This section details how Pew Research Center used data on age, fertility, mortality, migration and religious switching from 201 countries and territories around the world to make estimates of global religious populations in 2020, and to project how those populations will change in the years up to and including 2070. It uses a question-and-answer format to go through the methods used by Center researchers in the Future of World Religions report. Click the links for more information: Q1: How did Pew Research Center estimate baseline figures for 2020? Estimating religious composition Estimating age and sex structure by religion Estimating fertility by religion Estimating mortality by religion Estimating migration rates by religion Estimating religious switching Q2: How does Pew Research Center’s projection model work? Q3: How do the results in this study compare with prior Pew Research Center studies? The 2015 “Future of World Religions” projections The 2017 “Europe’s Growing Muslim Population” projections The 2022 “Modeling the Future of Religion in America” projections Q4: How do you define regions, countries and territories? This work was a collaboration between researchers at Pew Research Center and the Center for Advanced Studies of Population and Religion (CASPAR) at Krakow University of Economics. Q1: How did Pew Research Center estimate baseline figures for 2020? The starting point of a population projection is called a baseline, and in this study the baseline is the year 2020. The demographic methods we used to project populations of religious groups required collecting and standardizing massive amounts of input data about these groups (in each country and territory) for the baseline year, including each group’s population size as well as its age and sex composition. We also needed data for each group (in each country and territory) on patterns of fertility, mortality, migration and religious switching. The next sections describe how we measured these baseline characteristics. Estimating religious composition Using census and survey data information about religious identities, we sorted the global population into seven religious groups: Christians, Muslims, Hindus, Buddhists, Jews, followers of other religions and religiously unaffiliated people. The “other religions” category includes Baha’is, Daoists (also spelled Taoists), Jains, Shintoists, Sikhs, Wiccans, Zoroastrians and many small groups, some of which can be described as folk or traditional religions. The religiously unaffiliated population – sometimes called “nones” – includes people who in surveys and censuses answer a religious affiliation question by saying they do not identify with any religion or that they are atheist or agnostic. Most of the census and survey data came from questions that measure religion using a direct question like “What is your religion, if any?” The census and survey questions either ask this question by itself, or, less often, they first ask a filter question such as “Do you belong to a religion or religious denomination?”94 It’s worth noting that we rely on how people describe their own religious identity, regardless of whether that identity lines up perfectly with belief or practice. When someone identifies with a particular religious group, we classify them as being part of that group regardless of their beliefs and practices. For example, if someone identifies as Christian, we classify them as Christian whether or not they hold Christian beliefs that may be considered orthodox or attend worship services. Similarly, if someone identifies with no religion, we classify them as religiously unaffiliated even if they hold at least some religious beliefs and attend worship services. The censuses and surveys we use classify the religious identity of respondents at varying levels of specificity. For example, depending on the source, the most specific level of affiliation measured for a Southern Baptist could be Christian, Protestant, Baptist or Southern Baptist. Researchers coded all religious identities into one of the seven categories used in this study. Sources used to measure religious composition The baseline estimates of the 2020 religious population, which we used as the starting point for our projections through 2070, were first published in Pew Research Center’s 2025 report “How the Global Religious Landscape Changed from 2010 to 2020.” That report was based on more than 2,700 data sources, including censuses; large-scale demographic surveys (which often survey more than 10,000 people); general population surveys (which typically survey 1,000 to 3,000 people); information from population registers; the World Religion Database; and government reports. The vast majority of data underlying the 2020 baseline estimates came from four main types of primary sources: general population surveys; censuses; large-scale demographic surveys; and the World Religion Database. General population surveys were our primary source to measure the religious composition of an assortment of countries and territories that make up 44% of the world’s population. Census data was our primary source in countries where 42% of the global population lives. We relied on large-scale demographic surveys, including Demographic and Health Surveys (DHS), as our primary sources for 10% of the world’s people in 2020. Additionally, we used World Religion Database (WRD) estimates as our primary source in places that are home to 3% of the world’s people. Among the many types of information used to create WRD estimates are statistics gathered from religious communities and assessments from country experts. This type of information is especially valuable for countries like Cuba and North Korea, which have little or no nationally representative census or survey data on religious composition that is publicly available. In a relatively small number of countries and territories that together are home to less than 1% of the world’s people, we used some other kind of primary source. In Finland and Israel, for example, we drew from population registers maintained by the governments of those countries. (In Israel, religion data from the population register also is merged into census reports.) In Gulf Cooperation Council (GCC) countries, we relied on estimates produced for our 2024 report “The Religious Composition of the World’s Migrants” to describe the religious makeup of immigrants in the six GCC states, who made up about half their overall population as of 2020. For more information on the sources used for each country, read Appendix A. Special procedures for estimating the size of small religious groups Religious groups that make up a small share of a population are not necessarily measured reliably in surveys, which generally lack the capacity to measure groups that, for example, may have only a few thousand members in a country of many millions. Additionally, some foreign-born populations may be underrepresented in surveys for various reasons, including that immigrants may not be fluent in the languages in which a survey is conducted. And a survey’s sample frame