1. Introduction
Tis chapter ofers a systematic overview of the World Religion Project (WRP) that provides the informational baseline for the empirical study of religion and world politics. We discuss the underlying logic of coding data on religious adherence over a long period of time and across the globe.
We explain how data were generated, the kind of problems we confronted in generating the dataset, and the solutions we devised. Finally, we ofer a systematic description of religious adherence in the post–World War II era on a global scale. Te purpose of a review of the WRP is to provide a transparent and replicable report on the dataset that we use for analyzing the role of religion in world politics. Tis will provide information about the advantages and shortcomings of our approach.
2. Measuring Religion: The General Logic of the World Religions Project
In this section, we provide a brief discussion of the general rationale of the WRP, and some details about the data collection and management process.
A more detailed methodological exposition may be found in the appendix at the end of the chapter.
Te empirical study of religion and various aspects of world politics has been based on a number of datasets that examine religion in specifc
115
domestic political contexts (e.g., Putnam and Campbell 2012) or in inter-national contexts (Vanhanen 1999a b; Pearce 2005; Fox 2016). Other datasets ofer a cross-national and longitudinal perspective on the distribu-tion of religious adherents across time and space. Tese datasets formed the basis for many of the empirical studies discussed previously. Te Correlates of War (COW) cultural dataset (Singer 1997) includes decennial obser-vations on religious groups for states, but the obserobser-vations end at 1990.
Te criteria for classifying and diferentiating among major religions, reli-gious families, and relireli-gious denominations are not clear in these datasets.
Ellingsen (2000) provides annual estimates of the religious characteristics of states based on averaging values from three sources (i.e., Te CIA World Factbook, Britannica Book of the Year, and the Demographic Yearbook) and interpolating missing years. Tese data go to 2000; they identify the larg-est and the second-larglarg-est religious groups in the country; and they have a straightforward list of nine major religions (Animism, Atheism, Buddhism, Hinduism, Islam, Judaism, Christianity, Shintoism, and Syncretism).
Te WRP provides data at fve-year intervals from 1945 to 2010 on the religious adherents of states coded for fourteen major religions, and also data on religious families within some of the major religions. It draws from multiple sources and also provides novel measures of the religious charac-teristics of states in terms of religious diversity and religious similarity. Te WRP builds on existing datasets but improves on them in several impor-tant ways. We document these below. As such, it opens new and expanded opportunities for the study of the role of religion in world politics.
One of the distinguishing characteristics of the WRP is that our approach to the process of data collection was deductive in nature. Before collecting data on world religions, we had to come up with satisfactory answers to basic questions about the nature of religion and the classifca-tion of world religions. Tese answers were crucial in developing a “religion tree,” that is, a systematic classifcation of major world religions and of religious families within each of these major religions. Tis, as we quickly noticed, was in stark contrast to the common practice in the scholarly disciplines that deal with these matters. Tese generally lack an explicit set of principles for deciding what constitutes a religion or for classifying religions and religious families.
Te frst step was to develop a defnition of religion that is both consis-tent across disciplines and also is most relevant to the process of generating a world religions dataset. Te defnition of religion has been a focus of intense
debate not only among theologians and religious studies scholars but also among sociologists, anthropologists, and philosophers. We had no preten-sion of developing yet another defnition. Consequently, we conducted a comprehensive literature review designed to answer several questions that were essential to our enterprise.
1. What is a religion?
2. Can we fnd observable or tangible indicators that facilitate the systematic identifcation of religions?
3. Are there tangible indicators that allow us to distinguish among diferent religions, and among religious families/denominations within a given religion?
