Comparability — Argentina [top]

The relationship information is consistent across censuses.

Comparability — Austria [top]

The 1981-2010 samples are fully consistent in their categorizations, except that persons in collective dwelling are not included in the census sample. The 1971 sample combines children and children-in-law in a single category, and it does not identify grandchildren, who are included among "other relatives."

The parent/grandparent category was called "ancestor of head" in the Austrian samples, and it may include small numbers of other older-generation relatives.

Institutional inmates include persons in prisons, military installations, nursing homes, and other dependent situations. The unharmonized source variables provide detailed breakdowns.

Comparability — Bangladesh [top]

The census samples do not identify any specific relationship codes beyond heads, spouses, and children. In 1991 and 2001 other relatives and non-relatives are combined into a single category, while in 2011 these are separately identified. In 2001, some households included multiple heads; these second heads can be identified in the unharmonized source variable and are classified as other relatives/non-relatives here. A substantial number of cases had unknown relationship in 2011.

Comparability — Belarus [top]

In 2009, "child" includes biological and step-children. The 1999 sample refers only to "daughter or son."

Comparability — Benin [top]

In 2013, the "child" category is split to differentiate between biological children, adopted children, and children-in-law. Enumeration instructions in other samples are not explicit about the classification of adopted children and children-in-law. The census question in 1992 and 2002 refers to children of the household head, which suggests this category may only include biological ones.

Parents and parents-in-law are included in one category in 1979, while they are separately listed in other Benin samples. Grandchildren are only identified in 1992, 2002, and 2013.

The 1979 sample includes domestic employees and friends and visitors. The 1992, 2002, and 2013 samples identify instead a category for non-relatives.

Comparability — Bolivia [top]

The 1976 and 2012 samples separately identify grandchildren while the 1992 and 2001 census include grandchildren in "other relatives."

Comparability — Botswana [top]

In 1981 and 1991, "in-law" is considered to be "other relative, not elsewhere classified", since enumeration instructions do not clarify which specific relatives are included in this category.

The 2001 and 2011 samples have more detailed information: they identify separately biological, step-, and in-law children, and include additional categories for parent-in-law and grandparent.

Comparability — Brazil [top]

The level of detail for relationship varies considerably over time in Brazil, with 1991 and 2010 offering the most detail. Samples prior to 1991 combined some of the categories that were presented separately on the census questionnaires.

Comparability — Burkina Faso [top]

Relationship data are not available in 1985. Codes are largely consistent between 1996 and 2006. Although a code for parents was included on enumeration forms in both years, parents appear to be included with other relatives in the 2006 sample. Grandchildren are identified in 2006 only, while cousins are identified in 1996 only.

Comparability — Cambodia [top]

The 1998 and 2008 samples provide no distinction of the type of children, while biological, adopted or foster, step, or in-law children are identified in 2004, 2013 and 2019.

The 2004, 2013 and 2019 samples also provide specific categories for parent-in-law, sibling-in-law, niece or nephew, and servants, which are not separately available in 1998 or 2008.

Comparability — Cameroon [top]

The 1976 and 1987 samples provide only limited information on relationship. The 2005 sample provides detailed relationship codes, identifying grandchildren, parents, siblings, and in-laws. The 2005 sample typically distinguishes between the head's biological and step-children. If a children of the householder who are living with their own spouse or children, they are combined into a single category regardless of their biological relationship to the household head; we have classified them as "biological status unknown."

Comparability — Canada [top]

Grandchildren, and nephews and nieces are identified in 1971-1981 and grandchildren or great-grandchildren in 2011, but these cases are included with "other relatives" in 1991 and 2001.

In 1971 parents are combined with parents-in-law and siblings are combined with siblings-in-law. These four relationships are separately identified in 1981-2001. In 2011, parents-in-law, children-in-law, siblings, and siblings-in-law are seemingly grouped with other relatives; this sample also identifies parents and step-parents as a single category.

Only the 1981 sample identifies group quarters residents in RELATE. Those persons are coded as "non-relative" in other samples.

The "unknown" category in 1981 consists of persons who did not permanently reside in the household.

The modern Canadian samples are all samples of individuals; persons are not organized into households in the data.

Comparability — Chile [top]

The Chile 1960 to 2017 samples progressively identify more categories for the relationship to household head. Most of the changes are among relatives of the head outside the nuclear family.

