mechanisms of religion on health proposed by Son and Wilson (2011): 1) behaviors and lifestyles, 2) social networks and 3) social support and psychological resources. However, the data also suggests that either other channels through which religiosity affects health may exist or the mechanism of psychological resources goes far beyond of “good moods” and contains much more plentiful and profound connotations that is relevant to health.