4. How can we validate our criteria for a religion and/or religious families within a given religion?
Without going into too much detail, our literature review covered multiple felds of scholarship that focus on the study of religion such as history, reli-gious studies, anthropology, sociology, law, political science, and philoso-phy as well as specifc regional studies.2 Te upshot of this review was that we could not fnd a consensus on any of the issues listed above. Te defni-tion of religion we adopted (Alston, 1967, 142) was one whose elements appeared most often in other defnitions of religion, and which ofered the clearest indications of tangible elements of religions.3 Tis defnition states that a religion is a belief system held by an individual or a group that contains the following elements:
• belief in supernatural being/s (god/s)
• a distinction between sacred and profane objects
• ritual acts focused on sacred objects
• a moral code believed to be sanctioned by the supernatural being/s (god/s)
• characteristically religious feelings (awe, sense of mystery, sense of guilt, obligation, duty, adoration), which tend to be aroused in the presence of sacred objects and during the practice of rituals, and which are connected in idea with the gods
• prayer and other forms of communication with gods
• a worldview or a general picture of the world as a whole and the place of the individual therein; this picture contains some
specifcation of an overall purpose or point of the world and an indication of how the individual fts into it
• a more or less total organization of one’s life based on the worldview
• a social group bound together by the above
Clearly, this is a very general, but also quite vague, defnition of religion.
Other defnitions are equally focused on beliefs, moral codes, and rituals.
Empirically, one of the most important and commonly mentioned criteria for the classifcation of religions and religious families or denominations is self-identifcation. Tis, however, has limited application in the context of our project. Since we had to trace religions across time, surveys or census data were not available for most states and for most time points. Even when such data were available, the religious categories from which respondents had to select varied from one survey/census to another. In order to create a spatially and longitudinally stable defnition of religion, we had to develop observable or tangible criteria that can guide data collection. Tis required heavy reli-ance on secondary data. Given this state of the literature, our next objective was to derive a set of fairly tangible criteria that defne religions and religious families independent of subjective identifcation of people with a given belief system. Te review yielded a defnition of religion as a belief system shared by a community of people that is identifed by the following set of criteria.
a. Scriptures. A central text or a set of texts that—as a whole—
encapsulate the general principles of the belief system of a given religion. Te existence or type of scriptures is a key identifer of some but not all religions. Some religions do not have major scriptures; others incorporate scriptures from other religions with scriptures that are exclusive to a particular belief system. But the absence of central scriptures is also an important distinguishing feature of a religion, provided it possesses the other characteristics.
b. Institutions. A set of formal or semiformal institutions that are responsible for interpreting the basic beliefs for adherents, modi-fying them or changing them over time, training and ordaining spiritual leaders of their communities—and determining who is a believer and who is not. Te nature, size, and formality of institu-tions vary a great deal across religions. However, virtually every religion has a set of institutions or a group of individuals who interpret beliefs for adherents. In many cases, these institutions or
priests have primary responsibility for leading the practice of col-lective rituals. Tis set of institutions must be unique to that belief system and distinct from the institutions of other belief systems.
c. Historical development. It is possible to trace the origin, develop-ment, and difusion of religions in terms of historical turning points and/or identifable processes. At these turning points or processes, an individual or a group has formed the basic principles of a new religion. Tis individual or group is also responsible for forming certain institutions that guide religious practices, ordain spiritual leaders, and defne principles for inclusion or exclusion of believers.
Tis criterion is an important identifer not only of major religions but also of religious families and denominations. Since many reli-gions developed in connection with other relireli-gions, the historical context in which religions were formed helps identify their origins.4 d. A common class of beliefs, rituals, and practices. Tis criterion allows identifcation of a broad set of religions that is characterized by polytheism. It also characterizes religions that contain identifable rituals, which are followed without a clear set of institutions or historical evidence of how they were formed. Because such reli-gions have developed or existed in areas that are geographically distinct and noncontiguous, they cover a wide variety of ritualistic elements. Yet there are some basic commonalities in the (rather weak) institutional structure of such religions and their rituals.
Tese allow the grouping of such religions into a separate cat-egory. However, this criterion does not ofer a simple way of divid-ing such religions into denominations and subdenominations.