The 1960 combines spouses and unmarried partners, while later years distinguish them separately. The 2002 and 2017 samples separate children from stepchildren. The 1960 sample has an undifferentiated category for other relatives. The 1970 sample has a single category for all non-relatives. The 2017 sample identifies civil unions, which are assigned a distinct sub-code within "unmarried partner" in RELATE.

See the source variables for additional details.

Comparability — China [top]

The samples have generally consistent categorizations.

In-laws are not distinguished from blood relatives in 1982 or 1990.

The 2000 sample has separate categories for children, children-in-law, parents (including step-parents), and parents-in-law. However, grandchildren and grandparents are still not distinguished from their corresponding in-law categories. The 2000 sample also identifies siblings and siblings-in-law, which were grouped with other relatives in 1982 and 1990.

Comparability — Colombia [top]

The formal definition of household varied greatly over time, from census family (1964) to private household (1973, 1993, 2005) to dwelling (1985). Yet in practice, all the censuses share a common notion of persons eating, sleeping, or living together. Regardless of the definition of household used, relationship categories were those commonly recognized, usually consisting of a single word that was not explicitly defined on the questionnaire. The number of categories does not vary greatly, although over time several were added to further distinguish persons related to the household head by blood or marriage. The category "lodger" exists in 1964 and 1973, but it was dropped thereafter. In 1964 all relatives except spouses and children were grouped together as "other relatives." The 1973 census added two new categories for relatives that persisted in subsequent censuses: "parent/parent-in-law" and "grandchild." The 1985 census added a category for "child-in-law." The 2005 sample is the only one to identify siblings and siblings-in-law.

Comparability — Costa Rica [top]

The available relationship categories differ considerably across the Costa Rican censuses. Spouses are only differentiated from unmarried partners in 1963. The 1963 and 1973 censuses identify many fewer relationships than 1984, 2000 and 2011, which have categories for parents, grandchildren, siblings-in-law, and domestic help and their family. Parents are combined with parents-in-law only in 1984 and 2000; they are separated into two categories in the 2011 sample.

In 1963, outside of the handful of specified relationships, all other persons are grouped into one residual category that includes both relatives and non-relatives. This greatly impairs analyses of household structure, but since 1963 is a sample of individuals rather than of households, most such studies are impossible in any case.

Comparability — Cuba [top]

In 2002 and 2012, parents are combined with parents-in-law.

Comparability — Côte d'Ivoire [top]

The relationship information is consistent across censuses. The 1998 census has added resolution for other relative status, including brother or sister and nephew or niece.

Comparability — Dominican Republic [top]

The 1970 sample separately identifies unmarried partners of the household head; all other samples report a combined spouse/partner category. Children-in-law are not identified in the 1960 census. The 1970 census separately identifies spouses of children and partners of children, while the 1981 census includes a combined spouse/partner of child category. Subsequent censuses include only a child-in-law category. Parents and parents-in-law are first identified in a combined category in 1981 and are separately identified in subsequent years. Grandchildren are identified beginning in 1981. A combined category of stepchildren and foster children is included in the 2002 and 2010 samples. Siblings and grandparents are identified only in 2010. Servants identified in all years except 1960.

Comparability — Ecuador [top]

1962 identifies many fewer types of relatives than the later Ecuadorian censuses, which all identify grandparents, parents, and children-in-law. In-laws sometimes cannot be distinguished from blood relatives in those samples. Only the 2001 sample specifies that stepchildren should be included with "other relatives." The 2010 sample specifies that stepchildren should be included with biological and adopted children.

The "other relative" category in 2001 is large relative to earlier years and probably includes some persons that would have been categorized among the specified relationships (including stepchildren).

Comparability — Egypt [top]

Relationship codes are generally consistent over time. Parents are combined with grandparents in 1996. Persons coded as visitors could be relatives of the head.

Comparability — El Salvador [top]

The 1992 sample identifies fewer relationships than the 2007 sample. Parents are combined with parents-in-law and siblings and nephews/nieces are combined with "other relatives" in 1992; these relationships are separately identified in 2007. Children-in-law are distinguished from children in both censuses, while great grandchildren are explicitly included with grandchildren.

Comparability — Ethiopia [top]

Stepchildren are identified in 1984 and 1994. Those samples also identify stepchildren of the spouse (children of the head) in the source variables.