may exclude refugee camps (or other areas) where foreign-born populations are concentrated. In a given country, when any of the seven groups in our study made up a small share of the national population, we consulted a range of sources to refine our estimates. These sources included, but were not limited to, the World Religion Database, the religious demography sections of International Religious Freedom Reports from the U.S. State Department, the Yearbook of Muslims in Europe, and world Jewish population estimates published in annual editions of the American Jewish Year Book. In some Western European countries, we used surveys to measure the religious composition of the native-born population and a separate procedure to estimate the religious composition of the foreign-born population. In these cases, our foreign-born population estimates frequently rely on estimates we made for the 2024 Center report “The Religious Composition of the World’s Migrants.” Read that report’s methodology for details on how we estimated the religious makeup of international migrants. Children’s religion While censuses often collect information on the religion of people of all ages, most surveys collect information only on adults. In countries where our primary sources provide direct information about the religious composition only of adults, we use indirect demographic estimation methods to estimate the religious composition of children. In many of these countries, we use data on the age structure and fertility rates of different religious groups to estimate the religious composition of children. This approach assumes that children share their mother’s religion. Using surveys of people of reproductive age to make general population estimates Some demographic surveys, such as the Demographic and Health Surveys, gather data on everyone living in a household but measure religious affiliation only among the household members who are in their reproductive years. We used imputation methods to estimate the religious identities of other household members and of households that didn’t include anyone of reproductive age. In most cases, all the adults who are surveyed in a household share the same religious identity. In these cases, our research team assigned a matching religious identity to other household members. In cases where multiple religions were measured among multiple adults of reproductive age, assignment was based on a hierarchy of relationship closeness (in order: spouse, parent, child, other relative). For instance, if a male head of household was not asked about his religion because he was too old to be included in the fertility survey, but he lived with his Christian reproductive-age wife and his religiously unaffiliated nephew, we classified his religious affiliation as Christian. For households in which there were no members of reproductive age, and therefore no information about religious identity was gathered, multiple imputation was applied using the multivariate imputation by chained equations (MICE) algorithm in Stata. There was enough information about adults not of reproductive age in households with adults of reproductive age to allow us to identify age, sex, marital status, educational attainment, household size, region and urbanicity as predictors of religion and to use them in the imputation. Five iterations were executed with these predictors (independent variables) for the missing data on religion. One place where we did this was Nigeria. In the 2018 DHS survey of Nigeria, 11% of individuals lived in households without anyone of reproductive age, and we imputed their religious identities using the MICE approach. Why does Pew Research Center use projections for some 2020 numbers? Generally speaking, the Center uses the term “estimates” to describe population statistics for a year in the past and “projections” to describe demographic modeling of what a population is expected to be in the future. However, in some cases we created projections using older data to make 2020 estimates, because reliable data for 2020 was not available for all countries. For more information about methods used in projections of religious populations, read “How does Pew Research Center’s projection model work?” Why do Pew Research Center 2020 numbers sometimes vary from published census numbers? One reason our numbers may differ from published census figures is that we exclude responses from people who do not answer a religious identity question. Some countries that measure religion in their census make clear that answering the religion question is optional. Yet census agencies in these countries often report results based on the overall population, rather than just on the people who answered the religion question. For example, published religion results from the 2021 census of England and Wales show that 6% of people fell into the “not answered” category.95 However, since we don’t have a “not answered” category in this study, our numbers vary from these published census statistics.96 We don’t know which mix of our seven religion categories best describes the people who don’t answer religion questions. By treating this missing data as invalid, we effectively assume that the people who don’t answer census religion questions have a similar mix of religious identities as those who do provide answers. Another reason for variation in results is that we use the United Nations’ World Population Prospects (WPP) estimates of the number of residents in each country or territory, and these vary modestly from census population counts. Estimating age and sex structure by religion To estimate the religious composition of each five-year, age-and-sex cohort in a population (e.g., men ages 25 to 29, women ages 40 to 44), we begin with data on age, sex and religious affiliation from surveys and censuses. In many countries, individual general population surveys have relatively modest numbers of respondents in each five-year cohort. In those cases, we make age structure estimates using a combination of multiple datasets and demographic smoothing techniques. Estimating fertility by religion We estimated fertility rates using a combination of direct and indirect measures. Some censuses and surveys directly measure recent births or provide data on how many children a woman has ever given birth to (at the time of the survey). In other countries, we had to gather fertility data indirectly − for example, by using information on the current age of a mother’s children to estimate her past birth patterns. These varying sources of fertility data were used to calculate both age-specific fertility rates and total fertility rates for religious groups in each country.97 In many countries, religion-specific fertility data is available for the largest religious groups, but it is not available in sufficient detail for minority religious groups. In Nigeria, for example, more than 98% of women of reproductive age are either Christian or Muslim, and there is sufficient data to estimate the fertility rates of those two groups. For all other religious groups in Nigeria, researchers had