Pew Research Center has published a methodology explanation for its projections of religious populations from 2020 through 2070. The document details the collaborative work with the Center for Advanced Studies of Population and Religion at Krakow University of Economics and lays out how baseline estimates were built and how future change is modeled.
The baseline for the study is the year 2020. Researchers collected and standardized input data for 201 countries and territories on the size and age-sex composition of seven religious groups, along with patterns of fertility, mortality, migration and religious switching. Groups used in the analysis are Christians, Muslims, Hindus, Buddhists, Jews, followers of other religions and the religiously unaffiliated. “Other religions” includes Baha’is, Daoists, Jains, Shintoists, Sikhs, Wiccans, Zoroastrians and many small and folk or traditional groups.
Baseline estimates draw on more than 2,700 sources. The main primary sources are general population surveys, censuses, large-scale demographic surveys and the World Religion Database. General population surveys are the primary source for about 44% of world population, census data for about 42%, large-scale demographic surveys for about 10%, and World Religion Database estimates for about 3%. In a small number of places, population registers or migrant composition estimates from a 2024 report on world migrants were used.
For small religious groups, Pew refined estimates with additional sources such as the World Religion Database, U.S. State Department International Religious Freedom Reports, the Yearbook of Muslims in Europe and the American Jewish Year Book. Where primary sources covered adults only, indirect demographic estimation was used for children, often assuming children share their mother’s religion. Imputation methods, including multivariate imputation by chained equations, were applied when surveys measured religious affiliation only among reproductive-age adults.
Age and sex structures by religion are estimated from survey and census data, with smoothing techniques where sample sizes in five-year cohorts are modest. Fertility rates are estimated with direct and indirect measures, and religion-specific rates are used where data allow; otherwise national rates serve as proxies. Mortality is modeled by applying country-level life expectancy to all religious groups within a country, so between-group differences reflect geographic concentration.
Migration projections build on estimates of migrant flows by Guy Abel and colleagues, derived from UN international migrant stocks. Flows are averaged over recent five-year periods and adjusted for crisis-driven surges and for return migration, notably in Gulf Cooperation Council countries where return rates were increased. Religious composition of flows is applied using estimates from the Center’s 2024 report on the religious composition of the world’s migrants, with caps to avoid implausible outmigration levels.
Religious switching is measured from nationally representative surveys in 112 countries and territories covering 88% of the global adult population in 2020. Switching rates are calculated for adults ages 18 to 54, with sex-specific rates where sample sizes allow. Where data are insufficient, conservative assumptions such as 100% retention are applied, especially for small groups like Jews in most countries.
Projections use a multistate cohort-component method that tracks births, deaths and net migration for each religious group by age and sex, plus transitions due to switching. The main scenario assumes fertility differences between religious groups within countries converge gradually toward a common rate by 2100, constant religion-specific out-migration rates by age, sex and destination, and constant switching rates observed around 2020. Life expectancy is assumed to rise according to UN medium-variant assumptions.
The Center notes limitations and alternative scenarios. Alternative runs test no fertility differences by religion, no international migration and no religious switching. The methodology document also compares the current work with earlier Pew projections, highlighting changes in China measurement, inclusion of switching data for more countries, and updated fertility and migration inputs.
- Categories:
- Technology