Comparability — Fiji [top]

The 1966 sample combines all persons other than heads into a single category. Codes in subsequent years are generally consistent, although the 1976, 2007, and 2014 samples provide additional detail. There are categories for children-in-law, grandchildren and combined categories of parent/parent-parent-in-law from 1976 to 2014. The 1976, 2007, and 2014 samples distinguish between adopted and biological children; stepchildren appear to be included with adopted children. In 1986 and 1996, it is unclear whether the code for child includes only biological children or also adopted and stepchildren.

Comparability — Finland [top]

The Finland 2010 sample includes information about the relationship to the household head, but the data are composed of persons not organized into households. The variable identifies persons in a cohabiting family, which are classified as "unknown" relationship.

Comparability — France [top]

The definition of household and the relationships identified in the French censuses do not vary greatly over time. Grandchildren are combined with children in 1962-1975.

The general categories are fully comparable with an important exception concerning "child." The child category in 1962-1975 differs from all other samples in the database because it also includes grandchildren. The problem is less severe than it might seem because less than 1 percent of children in 1982 and 1990 were grandchildren.

Comparability — Germany [top]

Relationships are reported only for the 1970 sample, a sample that consists of persons not grouped into households. Children are combined with children-in-law and parents are combined with parents-in-law.

Comparability — Ghana [top]

Only the 2010 sample distinguishes between biological, step, and adopted/foster children.

In 1984 and 2000, the head-of-household category includes persons who were identified as "temporary head" in the census. Temporary heads can be identified in the unharmonized source variable.

Comparability — Greece [top]

The 2001 sample identifies many more categories of relatives and it distinguishes between spouses and unmarried partners. Unmarried partners are only identified in the 2001 and 2011 samples.

Only the 2001 sample specified that "child" includes adopted and step children but not foster children. The 2011 documentation refers to biological, step-children, and adopted children.

In 2011, relationship-to-head is inferred from the type of family nucleus and the relationships within nuclear families. For this sample, the specific relationship-to-head is unknown for persons outside the first nuclear family in the household.

Comparability — Guatemala [top]

In 1964, there are distinct categories for spouse and (unmarried) partner, while these are combined for the 1973 to 2002 samples. The 1981 and 2002 samples provide additional detail for types of children. Parents/parents-in-law and are only separately distinguished in 2002, and Siblings are only distinguished from other relatives in 2002. Children-in-law are only distinguished in 1981, 1994, and 2002.

Homeless persons are combined with persons in collective households in 1973, while they are separately identified in 2002.

Comparability — Guinea [top]

The 1983 sample identifies few relationship categories. The 1996 instructions specified that all relationships by adoption or marriage were to be included with "other relatives." The 2014 sample provides more detailed categories compared to 1983 and 1996 samples, particularly for "other relatives."

Comparability — Haiti [top]

Relationship availability changes each year.

The 1971 provides the least detail identifying only grandchildren among the head's extended family. Further, employees and friends are the only non-relatives identified. Consequently, it is unclear how boarders and other non-relatives are coded.

The 1982 sample combines other relatives and unrelated friends into a single category, while nonrelatives and relatives are separately identified in 2003. No information on parents is available in 1971; while the 1982 and 2003 censuses identify biological parents. Children are separately identified from stepchildren/children-in-law in 1982 and 2003. Grandchildren are identified in every year. In 1971 and 2003, the enumeration form the category for spouse explicitly included unmarried partners [concubine]; in 1982, only spouses are mentioned on the form, and some cohabiting partners may be included with other relatives/non-relatives.

Comparability — Honduras [top]

The Honduras 1961 sample includes information about the relationship to the household head, but the data are composed of persons not organized into households.

Step-children are explicitly included with other children in 1974 and 2001. Children-in-law are separately identified only in 1988 and 2001.

The 2001 sample is the most detailed for Honduras, as it includes relationships for siblings and siblings-in-law, and nephew or niece.

Comparability — Hungary [top]

The relationship data for Hungary are complex. The Hungarian samples 1970-2001 provided only relationship to family head, not to head of household. We inferred relationship to household head using different means in each sample. Users may wish to examine the unharmonized source variables to fully explore our decision-making and other data possibilities with these samples.

In 1970, all persons outside the primary family have an unknown relationship to the household head.