to use the country’s overall fertility rate as a proxy. In some countries, there was no religion-specific fertility data available at all. In these cases, researchers applied prevailing national fertility rates to all religious groups equally. Refer to Appendix A for a list of the sources used in each country and territory. Estimating mortality by religion Mortality patterns are measured with data on the death rates experienced by various age groups in a population over a given period. We can transform these period death rates into statistics about the expected longevity (or life expectancy) of people at birth. Because no global data is available on how life expectancy varies by religion within countries, we estimate global life expectancy at birth for each religious group by assuming that people in all religious groups have the average life expectancy of the country in which they live. Therefore, global differences in life expectancy between religious groups reflect their geographic concentrations. For instance, the relatively high life expectancy of Jews reflects the fact that the vast majority of Jewish people in the world live in either the United States or Israel – both of which are places where life expectancy is above the global average.98 Estimating migration rates by religion Pew Research Center’s projections of the populations of religious groups incorporate assumptions about future migration – specifically, our estimates of the rates at which people in each country in the world will migrate to every other country during five-year periods. We then break that down into estimates by age and sex for each religious group (Christians, Muslims, Hindus, Buddhists, Jews, followers of other religions and the religiously unaffiliated). How does Pew Research Center estimate migration rates for its projections? We used several steps to estimate migration rates. First, we looked backward in time and produced estimates of migrant “flows” (the number of migrants who moved from each country in the world to every other country, by origin-destination pair) by age, sex and religion in recent years. To produce these figures, we began with some of the most conservative estimates of recent migration from Guy Abel, a demographer at the University of Hong Kong and a leading expert on migration flows.99 For more than a decade, Abel and his colleagues have published estimates of migration flows in consecutive five-year periods (e.g., 2000-2005, 2005-2010). Over the years, these researchers have expanded their methods and have published additional estimates in publicly available datasets.100 All estimates of migrant flows are based on what demographers call “migrant stocks” – the number of people living outside their country of birth at a given time. The UN Population Division regularly produces estimates of international migrant stocks for each five-year period since 1990, broken down for each pair of origin and destination countries in the world. For example, the UN publishes estimates of the number of migrants who were born in Canada and lived in the U.S. and, separately, the number who were born in the U.S. and lived in Canada, as of mid-1990, mid-1995, and so on for every five years through mid-2020.101 The size of migrant populations in a given country changes over time, as migrants born elsewhere arrive and people born in that country move to other countries. Additionally, some migrants living in their destination country decide to return home to their birth country (“return migrants”). Deaths within immigrant populations also change migration numbers. The various methods that Abel uses to produce flow estimates vary in how they account for changes in migrant stocks over time due to return migration and deaths. Abel provides separate estimates of migration flows for male and female migrants moving between each possible origin-destination country pair in each five-year period since 1990. We built on that approach, using information about the age compositions of the overall populations of sending countries to estimate the number of migrants moving between country pairs by sex and five-year age group (ages 25 to 29, 30 to 34, etc.).102 The migration flows we used in our projections therefore include, for example, estimates of the number of female children under the age of 5 who moved from Canada to the U.S. between 2020 and 2025. For most origin-destination country pairs, we took the average of Abel’s 2010-2015 and 2015-2020 flow estimates to calculate future migration rates. However, we reduced some of Abel’s estimates to account for large migrant flows from those years that reflected refugee surges caused by wars and other crisis events that are not expected to continue indefinitely. (These include crisis events in Myanmar, Syria and Venezuela.) We substituted average flows out of these countries from earlier time periods.103 We also accounted for complex return migration flows between certain origin-destination country pairs. While we judged conservative flow estimates to be most appropriate for the vast majority of country pairs, there are some origin-destination country pairs that are well known to include many migrants who return home. This was particularly important to migration estimates in Gulf Cooperation Council (GCC) countries, which attract large numbers of temporary migrant laborers. The flow estimates from our method – which maximize the number of migrants staying in their country of destination – produced some return flows from the GCC countries back to migrants’ home countries that appeared unrealistically small. Based on information about the relevant immigration and labor policies in these countries, as well as surveys of migrants who had worked in the region and later returned home, we increased the number of return migrants from GCC countries to origin countries outside the Middle East-North Africa region to equal 60% of inflows from outside the region.104 When migrants move from one country to another, they bring their religious identities with them. So, the next step in the process was to estimate the religious composition of each flow from origin countries to destinations. To do this, we applied estimates of the religious composition of immigrant populations for most origin-destination country pairs by using data prepared for our 2024 report “Religious Composition of the World’s Migrants.”105 When analyzing the flow of return migrants – who are moving back home from a country they’ve lived in temporarily – the projections assume that they retain the religion they had when they initially migrated. For example, while data on the religious composition of migrants to and from GCC countries is incomplete, we have reason to believe that temporary migrants who return home to India after working in the United Arab Emirates will have a religious composition more similar to India’s than to the religious composition of migrants who were born in the UAE and move to India. As we did for countries where we adjusted the size of some migrant flows to account for crises, temporary labor patterns and other issues, there