Comparability — Armenia [top]

Persons who refused to respond are included in the unknown category.

Comparability — Austria [top]

The 2001 sample identifies subgroups of Christian Catholic and Orthodox in the unharmonized source variable.

Comparability — Bangladesh [top]

The categories are completely comparable between years.

Comparability — Benin [top]

The census asked each household person what religion their practice. The 1992 and 2002 samples indicate that minors are usually assigned the same religion as they parents, but it would also be appropriate to ask this question of the minors themselves.

Comparability — Brazil [top]

The 1991, 2000 and 2010 samples reported many more categories of religion than the earlier samples.

Comparability — Burkina Faso [top]

The categories are completely comparable between years.

Comparability — Cambodia [top]

Persons with "no religion" are included with "other." The universe changes from all persons in 1998, 2004 and 2008, to present persons in 2013. The universe for the 2019 sample reverts back to include all persons.

Comparability — Canada [top]

"No religion" in 1981 includes "other" religions not specified.

Comparability — Chile [top]

The universe changes from all persons in 1960 and 1970, to persons age 14+ in 1992 and 15+ in 2002.

Comparability — Côte d'Ivoire [top]

In addition to ''other'' religions, the 1998 sample has a separate category for ''other Christian.''

Comparability — Egypt [top]

The 2006 and 1986 census includes "Jewish" as an additional category, which was considered as "Other" in the 1996 census.

Comparability — Fiji [top]

The categories are largely comparable across years, but there is variation among the identified "Other Christian" groups.

Comparability — Germany [top]

The 1970 sample offers more detailed religion breakdowns of the "other Christian" and "other" categories in a separate unharmonized source variable.

Comparability — Guatemala [top]

"Other" in 1964 includes "no religion" not specified.

Comparability — Guinea [top]

The universe is all persons in 1983 and 2014, while it includes only household residents in 1996.

Comparability — Haiti [top]

The 1971 categories are less detailed than the later years. Weslyans are combined with Methodists in 1982, but can be separately identified in the source variable.

Comparability — Iran [top]

In the 2006 census, children who haven't attained puberty yet are considered to have the father's religion or any religion they are practicing.

Comparability — Ireland [top]

The universe for the Ireland 2002, 2006, 2011, and 2016 samples excludes persons who were absent on census day.

The 1971 and 1981 censuses include relatively more categories with respect to other samples. In Ireland 1991, 2002, and 2006 the "Anglican" category includes both the "Church of Ireland" and "Protestants," which were separately identified in 1971 and 1981. The 2016 sample only differentiates between Roman Catholic and other religions.

Comparability — Jamaica [top]

A small number of persons in certain collective dwellings were not asked the religion question in 1982 and 1991.

The categories in all three censuses are broadly comparable. The Muslim and Hindu religions are merged into one category in Jamaica 1982 and classified under "other". Judging from the 2001 sample, there the category likely contains somewhat more Hindus than Muslims.

Comparability — Laos [top]

The 1995 sample has a separate category for Animism, while the 2015 sample separates out Bahai. Animism, Christianity, Islam, and other religions. In 1995 and 2005, no religion is included in "other" while no religion is separately identified in 2015. Data has not been separated as records of questionnaire.

Comparability — Malawi [top]

For both samples, if a child's religion is not known, he/she is assigned the parents' religion, or the mother's religion if the parents have different religions.

Comparability — Mauritius [top]

The 1990, 2000 and 2011 distinguish between types of Hindu and Muslim religions, but these distinctions are preserved in the corresponding unharmonized source variables.

Comparability — Mexico [top]

The universe was all persons in 1960, 1970, 2010, and 2020, as well as persons age 5 and older in 1990 and 2000.

The Mexico 2000, 2010, and 2020 samples have much more detail. In 1960-1990, the possible responses to the religion question were specified on the census questionnaire. In 2000, the questionnaire had printed responses only for "none" and "Catholic," while the "other" responses were to be written into a space on the form. In 2010, only blank spaces were offered to write responses. For Mexico 2000, 2010, and 2020 samples instructions required to be specific and refer to the person's church, congregation, or religious denomination. The 2020 sample also takes into account other religions from Asia as well as popular cults.