In 1980 we know the relationship of subfamilies to the primary family in two-family households. Relationships are not known for persons in three- or four-family households, which are relatively rare. Within related second families, parents/parents-in-law, children/children-in-law, and grandchildren of the household head can be identified by using individuals' ages and relationships within subfamilies.

The 1990 data are similar to 1980, but they offer slightly richer information about the relationship of subfamilies to the head of household.

In 2001, the relationship-to-head is unknown for persons outside the first family. Moreover, other relatives and other non-relatives cannot be distinguished from each other and are combined at the most general digit of RELATE.

In 2011, the relationship-to-head was available only for the spouse or cohabiting partner; children and ascendants were identified using the relationship-to-family-head but only for those within the family of the household head.

Unmarried partners of the head are identified in all samples except 1980.

In 1980, 1990, and 2011, "ascending line relatives" are coded as parents/grandparents, but the category presumably also contains smaller numbers of other relatives.

Comparability — Indonesia [top]

The 2000 census offers the least detail, identifying only head, spouse, children; all other relatives and non-relatives are combined into a single category. The other samples all identify grandchildren, children-in-law, and either parents and parents-in-law or a combined category of parents/parents-in-law. In 1976, 1995, and 2010, biological children are distinguished from non-own children (step, adopted, and foster children).

In the 1976 census, virtually all mothers-in-law were identified as spouses in the original data. We have corrected the relationship codes for these individuals using information on mother's line number. They are identified by a separate code in the source variable.

Comparability — Iran [top]

The Iran 2006 and 2011 samples identify detailed relationships and distinguish siblings and parents-in-law from siblings and parents.

Comparability — Ireland [top]

The level of detail for relationship varies considerably over time in Ireland, with 1991 offering the most detail. The 1971-1986 samples offer extremely limited information on household relationships.

Comparability — Israel [top]

Relationship is consistent across censuses, except for the combination of parents with parents-in-law in 1983.

The instructions for 1972 and 1983 both discussed how to record a household with two wives.

Comparability — Italy [top]

Grandparents are only identified in the 2011 sample.

Comparability — Jamaica [top]

Relationships are generally comparable over time, with grandchildren, and children-in-law identified separately in all samples, while parents-in-law are combined with parents. The 2001 sample adds additional details, distinguishing common law partners from spouses, and identifying stepchildren and siblings. The unharmonized variable for 2001 distinguishes between the children of the head and spouse/partner, the children of the head only, and the children of the spouse/partner only. The category "domestic employee" in 1982 also includes relatives of an employee.

Comparability — Kenya [top]

The household, defined as a person or group of persons who live and eat together, is the fundamental enumeration unit in Kenyan censuses; members of a household might reside in more than one dwelling.

The 1979 sample provides only limited relationship information. The 1969, and 1989-2009 samples are largely consistent, and provide information on grandchildren and parents. The 1999 and 2009 censuses identify siblings of the head, while nephews and nieces are identified in 2009 only. The 2009 census included an "in-law" category; however the enumeration instruction explicitly stated that children-in-law and parents-in-law should be reported as other relatives. We have combined the in-laws with other relatives in the harmonized variable, but in-laws can still be identified in the unharmonized source variable.

Comparability — Kyrgyzstan [top]

Relationship categories are comparable over time. Grandchildren, children-in-law, parents, and parent-in-law are separately identified in both censuses.

Comparability — Laos [top]

Adopted children and step-children are combined with other children under "child". Persons in collective households are classified as "group quarters".

Comparability — Lesotho [top]

In 2006, the source variable differentiates between a present and absent head of household, which are both classified in RELATE as "Head."

Comparability — Liberia [top]

The 2008 sample included a code of "other spouses" for the second and higher order spouses of the household head. These wives are coded as spouses of the household head in RELATE, but can be identified in the unharmonized source variable. Grandchildren are identified only in the 2008 census. The enumeration instructions in 2008 indicate that the category child should only include the biological children of the household head. Other children should be coded as other relatives or non-relatives.

The 1974 census included a category for "mates"; all persons receiving this code were female and nearly all were married. Enumerators were instructed to use codes like "lodger's wife" for the wives of non-relatives and it is possible that this category was used in these cases. Mates are included in the other non-relative category of RELATE and can be identified in the source variable.

Comparability — Malawi [top]

Relationship categories are largely comparable over time, identifying the head, spouse, children, other relatives and non-relatives. The 1987 and 1998 samples also identify institutional members.