were some cases when we found it necessary to modify future religious compositions of migrants, including from some sending countries with low populations of the migrants’ religion, so that projected flows would be plausible. For example, in our 2024 report we estimated that almost 3% of all migrants who were born in Pakistan are Buddhist. If this share were applied to our 2020-2025 migrant flows out of Pakistan, though, it would result in more Buddhists leaving the country during those years than we estimated lived there in 2020. To avoid projecting migration patterns that are implausible, we capped outmigration for all religious groups from all origin countries to never exceed 10% of a group’s estimated population size in any five-year period.106 Finally, once we had a complete set of estimates of the number of people in each migrant flow from country to country by religion, age and sex, we transformed these counts into rates of migration (the percentage of people of each religion in a given country who migrate to another country in each five-year period). For example, we estimated that nearly 139,000 Christians and about 6,000 Muslims moved from Nigeria to the U.S. between 2020 and 2025. As there are 93 million Christians and 120 million Muslims living in Nigeria in 2020, the Christian migration rate out of the country to the United States was roughly 1,500 per million between 2020 and 2025, and the Muslim migration rate to the U.S. was 50 per million. Using this approach, we calculated migration rates from each sending country for each of the seven religious groups. A conservative assumption about future migration rates It is impossible to know the future of migration patterns. Migration is influenced by government policies, economic opportunities, family reunifications, and harder-to-predict calamities like wars and natural disasters. Given the difficulties of anticipating these phenomena, our projections are based on conservative estimates of the number of migrants who’ve moved in the recent past. We assume, then, that migration rates will remain constant during each five-year period. Assuming that migration rates will remain constant is not equivalent to assuming that the size of migrant flows will remain constant. Consider, for example, the flow of Christian and Muslim migrants from Nigeria to the U.S. that was discussed above. The Christian and Muslim populations in Nigeria will change in size over the coming decades, but at different speeds. Consequently, when we apply a constant rate of migration to these populations in the projections, it results in different numbers of Nigerian migrants going to the U.S. in different five-year periods. For instance, we project that 175,000 Christians will move from Nigeria to the U.S. from 2065 to 2070, more than the count from 2020 to 2025 (139,000). For Muslims, the 2065-2070 figure is 12,000 (vs. 6,000 in 2020-2025). Measuring religious switching In this study, religious switching refers to a change from the religious group in which a person says they were raised (during their childhood) to their religious identity now (in adulthood). We use the term “switching” rather than “conversion” because many people who switch identities leave a religion and become religiously unaffiliated, and also because the changes often take place without any formal declaration or ritual. For our analysis we use nationally representative survey data on switching from 112 countries and territories. These surveys ask adults about both their current religion and the religion in which they were raised. These include 93 Pew Research Center surveys conducted between 2009 and 2024; 18 studies carried out under the auspice of the International Social Survey Programme (ISSP) in 2008 and 2018; and the 2018 Chinese General Social Survey.107 As we are interested in capturing data on recent patterns of religious switching, we exclude responses from the oldest respondents in a survey, who tend to have made any changes in their religious identities further back in time than younger respondents. We focus on the switching experiences of adults ages 18 to 54 at the time of the survey. In several places, including Belarus, Russia and Ukraine, we use the switching experiences of adults under 30 (at the time of the survey) as our input data, because the experiences of slightly older adults in these countries seem to reflect a period effect, perhaps connected to the collapse of the Soviet Union. Meanwhile, because men and women often follow somewhat different switching patterns, we calculated rates of switching separately for men and women. In most cases, available surveys include enough respondents to estimate switching rates for both men and women of the same religious group. But in some cases, the sample size is sufficient for one sex but not the other in a particular age group, or only for both sexes combined. In such cases, we use switching information from all respondents ages 18 to 54 as a proxy for the sex group(s) with insufficient sample sizes. For instance, in Germany, due to sample size constraints, we use pooled switching information from all Muslims 18 to 54 to estimate switching patterns for Muslim men and Muslim women alike. For places with multiple surveys available, sources were chosen based on considerations including data recency, sample size and consistency with estimates from other surveys. How we prioritized these considerations (say, data recency vs. sample size) depended on the circumstances. For example, to estimate rates of religious switching in the United States, we used data from the 2023-24 Religious Landscape Study instead of the 2024 American Trends Panel because the former has a larger sample size, allowing us to estimate switching patterns among U.S. Hindus, Muslims, Buddhists and Jews. (Refer to Appendix A for list of sources in each country.) We also considered whether the switching patterns observed in the survey align with recent religious trends. For instance, in Germany, we chose to use switching data from the Center’s Spring 2024 Global Attitudes Survey rather than the Center’s 2017 Western European Survey, despite the latter’s larger sample size. We selected the former because its lower Christian retention rates are more consistent with the recent pace of decline in Christian affiliation in Germany. We chose the Spring 2024 Global Attitudes Survey as the final data source for Italy for the same reason. In Mongolia, Mozambique, the Democratic Republic of the Congo and Uganda, results from the only survey available in each country that measures switching did not match what other surveys (and national censuses) reveal about religious trends in those countries. As a result, we took a conservative approach and assumed no switching (i.e., 100% retention rate) in these countries for both men and women. For example, in Mongolia, the only publicly available data on religious switching – a 2018 study conducted with the ISSP – suggests that Buddhism in Mongolia has recently experienced considerable net growth through religious switching, enough to offset any declines due to