Several authors describe the limitations of the early phase of literature in this area of research and the recent evolution to more sophisticated data collection and research designs of new studies, including experimental research designs. Waite and Lehrer (2003, p. 256) note that many of the early studies in this literature [religion and health] suffer from methodological shortcomings, including small, unrepresentative samples, lack of adequate statistical controls and a cross-sectional design that confounds causality.
Limited National Data Collection
The shortcomings are due, in part, to limitations in the availability of national data sources with relevant information on religion, spirituality, and health. Nationally representative longitudinal studies that collect detailed health outcomes generally lack information on religious denomination and attendance, let alone more extensive measures of religion and spirituality. Therefore, links between religious beliefs or practices and the prevention or treatment of specific health issues cannot be examined using these national surveys. At the same time, studies that collect detailed measurement of religion rely on convenience samples that can have limited variability in religiosity and/or are too small to detect statistically significant differences. Further, most of the data collection is cross-sectional. The lack of systematic nationally representative longitudinal data collection with both extensive religiosity measures and health assessments (medical test results as well as self-ratings of health) makes drawing generalizable conclusions in this research field particularly challenging (Fagan, 2006).