In 1998 and 2008, only biological children are identified separately. Adopted and stepchildren are identified instead as other relatives.

Comparability — Malaysia [top]

The 1970 sample combines parents and grandparents and does not identify grandchildren or any in-laws. The later samples are consistent in the categorizations.

Comparability — Mali [top]

Relationship categories were fully consistent between 1987 and 1998. Grandchildren and in-laws are grouped with other relatives. Relationship detail increased in 2009 when grandchildren, children-in-law, aunts and uncles, and nephews and nieces were separately identified. Parents and siblings are identified in all years.

Comparability — Mauritius [top]

The samples are consistent. In 1990-2011, adopted and step-children are grouped with biological children. Parents are interpreted as not including in-laws.

Comparability — Mexico [top]

The definition of household, its composition, and the major categories of relationship are comparable across census years, but the level of detail varies considerably. The 1960, 1970 and 1995 samples have very little category detail; the 1990, 2000, 2005, 2010, 2015, and 2020 samples have extremely detailed relationship information. The 2010, 2015, and relationship information is drawn from two separate variables: one variable identifies major relationship categories; the second provides detailed relationship for other relatives and non-relatives as recorded. The original variables are available in the unharmonized variables.

In all years, households consist of persons who habitually reside together, whether present or absent at the time of enumeration. Basic relationship terms, such as head, spouse, and child, are those commonly understood and often remain undefined on the census forms. Beginning in 1970 the basic terms were supplied on the census form, to be filled in as appropriate. If a respondent named any other relationship, the enumerator wrote in the term, which was coded at the data processing stage.

Comparability — Mongolia [top]

The samples are entirely consistent.

Comparability — Morocco [top]

Relationship detail increases over time. All censuses identify parents and children-in-law. Grandchildren and siblings are included with other relatives in 1982, but are separately identified in the 1994 and 2004 samples.

In 2004, virtually all spouses of the household head who were older than the head were identified as parents in the source microdata. Using information on age, marital status, and position in the household, we have corrected the relationship codes for these individuals. Researchers can identify the persons whose relationship is imputed as a spouse in the source variable.

Comparability — Mozambique [top]

The samples are largely consistent over time. The 2007 sample has a separate category for stepchild.

Comparability — Nepal [top]

The 2001 and 2011 samples provide relatively detailed relationship codes, including siblings and grandchildren.

In 2001, daughters-in-law have their own category, while sons-in-law are not identified. Two categories in the source variable are used to identify sisters-in-law (most are the brother's wife), compared to just one category for brothers-in-law. For consistency, both categories of sister-in-law have been combined with brothers-in-law into the general "sibling-in-law" category. The different types of sister-in-law can still be distinguished using the unharmonized source variable.

The categories for sibling-in-law, grandparents, nephew/niece, and cousin are only separately identified in 2001. For both 2001 and 2011 "non-relatives" are combined with unspecified "other relatives."

Comparability — Netherlands [top]

Heads and spouses are not distinguished in the Netherlands samples, and they are coded together as heads. The data are samples of individuals -- not of entire households -- which mitigates the impact of this situation.

The samples have extremely limited information: essentially only identifying heads/spouses, children, and other persons. The residual category therefore includes both relatives and non-relatives. The 1960 and 1971 samples also identify institutional residents.

Comparability — Nicaragua [top]

The 2005 census provides the most detail, distinguishing biological children from step-children, parents from parents-in-law, and great grandchildren from grandchildren, as well as identifying siblings and grandparents. Children-in-law are identified only in 1995 and 2005. The 1995 and 1971 censuses identify grandchildren and a combined category of parents/parents-in-law.

Comparability — Pakistan [top]

Relationship detail declines considerably over time. The 1973 sample identifies grandchildren, children-in-law, and a combined category of parents/parents-in-law. The 1981 and 1998 samples identify only the five basic categories: head, spouse, child, other relatives, and non-relatives.

Comparability — Palestine [top]

Relationship categories are comparable over time, and include information on parents, siblings, children-in-law, and grandchildren.

Comparability — Panama [top]

The level of category detail increases over time. The 1960 sample groups all non-children into "other relatives." 1970 identifies grandchildren and parents, combining the latter with parents-in-law. Later samples distinguish parents-in-law and children-in-law. The 2010 sample provides information on most categories of relatives.