other factors. However, larger-scale census data indicates a slight decline in Buddhists as a share of the population. Due to concern that the Mongolia survey data could be misleading, our projections assume no switching for Mongolia.108 In Vietnam, the two available surveys, the Center’s 2023 East Asia Survey and the 2018 ISSP survey, produce conflicting accounts of religious switching patterns. Furthermore, neither dataset aligns with the country’s moderate decline in the share of Buddhists and increase in the religiously unaffiliated population, as measured by Vietnam’s two most recent national censuses. Given these mixed signals, we took a conservative approach and assumed no religious switching in the country. In addition, estimates of religious switching patterns often can be rough due to small sample sizes. Many national surveys with switching data have a total sample size of 1,000 to 3,000 respondents ages 18 and older, and the estimated percentages of people 18 to 54 who were raised in, left, entered or currently are in a religion can be imprecise even for major religious groups. This is especially true for estimating the rates of people who were raised in small religious groups. In places where the sample size is insufficient to reliably estimate movement out of small groups, we assumed there was no switching out of that religion (i.e., a 100% retention rate). This assumption factored into our switching estimates for Jews in the vast majority of countries and territories. Outside of the U.S. and Israel, Jews made up less than 1% of each place’s population in 2020. Consequently, we applied this “no switching out” assumption to Jews in 199 countries and territories because the available surveys did not include enough respondents who were raised Jewish. As a result, our projections likely underestimate the losses, or overestimate the gains, that Jews are expected to experience through religious switching.109 Imprecise estimates also present challenges for estimating the rate of entering a small religious group among people raised in a major group in a country. One example is estimating the percentages of men and women in Paraguay who were raised Christian and now identify as Muslim. Although the survey we used includes a sufficiently large sample of respondents who were raised Christian, cases of conversion to Islam were relatively uncommon, and it remains unclear to what extent the observed switching patterns are attributable to random error, particularly in cases when similar rates are not observed in other age groups or in the opposite-sex group. Because Christians make up more than 90% of Paraguay’s population while Muslims account for less than 1%, a few misreports of Christians becoming Muslims would have an outsize impact on the projected size of the country’s Muslim population. To reduce the noise resulting from potential outliers, we carefully evaluated estimates of movement into smaller religious groups by comparing results across sex and age groups, examining consistency across available surveys, and corroborating the findings against additional evidence, including news articles and missionary reports, regarding the prevalence or rarity of such switching cases. In the case of Paraguay, survey data included a single case of switching into Islam from Christianity by one young man (and zero cases among women of any age). Because this pattern was not observed among men in other age groups, and because there is little documented evidence of Paraguayan Christian men converting to Islam, we decided to manually adjust the rate of switching into Islam from Christianity among men to 0%. We made similar adjustments in several other countries. For example, in Zambia, we manually adjusted the rate of switching into Judaism from Christianity to zero. Sample coverage of religious switching input data Our religious switching input data is based on surveys conducted in 112 countries and territories, encompassing 88% of the global adult population in 2020. For the remaining 12%, or 89 countries and territories, our figures assume that no switching takes place because we have no data measuring religious switching in these places. This may be fairly accurate in some of the places with no available data, such as Saudi Arabia and Yemen, where laws penalize citizens for leaving Islam for a different religion. Our ability to analyze switching varies considerably by religion, ranging from nearly all Hindus worldwide to 71% of Buddhists. The difference is more pronounced when looking at the number of countries with data on switching by religion. For instance, there is sufficient data on people who were raised as Jews to reliably estimate rates of switching out of Judaism in only two countries, Israel and the U.S. In countries where such data is not available, we assume that no Jews leave the faith. Q2: How does Pew Research Center’s projection model work? The technical calculations for the projections in this study were made in collaboration with the Center for Advanced Studies of Population and Religion (CASPAR) and its director, Marcin Stonawski, using an advanced variation of the standard demographic method of making population projections. The standard approach is called the cohort-component method, and it takes the age and sex structure of a population into account when projecting the population forward in time. This has the advantage of recognizing that an initial baseline population can be relatively “young,” with a high proportion of people in younger age groups (such as Muslims), or relatively “old,” with a high proportion of older people (such as Buddhists). Cohorts are groups of people that had a collective experience in a particular time. A birth cohort, the type of cohort referenced in this context, is composed of people born during the same period. Birth cohorts can also be described as males or females who have reached a certain age in a particular year. For example, the separate cohorts of females ages 15 to 19 in the year 2020, and of males ages 15 to 19 in the year 2020, shared the experience of being born between 2001 and 2005. Components are the three ways in which populations grow or shrink: new entrants via births, exits via deaths, and net changes from migration. Each cohort of the population is projected into the future by adding likely gains (from births and people immigrating to the country) and subtracting likely losses (from deaths and people emigrating from the country) year by year. The very youngest cohorts, those ages 0 to 4, are created by applying age-specific fertility rates to each female cohort in the childbearing years (ages 15 to 49).110 The cohort-component method has been in existence for more than a century. First suggested by the English economist Edwin Cannan in 1895, then further improved by demographers in the 1930s and ’40s, it has been widely adopted since World War II. It is used by the UN Population Division, the U.S. Census Bureau, other national statistical offices, and numerous academic and research institutions. An advanced variant of this approach, known as multistate