Comparability — Pakistan [top]

The 1998 sample identifies persons in a "scheduled caste."

Comparability — Papua New Guinea [top]

The census asked each person what church or religion they belong to. Persons who do not belong to a church are considered to have "no religion".

Comparability — Paraguay [top]

In all samples respondents were asked to indicate if their religion was Catholic, other, or none. In the 1962 sample, "Catholic" and "Christian not Catholic" are the only unique response categories. The 1992 and 2002 censuses asked respondents to specify non-Catholic religions. These samples have several unique response categories. The age universe changes after 1962 to persons age 10+.

Comparability — Peru [top]

The 2017 census reports more category detail among Christian groups. In 2007, '' Evangelical protestant'' includes both the ''Christian'' and ''Evangelical,'' which were separately identified in 2017.

Comparability — Philippines [top]

The 2000 sample reported many more categories of religion than the 1990 sample; similarly, the 2010 sample included additional categories with respect to the 2000 sample.

Comparability — Romania [top]

The 1992 and 2002 samples distinguish between "no religion" and "atheist."

Comparability — Rwanda [top]

The 2002 and 2012 census questions suggest that religion for children would be that of their parents, unless each parent reports a different religion. In the latter situation, enumerators were to ask each child about their religious beliefs. If a person refused to answer the question, they were considered to be reporting "No religion".

Comparability — Senegal [top]

Senegal in 1988, 2002 and 2013 samples identifies distinct groups (Khadrya, Layene, Murid, Tidiane) of Muslim separately. These can be distinguished in the unharmonized variable.

The universe in 2013 includes only residents in ordinary households, but it includes all persons in 1988 and 2002.

Comparability — Sierra Leone [top]

The 2015 census restricts the universe of respondents to private households.

Comparability — South Africa [top]

The 1996 census asked the question only of persons in private households. Response to the religion question in 1996 was optional, as stated on the questionnaire.

Comparability — Suriname [top]

The 2004 and 2012 samples are generally comparable, except that Christians appear grouped in 2004 while specific Christian religions are separately identified in 2012. The 2012 sample also distinguishes types of Hindu and Muslim religions, which are available in the unharmonized source variable.

Comparability — Switzerland [top]

In 2011, the Christian categories are less detailed compared to previous years.

Comparability — Thailand [top]

The question is consistently directed to all persons in all samples. In the 1970 and 1980 censuses, the child's religion was considered to be the same as the father if the child was not able to give an answer; in the 1990 and 2000 censuses, the child's religion was considered to be the same as either the father or the mother, if the child was not able to give an answer and the parents had different religions.

Comparability — Togo [top]

The 2010 census reports more category detail among Christian groups. The 1970 sample does not identify a "no religion" group, presumably including them among "other religions."

Comparability — Trinidad and Tobago [top]

The 2011 sample distinguishes between "no religion" and "atheist." The 2011 census asked the question only of persons in private households. There are significantly less detailed response categories for the 1970 and 1990 censuses.

Comparability — Uganda [top]

Persons with "no religion" are coded to NIU in 1991 and in "other" in 2002, while they are separately identified in 2014.

Comparability — United Kingdom [top]

Data for religion were spread across three separate variables for the different regions of the United Kingdom. Some categories for Scotland and Northern Ireland are combined with others in RELIGION but can be recovered using the unharmonized source variables.

Comparability — Vietnam [top]

Relative to most other countries, Vietnam has a high proportion of persons claiming no religion.

The category 'Buddhist' comprises various types of Buddhism, including Hoa Hao.

Comparability — Zambia [top]

There is a minor universe difference between samples.

In 2000, the question was asked of all persons. In 2010, the question was asked of present household members and visitors. The 2000 enumeration instructions indicate that this question should have been asked only of present household members and visitors, but religion is reported for almost all absent household members; they are included in the universe.