Role of Religiosity in the Lives of the Low-Income Population: A Comprehensive Review of the Evidence. Data Sources


The data sources used in low-income religiosity studies across outcome indicators are highlighted in Table 8-1. Notably, for most of the topical areas, except for health, there is research that uses nationally representative panel studies to follow individuals over time. The limitation is that most of the national longitudinal data sources do not include rich multi-variable religiosity measures and they are limited to measures of religious denomination and attendance. These national panel studies described throughout this report include, for example, the National Longitudinal Study of Youth (NLSY) and the Panel Study of Income Dynamics (PSID). Similarly, nationally representative studies that include detailed modules focused on religion, such as the General Social Survey (GSS), do not include a comprehensive set of policy-relevant outcome measures.

New longitudinal studies, such as the National Study of Youth and Religion, are starting to fill in the research gaps. There are also newly designed studies that concentrate specifically on religion and family relationships, but data have only been collected at one point in time. These surveys can include multiple family members that are interviewed, such as both members of couples or parents and children. The newer surveys generally use more detailed measures of both organizational and individual religiosity and spirituality that include general measures as well as religiosity measures specifically relevant to family process/relationship outcome measures. In addition, these surveys include parallel secular measures. These newer data sources can be extended so that families can be followed over time.

Across the outcomes, there are also data collection efforts that draw representative samples from low-income communities. These studies measure the effects not only of individual religiosity, but also of community religiosity. For this review, studies focusing on low-income communities occur across all outcome areas, although there are only one or two community studies within each outcome.

Within the health area, there are two studies that draw on administrative records. Administrative data can be used to draw specialized samples or be used to formulate outcome measures. For example, in one study, administrative data are used to draw a random sample of mothers receiving welfare assistance. Another study uses patients medical records as an outcome measure in a randomized trial of educational programs hosted at churches to promote breast cancer screening.

Lastly, with the exception of marriage, there are studies across the outcome measures that draw on convenience samples, which can be drawn from existing studies taking place within correctional institutions, hospitals, or clinics to study the role of religiosity and spirituality. Convenience samples can also be collected specifically to study the effects of religiosity and spirituality on a particular population. Samples can be drawn from the social service population, such as homeless shelters. These samples are most common in the health, substance abuse, and criminal justice fields.

Table 8-1.
Summary of Religiosity Findings for the Low-Income Population: Number of Research Studies, Data Sets, and Research Methods
  Marriage Parenting Youth Health Substance Use Crime and Violence
Number of Research Studies Identified 11 13 17 37 10 5
Data Set  
Nationally representative (or large cities) longitudinal panels n n n   n n
Nationally representative cross-sectional n n n      
Low-income neighborhoods n n n n n n
Administrative records       n    
Social service clients     n n    
Other convenience samples   n n n n n
Program intervention participants       n n  
Research Method            
Single equation linear/nonlinear models OLS, logit/probit, seemingly unrelated regressions n n n n n n
Testing for basic mediators n n n n n n
Simultaneous equation models Includes structural equations       n n  
Instrumental variables     n      
Basic linear unobserved effects models Fixed effects n   n      
Propensity score matching     n      
Duration analysis Event history/hazard models n          
Experimental study design       n    
Quasi-experimental study design       n n  

View full report


"report.pdf" (pdf, 818.46Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®