cohort-component projection, became viable starting in the 1970s thanks to the availability of mainframe computers and work by the American geographer Andrei Rogers, among others. The multistate approach permits simultaneous projection of the religious groups included in this study, taking into account variation by religious group in age, sex, childbearing patterns, and propensity and direction of migration. This approach also enables modeling of religious switching as a transition between religious groups.111 For this study, we made projections for 201 countries and territories that had populations of at least 100,000 people in 2010 or 2020. Collectively, these places were home to 99.98% of the world’s population in 2020. Projection inputs for each country – including data for each religious group on fertility, age structure, migration and religious switching rates – were used for multistate cohort component projections going out to the year 2070. The country-level 2020 population data and fertility trajectories are based on the 2024 revision of the UN’s World Population Prospects. The projection model assumes that fertility differences within countries between religious groups will gradually converge by the year 2100, as educational differentials between the groups diminish. The main projection scenario extends the recently observed (estimated) religious switching rates into the future. This report focuses on the main scenario with fertility convergence (i.e., narrowing fertility differentials between religious groups); constant religion-specific out-migration rates by age, sex and destination; and constant religious switching rates. To illustrate the relative influence of demographic factors, the report also presents alternative, hypothetical scenarios with: (1) no differences in fertility by religious groups; (2) no migration flows between countries; and (3) no religious switching between religious groups. Fertility In the main scenario, all religious groups within each country are assumed to see their fertility levels slowly converging toward identical fertility rates by 2100 – 80 years from the baseline year of the projections. The assumption that fertility differences gradually will diminish within countries is based on evidence that when people live in the same economic and social milieu, their fertility patterns tend to become increasingly similar over time. The adoption of a relatively conservative, 80-year time frame for within-country convergence reflects the assumption that geographic clustering, differing education levels and other factors may perpetuate distinctive childbearing patterns among some religious groups. At each step of the main projection scenario, fertility for the total population of a country follows the U.N. Population Division’s medium variant assumptions from its 2024 Revision of World Population Prospects. An alternative scenario assumes that there are no religion-specific differentials in fertility. In this scenario, all religious groups are assumed to have the same level and pattern of fertility as the total population of a country throughout the projection period. By comparing the results of this alternative scenario with the main scenario, we can highlight the expected effects of differences in fertility. Migration The main scenario assumes constant religion-specific out-migration rates by age, sex and destination calculated for the baseline period. In each iterative time period of the projections, out-migration and immigration flows are calculated by age, sex and religion based on the baseline rates and destination matrices. An alternative, hypothetical migration scenario assumes no international migration anywhere in the world. Though clearly implausible, this scenario enables comparisons that show the expected effect of migration on the religious composition of countries and territories. Religious switching The main projection scenario includes constant religious switching rates starting from the 2020 baseline estimates. These switching rates are used to calculate the number of people in certain age and sex groups who move from one religion to another at each five-year interval of the projections. For comparison purposes, an alternative scenario models what would happen if nobody changed their religion during their lifetimes. In countries for which switching data is available, researchers generated recent rates of switching. The main projection scenario assumes that emerging cohorts will switch from their childhood status at the same rate observed in recent survey data. As new cohorts progress from ages 15 to 29 in the projections, they experience these rates of change. For example, if 20% of the adults measured in a recent survey say that they grew up Christian but are now religiously unaffiliated, the model projects that 20% of Christian children ages 10 to 14 going forward will disaffiliate by the time they are 30 to 34.112 The model takes into account recent patterns of switching in all directions. For projection purposes, each person was allowed one switch in their lifetime, which is all that is directly measured in the surveys this method draws upon. Though this does not capture the complexity of individual religious identity and switching, it captures aggregate patterns of change. Mortality Based on the UN’s medium scenario assumptions, this study projects that life expectancy at birth will gradually increase in all countries. This scenario assumes that each country, regardless of its current economic condition, is moving toward better living standards and, therefore, longer life expectancy at birth. We use the UN’s data on the different mortality patterns of men and women in each country. Projections software package The R software package “Measuring Religious Change” (measuringreligiouschange), developed by a team from CASPAR, is a tool for generating demographic projections of religious group populations around the world. It is available at https://github.com/CASPAR-PL/measuringreligiouschange. The software package includes the input data used in this study. The software package can generate religious group population projections at the global, regional and national level resulting from initial age-sex composition of populations, demographic processes (fertility, mortality and migration) and religious processes (transmission of religion between generations and religious switching). The package is a flexible tool that can be used to numerically reproduce the projections presented in this study, generate other projection variants based on the included input data, or create projections from user-supplied input data and adjustments to demographic and religious processes. This software may be most useful for users with training in demographic analysis and some experience using and programming in R.113 Disclaimers Due to the difficulty of anticipating future trends, some cautionary words are in order. Population projections are estimates built on current population data and assumptions about demographic trends, such as declining birth rates and rising life expectancies. The