Comparability — Papua New Guinea [top]

There are slight differences in the treatment of non-biological children across samples. The 1980 and 1990 samples identify adopted children, while these are combined with step-children in 2000. Children-in-law are only reported in 1980.

Comparability — Paraguay [top]

The relationship categories are largely comparable over time. The 1962 and 1972 samples contain very few categories: only head, spouse or partner, child, and other. The 1962 sample has a category for "Boarders." The 1982-2002 samples contain more detailed relationships such as child-in-law and parent-in-law.

Comparability — Peru [top]

Relationship categories are consistent over time. Grandchildren and children-in-law are identified separately, but parents are combined with parents-in-law.

Comparability — Philippines [top]

The data are fully comparable across samples apart from some extra detail among non-relatives in 1995-2010.

Comparability — Poland [top]

Relationship-to-head is inferred from that to the family head in 1978. This sample allows identifying specific relationships only for the household head, spouse/partner, and child in the first subfamily; persons in other subfamilies are classified as "other relative or non-relative."

The 2002 sample has detailed relationships for all household members.

Comparability — Portugal [top]

The 1981 census identifies fewer relationship categories than the later censuses. It lacks categories for siblings, grandparents and stepchildren, which are present in the later samples. The 2010 sample includes separate categories for "Spouse" and "Partner in a consensual union."

Comparability — Puerto Rico [top]

Relationship codes are generally comparable across all years. The 2005 sample does not include group quarters. Category detail peaks in 1980-2010. The 1970, 2005, 2010, 2015, and 2020 samples do not identify grandparents, aunts and uncles, nephews and nieces, or cousins. The 2005 questionnaire included a single category combining all types of in-laws: child, sibling, and parent. Because of the ambiguity in relationship to household head, the RELATE variable groups in-laws with other unspecified relatives. In-laws can be identified using the unharmonized source variable.

Comparability — Romania [top]

The 2002 sample distinguishes between spouses and unmarried partners and the 1977 sample identifies nephews and nieces; otherwise the categories between the 1977-2002 samples are consistent. In each of these samples children include foster children as well as step, adopted, and biological.

The 1977 sample contains some extra category detail in the unharmonized source variable.

In 2011, only the head of household, spouse/partner, and child are separately identified. Other household members are grouped into "Other relative, not elsewhere classified" or "Non-relative". Foster children are considered as non-relatives in 2011.

Comparability — Russia [top]

Spouse includes unmarried partner in 2010. Children comprise biological, adopted, and step children. Step parents, siblings, grandparents, and grandchildren are included in the corresponding biological relationships in 2010.

Comparability — Rwanda [top]

The 1991 sample lacks a general category for "other relative." It is unclear if unspecified other relatives were grouped with non-relatives, which also includes persons in collective dwellings.

The 2002 sample separates the residual relatives, non-relatives, and group quarters residents into separate categories. Persons who were visitors at the time of the census did not answer the relationship question in 2002. We have categorized them as "visitors" in RELATE, but some may have been related to the household head.

The 2012 sample separately identifies residual relatives and non-relatives, but it does not distinguish group quarters.

Comparability — Saint Lucia [top]

Relationship categories increase in detail between the 1980 and 1991 samples. The 1980 sample combines all relatives except spouses and children, while the 1991 sample separately identifies grandchildren, children-in-law, and parents/parents-in-law. Boarders, domestic employees, and their relatives are identified in 1980 only.

Comparability — Senegal [top]

The relationship codes are not comparable across censuses. In 1988, enumerators were instructed identify subfamilies (for instance married children). Relationship to household head was collected only for primary family members and subfamily heads. Relationship to family head was collected for all other subfamily members. Unfortunately, the detailed information on subfamily relationships is unavailable in the IPUMS data for most persons in subfamilies including subfamily heads (about one-quarter of the full sample). Instead, subfamily members receive the code "other relative" or "other relative or non-relative". As a result, a large number of adult children, children-in-law, grandchildren, parents, siblings, and nephew/nieces cannot be identified in 1988. Non-relatives are separately identified only if they reside in the primary family or are visitors to the household. Persons in subfamilies are identified in the variable SUBFNUM.

The 1988 sample had a significant number of persons originally identified as a second household head; they are recoded to "unknown" relationship.

The 2002 census collected relationship to household head for all persons and roughly one-fifth of the sample is coded as an "other relative" or "non-relative".

The 2013 census offers more detailed relationships, even for persons not related to the head of household.

