Religious landscape in Brazil: Comparing different representative nationwide approaches to obtain sensitive information in healthcare research.

Although the basis of religious studies start with demographics, nation-wide data are often extracted from face-to-face interviews (leading to a social-desirability bias) and in studies not originally designed to assess religion. This study aims to understand the religious landscape in Brazil and to investigate the feasibility of carrying out a representative nation-wide survey without interviewers, comparing it with other representative face-to-face surveys. We conducted a nationwide online survey representing all regions in Brazil. These results were compared with five other Brazilian representative surveys. A total of 1169 individuals completed the online questionnaires. The percentage of participants according to the gender and in relation to the Brazilian region was quite similar for all surveys. However, the online survey had the higher level of education among all surveys. In relation to the religious characteristics, the percentage of each variable varied from survey to survey. Those surveys originally designed to assess religion tended to yield more religious affiliations and less ceiling effects. The online survey was able to identify more diverse religious affiliations and more balanced responses in the religious attendance and importance of religion in life. The present study found that, even in nation-wide representative surveys, there are important differences in the results obtained while investigating religion. These differences could be associated with the type of data collection (face-to-face and online), the design of the study (originally designed to investigate religion or not) and options and the type of the questions used.