projections are what will occur if the current data is accurate and the trends play out as expected. But many events – scientific discoveries, armed conflicts, social movements, political upheavals, natural disasters and changing economic conditions, to name just a few – can shift demographic trends in unforeseen ways. This study does not attempt to model how religious switching patterns may change in the future as countries experience changes in education, urbanization, political governance and economic development. It is uncertain how economic growth might impact levels of religious affiliation. Economic development could be accompanied by increasing rates of religious disaffiliation in countries that currently show little movement toward disaffiliation. Conversely, countries that currently have large numbers of religiously unaffiliated people could experience circumstances that lead to a rise in religious affiliation. Q3: How do the results in this study compare with prior Pew Research Center studies? The 2015 ‘Future of World Religions’ projections The Center published its first global demographic projections of the world’s religious groups in 2015, projecting global religious change between 2010 and 2050. Our latest projections, focusing on global religious changes between 2020 and 2070, differ from the earlier effort in several aspects. The religious landscape of many countries shifted considerably from 2010 to 2020, and our new baseline data reflects these changes as well as updated measures of recent fertility, religious switching and migration patterns for each religious group. We also have made some methodological changes. These changes were instituted for our previous report “How the Global Religious Landscape Changed from 2010 to 2020.” We adjusted our data sources and methods to make our estimates for 2010 and 2020 as reliable and comparable as possible. Most consequentially, our estimates of the size of religious groups in China – a country that holds roughly one-sixth of the world’s population – rely on different measurements in the two studies. In the latest study, our baseline estimates for China are based on religious (zongjiao) affiliation, as these are the most comparable with what we use in every other country, as well as the most readily comparable over time in China. In our 2015 projections report, we took a broader approach and relied not only on zongjiao affiliation but also on non-zongjiao belief and practice when estimating the size of China’s religious groups. Our old estimates of China’s 2010 religious composition relied on analysis of questions asked in a 2007 survey that have not been asked in a comparable manner in subsequent years. Starting with our 2025 report on global religious change, we have used our new approach to measuring religion in China for both 2010 and 2020. As a result, the estimates from these two approaches differ considerably. For example, using the earlier approach, we estimated that religiously unaffiliated people made up 52% of China’s population in 2010, compared with 87% under our revised approach. Given China’s large population, this difference has a substantial impact on global estimates of the religiously unaffiliated population. (For more details on our earlier and revised approaches, read “Our new approach to measuring religion in China” and “Our previous estimates of China’s religious composition.” For general background, read our report “Measuring Religion in China.”) Another change in our measurement approach in all countries is that we no longer include a separate category for people who practice what some scholars call “folk religions,” because they are not consistently captured in surveys and censuses, and because the religious beliefs and practices of followers of folk religions vary widely from place to place. Due to the limited comparability of this category across countries, they are instead included in the “other religions” category. Read “Why we revised our estimates for 2010” for more details. Our latest projections also incorporate improvements in our input data. For example, in this 2026 study, we estimated religious switching in 112 countries, up from 70 countries in the 2015 report due to increased data availability over the past decade. For example, we now incorporate switching data from China, which we did not do in our earlier study. Because of the changes in the methods and data used in the religion projections published in 2015 and those published in 2026, the results of the two studies are not directly comparable. The 2017 ‘Europe’s Growing Muslim Population’ projections In this study, we project that Europe’s Muslim population will continue to grow over the coming decades. The expansion of the Muslim population is particularly pronounced in some parts of Europe. In our 2017 study “Europe’s Growing Muslim Population,” we projected Muslim population growth in the 27 member states of the European Union, plus Norway, Switzerland and the United Kingdom.114 For shorthand, we will refer to this group of nations as “Western and Central Europe,” though we acknowledge that this term is inexact.115 In the medium scenario of our 2017 projections, Muslims were expected to account for 11.2% of these nations’ populations combined in 2050. In our new projections, Muslims are projected to reach 9.6% of these nations’ populations combined in 2050, up from 5.0% in 2020. The difference looks even larger when it comes to the projected number of Muslims. In the earlier report, we estimated that under the hypothetical scenario that all refugee flows into Western and Central Europe were to stop as of mid-2016 but that recent levels of “regular” migration to Europe (i.e., migration of those who come for reasons other than seeking asylum) would continue, there would be 58 million Muslims in Europe in 2050. This figure is higher than our latest projection of 49 million. The discrepancy between our earlier and latest estimates of the Muslim population in Western and Central Europe can be explained by differences in the input data we relied on for the projections. It also highlights the fact that projections are not predictions; their accuracy depends on how closely our assumptions about demographic patterns that were observed in the past align with those that unfold in the future. For example, many European countries have adopted more restrictive immigration policies since the surge of migration in 2015 and 2016. In some cases, countries have tightened their rules for admitting not only asylum seekers, but also migrants who move to Europe for economic, educational or family reasons. The earlier projections to 2020 and beyond assumed that levels of regular migration would continue. Our new 2020 baseline estimates for Muslim populations in Western and Central Europe are based on new data that captures variation from our earlier projection. Our new projections also reflect updated estimates of fertility differences between Muslims and non-Muslims in Western and Central Europe. In the 2017 report, we estimated a total fertility rate in the 2015-2020 period of 