Tese criteria provide a foundation for an operational defnition of reli-gion. A belief system with these four characteristics is clearly identifed as a religion. As noted, there are religions that lack a major scripture. But the acceptance of a common scripture as a central guide of beliefs, rituals, and practices is almost a sufcient—though not necessary—identifer of a reli-gion. Te same applies to a set of institutions that defnes the ethical and ritualistic codes for the community. Note that the key elements of Alston’s defnition do not apply to nontheistic religions or nontheistic versions of such religions as Buddhism, Hinduism, Jainism, and Confucianism. Nor does the last item in his defnition tell us what that group or community is or what exactly binds such a community together.
Te operational defnition we ofer allows us to identify religions empir-ically. Using these criteria we scanned the literature identifying major reli-gions, religious families within major relireli-gions, and denominations within the religious families, which resulted in a three-level structure of the reli-gion tree: major relireli-gions, religious families, and denominations. For exam-ple, Christianity contains several religious families such as Catholicism, Protestantism, Eastern Orthodox, and Anglicanism. Protestantism contains a number of denominations such as Presbyterian, Methodist, Mennonite, and so on. It became clear very quickly that it would be unlikely to obtain systematic, consistent, and reliable data going back in time on the denomi-national level. Terefore, our project contains data only on the frst two levels of the religion tree: major religions and religious families.
Once the structure of the religion tree was set, we surveyed multiple classifcations of religions, focusing in particular on those that were con-sistent with our observable criteria. Tis process was challenging due to the substantial variation in religious classifcations in diferent sources and disciplines. Some classifcations were fne-tuned, going down to the level of ethnically or “tribally” specifc rituals unique to some regions (or within a given country). Other classifcations were quite general, subsuming broad categories of religious groups. Given this state of afairs, we developed a
“candidate” religion tree of the most plausible lists of religions/religious families in the literature.
Next, we formulated a questionnaire with several groups of questions.
One group requested that experts assess the validity of the criteria for identifying religions and religious families. Te other group asked about whether a given candidate religion qualifes as one of the major world religions. Likewise, for each candidate for a major world religion that was derived from the literature search, we asked whether respondents consider this to be a “proper” religious family. Te survey also contained a set of open questions that asked respondents to specify additional criteria for identifying major religions or religious families, and for adding or deleting religions or religious families. We surveyed noted religion scholars from various disciplines as a way of validating or modifying our religion tree.
Table A1 (in the appendix) provides a summary of the survey responses to the key questions.
Te survey yielded a number of results. First, we received strong con-frmation of four of the fve criteria we had thought to use in identifying
major religions and religious families. In the open questions asking whether we had omitted any important criteria, we found no signifcant responses that would cause us to reconsider these criteria. Second, two-thirds of the twenty-one candidates to the position of “major world religion” received a validity score of six or higher. With respect to the remaining candidates, answers to open questions suggested collapsing them with the “surviving”
candidates. Respondents also mentioned a very low number of adherents as an alternative reason for entering a low validity rank for certain reli-gions. Some of the candidate religions were collapsed into a broader group.
For example, various animistic religions were collapsed into the animist group. Syncretic religions represent a wide array of belief systems, but they have some common underlying characteristics, so we decided to group them together. Also, the “other religions” category is a residual category that refects either religions with relatively few adherents, religions that did not make it into the list, or a complement of a population that is known to practice some religion but there is no information about which religion/s they do practice (more below).
Tird, with respect to religious families, we also found a substantial degree of agreement for most of the candidate families. In this case, low-validity religious families were grouped into “other” (e.g., “other Christians,” “other Muslims,” etc.). Fourth, we found no relationship between the degrees of agreement across diferent disciplines of respondents. Finally, the number of responses to some questions varied signifcantly (range = 30−67), because many of the respondents had expertise that covered a specifc region or a specifc religion. Tis made some respondents hesitant to respond to ques-tions or items that were outside of their professional purview.
Table A4.1 in the appendix to this chapter presents the major world reli-gions and religious families included in the dataset. It also includes infor-mation regarding the rate of agreement among experts in our expert poll regarding the validity of a given religion/religious family. Te experts were asked to rate on a scale of 1–10 the degree to which they believe a given cat-egory constitutes a major religion or a religious family within a given religion.
