Measuring Cross-National Variations in Religiosity and Attitudes Toward Science and Technology Using Machine Learning

Abstract

ABSTRACT This study uses resampling methods and machine learning to measure how religio-scientific groups are distributed across regions, countries/territories, and religious groups. Across 76 societies (N = 143,092), the distribution of class membership is as follows: traditional (31.9 percent), modern (23.7 percent), post-secular (30.3 percent), and postmodern (14.1 percent). Although most societies are dominated by a single class, there is evidence of significant heterogeneity within societies in class prevalence. Those with post-secular views are both religious and feel favorably toward science; however, when faced with a conflict between religion and science they tend to support religion. Ultimately, societies with large traditional and post-secular classes are significantly more likely to support religion given a conflict with science; in contrast, the reverse is true for societies with large modern and postmodern classes.