2.6 children per woman for Muslims and 1.6 children per woman for non-Muslims. Our current estimate, based on more recent measures of fertility patterns, is that the 2020-2025 total fertility rate for Muslim women in this region is 2.0 children per woman, compared with 1.4 children per non-Muslim woman. The 2022 ‘Modeling the Future of Religion in America’ projections Our latest projections indicate that the United States will no longer have a Christian majority by 2070. In the coming decades, Christians are expected to decline to less than half of the U.S. population, while the percentage of religiously unaffiliated people will rise substantially (though remaining smaller than the Christian share of the U.S. population). These findings about the future U.S. religious landscape largely align with projections we published in the 2022 study “Modeling the Future of Religion in America.” In the earlier report, we presented projection results from a “steady switching” scenario that assumed the switching patterns observed among Christians, religiously unaffiliated people and followers of all other religions would continue in the future at the same rates observed at the time of the study. This projection scenario is similar to the “main” projections scenario in our 2026 study, which also assumes that recently observed switching rates among religious groups continue. In the current study, we project under the “main” scenario that Christians will make up 48% of the U.S. population in 2070, while religiously unaffiliated “nones” will constitute 41%. By comparison, in the 2022 study, we projected (under the “steady switching” scenario) that Christians would make up 46% of the U.S. population in 2070, while “nones” would reach 41%. Although these two studies appear to share the same 2020 estimates and arrive at the same figure for the religiously unaffiliated share of the U.S. population in 2070, they are not directly comparable. The studies differ in many respects, including the input data used to model religious switching, intergenerational transmission and fertility differences among religious groups. Moreover, the 2022 report focused only on the two largest religious groups – Christians and the religiously unaffiliated – while collapsing Jews, Muslims, Hindus, Buddhists and other smaller groups into the broader category of “other religions.” By contrast, the latest study considers the future trajectories of more groups (including Jews, Muslims, Hindus and Buddhists) individually. On the other hand, in the 2022 study, focusing only on the largest religious groups in a single country (the United States) allowed us to explore a wide range of projections scenarios, including scenarios with different assumptions about switching patterns and rates of religious transmission from mother to child. This was possible because of the availability of repeated, detailed measures of religious switching in the United States. Due to data limitations, it is not feasible to extend such scenarios to all other countries. Q4: How do you define regions, countries and territories? The word “country” in this study refers to all countries recognized as such by the United Nations. The word “territory” in this study does not have a technical definition. Rather, it is a general term for distinct geographical entities that the UN doesn’t recognize as countries, but for which it provides separate population estimates. Territories in this study include such entities as Hong Kong and Macao (special administrative regions of China), Greenland (an autonomous constituent country within the Kingdom of Denmark) and the Commonwealth of Puerto Rico (an unincorporated territory of the United States). List of countries and territories in each region This study groups 201 countries and territories into six major regions: Asia and the Pacific; Europe; Latin America and the Caribbean; the Middle East and North Africa; North America; and sub-Saharan Africa. The 50 countries and territories in the Asia-Pacific region are: Afghanistan, Armenia, Australia, Azerbaijan, Bangladesh, Bhutan, Brunei, Cambodia, China, Cyprus, East Timor (also called Timor-Leste), Fiji, French Polynesia, Guam, Hong Kong, India, Indonesia, Iran, Japan, Kazakhstan, Kiribati, Kyrgyzstan, Laos, Macao, Malaysia, Maldives, Federated States of Micronesia, Mongolia, Myanmar (also called Burma), Nepal, New Caledonia, New Zealand, North Korea, Pakistan, Papua New Guinea, Philippines, Samoa, Singapore, Solomon Islands, South Korea, Sri Lanka, Taiwan, Tajikistan, Thailand, Tonga, Turkey, Turkmenistan, Uzbekistan, Vanuatu, and Vietnam. The 42 countries and territories in Europe are: Albania, Austria, Belarus, Belgium, Bosnia and Herzegovina, Bulgaria, Channel Islands, Croatia, Czech Republic, Denmark, Estonia, Finland, France, Georgia, Germany, Greece, Hungary, Iceland, Ireland, Italy, Kosovo, Latvia, Lithuania, Luxembourg, Malta, Moldova, Montenegro, Netherlands, North Macedonia, Norway, Poland, Portugal, Romania, Russia, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Ukraine, and United Kingdom. The 37 countries and territories in Latin America and the Caribbean are: Argentina, Aruba, Bahamas, Barbados, Belize, Bolivia, Brazil, Chile, Colombia, Costa Rica, Cuba, Curacao, Dominican Republic, Ecuador, El Salvador, French Guiana, Grenada, Guadeloupe, Guatemala, Guyana, Haiti, Honduras, Jamaica, Martinique, Mexico, Nicaragua, Panama, Paraguay, Peru, Puerto Rico, St. Lucia, St. Vincent and the Grenadines, Suriname, Trinidad and Tobago, U.S. Virgin Islands, Uruguay, and Venezuela. The 20 countries and territories in the Middle East and North Africa are: Algeria, Bahrain, Egypt, Iraq, Israel, Jordan, Kuwait, Lebanon, Libya, Morocco, Oman, Palestinian territories, Qatar, Saudi Arabia, Sudan, Syria, Tunisia, United Arab Emirates, Western Sahara, and Yemen. The two countries of North America are: Canada and the United States. The 50 countries and territories of sub-Saharan Africa are: Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cape Verde, Central African Republic, Chad, Comoros, Democratic Republic of the Congo, Republic of the Congo, Djibouti, Equatorial Guinea, Eritrea, Eswatini (formerly Swaziland), Ethiopia, Gabon, Gambia, Ghana, Guinea, Guinea-Bissau, Ivory Coast (also called Cote d’Ivoire), Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritania, Mauritius, Mayotte, Mozambique, Namibia, Niger, Nigeria, Reunion, Rwanda, Sao Tome and Principe, Senegal, Seychelles, Sierra Leone, Somalia, South Africa, South Sudan, Tanzania, Togo, Uganda, Zambia, and Zimbabwe. A note on rounding In most countries, it is not possible to precisely count the number of people who identify with each religion. Because our estimates are not precise, the tables in this report and accompanying data files show rounded population counts (often to the nearest 10,000) and show counts below 10,000 as “<10,000.” In the text, many estimates are rounded to the nearest million or percentage point.
Methodology
Full Article
Original Source
Read the full article at Pewresearch →KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.