Relationship category detail changed slightly over time. Grandchildren and grandparents are identified in the 2002 sample, while nephews/nieces are identified in 1988 and only when they reside in primary family. The 2013 sample identifies nephew or niece, aunt or uncle, and grandchildren. Siblings and parents are identified in 1988-2013.

Comparability — Sierra Leone [top]

Both the 2004 and 2015 Sierra Leone samples contain detailed information on relatives, including grandchildren, parents, siblings, and nephews/nieces. In both samples, "in-laws" were collected into a single-category and are combined with other relatives in RELATE.

In 2004, grandparents, aunts, and uncles are not distinguished, these are likely included in the "other" category; whereas those relationships are specified explicitly in 2015. Similarly, the 2015 sample distinguishes biological children and stepc-hildren, whereas the 2004 sample groups these into a single "child" category.

Even though both the 2004 and 2015 samples include a category for "grandchild," grandchildren make up a much higher proportion of the 2004 sample (7.9% vs. 0.9% in 2015), suggesting that some grandchildren may have been included in a different category in 2015 (most likely in "other").

Comparability — Slovakia [top]

The samples are fairly comparable with small detail variations across them. The 1991 and 2001 category for "parent or parent-in-law" includes grandparents; while the 2011 sample separates parents and parents-in-law. The 1991 and 2001 samples have a separate category for tenant, while these are included with other non-relatives in a single category in 2011.

Comparability — South Africa [top]

The 1996 sample identifies fewer relationships than later years. In-laws are included among "other relatives" in 1996, but several types are given distinct codes in 2001, 2007, 2011, and 2016. Children are combined with step and adopted children in a single category in 1996, but non-biological children are independently identified in 2001, 2007, 2011, and 2016. In the 2011 and 2016 samples, there is also an additional category for children-in-law.

Great grandchildren are explicitly included with grandchildren in 2001-2016.

Comparability — South Sudan [top]

The 2008 South Sudan census identifies grandchildren, nephews/nieces and combined categories of parents/parents-in-law and siblings/siblings-in-law.

Comparability — Spain [top]

The 2001 sample distinguishes between spouses and unmarried partners and between siblings and siblings-in-law. Grandchildren are identified in 1981 and 1991, but are included with "other relatives" in 2001.

Comparability — Sudan [top]

Sudan 2008 contains identifies grandchildren, nephews/nieces and combined categories of parents/parents-in-law and siblings/siblings-in-law.

Comparability — Suriname [top]

The Suriname 2004 sample includes information about the relationship to the household head, but the data are composed of persons not organized into households.

The 2004 sample combines step and foster children; in 2012, step-children are grouped with biological and adopted children, while foster children are separately identified. Grandparents include grandparents-in-law.

Comparability — Switzerland [top]

Relationship categories are consistent over time. Children-in-law are combined with children, parents are combined with parents-in-law, and grandchildren are combined with other relatives in all years.

The 2011 sample, which is a sample of individuals not organized into households, includes information on family relationships in the unharmonized source variable CH2011A_0400, but this variable was not integrated into RELATE because it does not identify the head of household.

Comparability — Tanzania [top]

Relationship category detail increased between the 1988 and 2002 samples. Parents and grandchildren are distinguished from other relatives only in 2002 and 2012, and in-laws are never identified.

Comparability — Thailand [top]

The samples differ in their treatment of children, parents, parents-in-law and grandchildren, and non-relatives. Adopted, step, and biological children are separately identified in 1980 and 1990. Step and adopted children are combined in 1970 and coded as stepchildren in RELATE (stepchildren are much more numerous than adopted in other Thai samples). The 2000 sample combines all categories of children. In 1970, parents are separately identified, while parents-in-law are included with other relatives. The 1980 and 1990 samples combine parents with parents-in-law, while the 2000 sample combines parents, parents-in-law, and grandparents. Grandchildren are separately identified in 1970 and combined with grandchildren-in-law in later samples. Detail on non-relatives declines slightly over time.

Comparability — Togo [top]

The 2010 sample provides separate categories for parent and parent-in-law; while in 1960 and 1970, those enumerated as parent are assumed to be biological relationships. The availability of other specific relative and non-relative categories differs considerably across the 1960, 1970, and 2010 samples.