In general, we used a rule of thumb of an agreement rate of six or above for a given religion. Te only exception was the non-religious category, which was rated lower. Tis category includes atheists, agnostics, and people who stated (or were estimated) as having no particular religion. Te low ratings for this category were due to low response rates and (possibly) a confusing defnition
of this category in the questionnaire. Nevertheless, virtually every source we used provided data for this category. We also used a residual (other religions) category, but did not include this in the survey. Tis category includes adher-ents of religions that did not make it into the fnal religion tree as well as data reported by most sources under the same (other) label.
3. Trends in World Religions, 1945–20105
Before presenting the data, some caveats about trends within nations and within religions over time are in order. First, our approach is state-reliant. We started our data collection of the distribution of religious adherents within a given territory when this territory became an independent state, following the COW defnition of system membership. Tis means that the interpretation of the data presented below should refect the changes in the size and char-acter of the interstate system over the period 1945–2010. Second, the avail-ability and reliavail-ability of data have improved considerably over time. We have more sources and fewer missing data for later years compared to previous periods.6 Terefore, the fgures provided below may refect some temporal bias. Tird, domestic political changes may have also afected reporting on religious afliations. With these caveats in mind, we turn to a discussion of some major trends in religious adherence from 1945 to 2010.
3.1 Global Patterns
Figure 4.1 describes the distribution of adherents of some of the key religions over time. For convenience of presentation, the fgure does not include all of the major world religions in the dataset. However, the discussion covers all of them. Several points are noteworthy. First, the proportion of religious adher-ents has increased over time from a low of about 63.5 percent of the world’s population in 1945 to a high of about 88.9 percent in 2010. Non-religious percentages varied between 9 percent and 16 percent of the world’s population over the same period, with the highest percent of non-religious people reported in 1975: 16.6 percent. Tis refects two statistical artifacts rather than a major trend of increased religiosity. One concern, as noted above, is that a signifcant proportion (30 percent) of the world’s population did not live in independent states in 1945. In contrast, by 2010, 99.9 percent of the world’s population resided within independent states. Terefore, the increased trend in religiosity
0.00 0.˜0 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90
˜.00
˜945 ˜950 ˜955 ˜960 ˜965 ˜970 ˜975 ˜980 ˜985 ˜990 ˜995 2000 2005 20˜0
Pct. of World Population
Non-Religious Year
Christianity Islam Buddhism Hinduism Other Religions
Fig. 4.1. The distribution of major world religions, 1945–2010
refects the fact that many of the new states that emerged over the 1945–2010 period have an especially high proportion of religious adherents. Improved data quality over time may also afect these trends. One can argue, however, that since the mid-1970s, the changes in the size of the international system have been relatively marginal, and that these changes refect a rearrangement of the system (secessions and partitions of multiethnic entities—e.g., the Soviet Union, Czechoslovakia, Ethiopia, Yugoslavia) into more ethnically homoge-neous states. So at least some of the increase in religiosity refects an actual trend rather than a combination of statistical artifacts. We leave this question open to future investigations.
Second, once we consider changes in the composition of the interstate system, Muslims are the only religious group that shows a consistent and signifcant upward trend relative to the world population. Hindus also have exhibited a growth trend, but this has been more modest. Te other major religious groups capture a relatively stable proportion of the world’s population (e.g., Christians, non-religious) or display a declining trend (e.g., Jews, animists, syncretics). Figure 4.2 provides two important items of information. Te frst (top panel) shows percent changes in the
Second, once we consider changes in the composition of the interstate system, Muslims are the only religious group that shows a consistent and signifcant upward trend relative to the world population. Hindus also have exhibited a growth trend, but this has been more modest. Te other major religious groups capture a relatively stable proportion of the world’s population (e.g., Christians, non-religious) or display a declining trend (e.g., Jews, animists, syncretics). Figure 4.2 provides two important items of information. Te frst (top panel) shows percent changes in the