Subfamilies are only identified in the 1970 sample. In that sample, the relate source variable indicates relationship to the household head only for members of the primary family, persons not in a subfamily, and subfamily heads. For subfamily members other than the subfamily head, the relate source variable indicates relationship to the subfamily head. Where possible, the relationship to the household head of subfamily members other than the subfamily head was inferred on the basis of the subfamily head's relationship to the household head (e.g., a child of a child of the household head is the grandchild of the household head).

Comparability — Trinidad and Tobago [top]

The Trinidad and Tobago samples have varying level of detail for relationships with the household head. In 1970, only head, spouse/partner, and child are identified. The 1980, 1990, and 2000 samples also consider grandchildren and spouse/partner of a child.

The 2011 sample is the most detailed for Trinidad and Tobago, as it includes additional relationships: spouse and partner of the household head (separately), biological, adopted, and step-children, and parents or parents-in-law.

Comparability — Turkey [top]

The 1985 and 1990 census included a joint category of children-in-law and grandchildren. We have used marital status to distinguish between these two categories of relationship, because children-in-law are married by definition and the vast majority of grandchildren have never married. The original codes are preserved in the unharmonized source variables. In 2000, these categories are separately identified.

Unlike the earlier samples, the 2000 census combines other relatives and non-relatives into a single category. With this exception, relationship codes are more detailed in 2000 than in earlier censuses: parents, parents-in-law and siblings are separately identified. The 1985 and 1990 censuses report combined categories for parents/parent-in-law and include siblings with other relatives.

Comparability — Uganda [top]

The 2002 and 2014 samples identify step-children. Parents-in-law and siblings-in-law appear to be included with "other relatives" in 1991, but are combined with parents and siblings in 2002 and 2014. Grandchildren are identified in 2014 but not in 1991 or 2002. In the 2014 sample, the source variable distinguishes between "usual head of household present" and "usual head of household absent." Slightly more than one in ten households has an absent head. The reason for the relatively high frequency of absent heads appears to be that polygamous households were split such that there would only be one spouse in each household, and the head was likely recorded as absent in all but one of these component households. In addition to being available in the RELATE source variable, information on the head's presence or absence is incorporated into the RESIDENT variable.

Comparability — United Kingdom [top]

Both samples offer reasonably detailed relationship codes, but the categories differ. The 1971 and 1991 sample identifies in-laws. In-laws were not mentioned in the 2001 census questionnaire and it is assumed that they are included in "other relatives" in that sample.

The 2001 sample identifies step-children, step-parents, and grandparents; 1971 and 1991 identifies nephews and nieces.

Comparability — United States [top]

The general code is comparable across all years. Group quarters (collective dwellings) is a primary distinction in the relationship variable for the U.S. samples. The 2005 sample does not include group quarters.

Category detail peaks in 1980-2000. The 1960-1970 and 2005-2020 samples do not identify grandparents, aunts and uncles, nephews and nieces, or cousins.

Some of the following categories posed unique problems due to changing definitions and categories across census years, and they are described in detail.

Comparability — Uruguay [top]

Category detail increases over time. Children-in-law are separately identified in every sample except 1963. Grandchildren are combined with other relatives in 1963-1985. Parents and parents-in-law are combined in 1963-1996, and separately identified in 2006 and 2011. The 2006-2011 samples also identify siblings and siblings-in-law.

Biological children and stepchildren are combined in 1975 and 1985, and separately identified in later samples. The 2006 and 2011 censuses distinguish between two types of biological children: those related to the head and spouse, and those related only to the head. This distinction is preserved in the unharmonized source variables.

Comparability — Venezuela [top]

Category detail increases over time. 1971 identifies many fewer types of relatives than the other Venezuelan censuses, grouping all relatives other than children into one category. The 1981 sample offers detail within other relatives, including grandchildren, parents, children-in-law and siblings. The 1991-2001 samples add nephews and nieces.

Comparability — Vietnam [top]

"Non-relatives" are combined with "other relatives" in 1999 and 2009 Vietnam. This makes these samples different from the 1989 Vietnamese sample and nearly all other samples.

Comparability — Zambia [top]

Category detail increases over time. 1990 identifies many fewer types of relatives than the other Zambian censuses, grouping all relatives other than children into one category. The 2000 sample offers detail within other relatives, including grandchildren, parents, children-in-law and siblings. The 2010 sample add parents-in-law and cousins. Stepchildren are distinguished from biological children in all years.