Codes and Frequencies
Description
PARRULE describes the criteria by which the IPUMS International variables MOMLOC and POPLOC linked the person to a probable mother and/or father.
IPUMS International establishes child-parent links according to five basic rules, and PARRULE gives the number of the rule that applied to the link in question. A link to any parent automatically generates a second link to that parent's spouse or partner, so only one rule is needed to describe both MOMLOC and POPLOC.
The design of the interrelationship variables is described in this paper on IPUMSI family linking methodology.
Comparability — Index
Comparability — General
PARRULE is relatively comparable across all samples, but the distribution of codes and the quality of the links varies due to the nature of the underlying data. In some instances customization of the linking rules was necessary because of category differences in RELATE (relationship to the head of household) and other enumeration practices that affected the data.
The various factors that can affect the construction of the pointers in the different samples are summarized in the Sample Pointer Details Table. It provides a means of identifying samples with comparable input information.
No constructed family interrelationship variables are available for the individual-level samples that do not contain entire households of persons (see the samples page).
PARRULE works the same general way for all samples. Starting from the standpoint of the "child," the program searches the household for a possible parent who meets minimum age and relationship requirements. The exact criteria vary by "rule" -- roughly defined as allowable relationship pairings. Within each rule the strongest links are determined first, based on person order and the age, marital status, and childbearing information of the potential parent. At a given link strength, every eligible "child" in the household looks first for a mother and, if that fails, for a father. Then the next weaker test is executed, with each person again seeking a mother first and then a father. Thus mother links are consistently advantaged, but only where potential mothers and fathers have similarly strong information for making a link. Once a child links to a parent, they discontinue searching. If that parent is linked to another person via the location-of-spouse pointer variable, SPLOC, the child is also immediately linked to that person.
Possible pairings of children and parents depend on each person's relationship to the household head. A small number of samples have relationship information for subfamilies within households. For those samples, MOMLOC and POPLOC use the relationship to the head of subfamily (SUBFREL), instead of relationship to head of household. Persons in different subfamilies can link to one another only under Rule 1 (head and spouse links), and only where RELATE identifies the cross-subfamily relationships. Subfamily information is available for only a subset of relationships in Fiji 2007 and is not used to construct pointers in that sample.
The implementation of the links also depends on the relative positions of the persons within the household. Persons who are adjacent in the roster of the household are considered stronger candidates for a link than non-adjacent persons. A potential mother is considered adjacent to a child so long as the woman precedes the child in the household and no persons separate the potential mother from her child, except for her spouse or partner and any children already linked to that mother. Links to fathers follow the same definition of adjacency. In most samples the person order has meaning, but the strength of the inferences from adjacency varies across data sets. Every child looks first for an adjacent mother or father. If no adjacent parent is found, the search continues from the top of the household.
We emphasize consistency across samples in the algorithms that determine MOMLOC, POPLOC, and PARRULE. We use the same age constraints and relationship pairings to the extent possible. In some samples, however, we were compelled to customize the links beyond what could be implemented more generally. For example, we allowed unspecified other relatives to link to children in samples that do not identify grandchildren, and allowed heads to link to unspecified other relatives in samples that do not identify parents of the head.
A primary concern guiding the development of the pointers was to prevent all children in complex households from linking to a single parent when there were multiple legitimate candidates. Whenever possible, we relied heavily on reported children ever born (CHBORN) and children surviving (CHSURV) to determine how many children should link to a particular woman and to her spouse or partner. We refer to this as the "child cap" for a parent or couple. In most contexts, the linking algorithm allows the cap to be exceeded, but typically only after other potential parents have received their share of eligible children. Thus, it plays a powerful role in the allocation of children.
Unfortunately, some censuses do not collect childbearing data, while others collect information only for married or reproductive-age women. Virtually no countries collect data for men. In instances when we could not use empirical childbearing data to "cap" links to a potential parent, we estimate the share of available children a man or woman should claim based on their marital status, age differences from potential children, and the known childbearing data of other household members.
The calculated child cap distributes the available children not already accounted for by empirical childbearing information for women in the household. Male and female caps are calculated independently -- except when a married man receives the empirical cap of his wife. Ever-married potential parents divide up all the available possible "children." The sum of the married parent caps typically accounts for all the possible children in a household. The cap calculation for never-married potential parents is much reduced if any married parents are present, but it is non-zero. Depending on the order of links and age differences among possible children and parents, never-married parents can be linked in households containing married parents, but it is relatively unlikely by design. Careful analysis of the data convinced us that many more good links than bad were created by implementing these child caps, but they are imperfect.
MOMLOC and POPLOC identify social relationships (such as stepmother and adoptive father) as well as biological relationships. For this reason, it can be appropriate to sometimes exceed the number of children born to a woman or to allow improbable age differences. Researchers wishing to exclude these social relationships from their analysis can identify them using the variables STEPMOM and STEPPOP. Links exceeding children ever born can be identified in rules 2 and higher. Users should note STEPMOM and STEPPOP will underestimate the number of stepparents in the population.
A small number of samples contain parental pointer information collected as part of the census or constructed by the national statistical office. The samples are not a highly representative group, and the pointers typically restricted themselves to biological parents only, but we did use these data to help calibrate the IPUMS pointer variable algorithms. These "empirical pointer" data are available among the unharmonized variables, and are mentioned in the country-specific comparability sections below. Empirical pointers are available for: Brazil 1991, Belarus 1999-2009, Cuba 2002-2012, Fiji 1976-2007, Guatemala 1981, Iran 2011, Kenya 2009, Romania 1991 and 2002, South Africa 2001-2011, Spain 1991 and 2002, Spain labor force surveys, Thailand 1980 and 1990, and Togo 1960. Spain 2011 is a special case: we use the empirical pointers in place of the IPUMS-constructed pointers, because the sample lacks information on relationship to head, preventing us from following our usual procedures.
In general, complex households, polygamy, and lack of childbearing information were the biggest hurdles to accurately identifying parents. The more these factors pertain to a sample or a particular household -- and the fewer relationship categories identified in a sample -- the less certain the pointer links are likely to be. One of the bigger practical effects can be ambiguity in the linking of grandchildren to children and the linking of children to spouses in polygamous households. Typically, all children will be linked in those situations, but not necessarily to the correct parent.
The parental links in samples without childbearing variables are more uncertain, and this affects the general comparability of both MOMLOC and POPLOC across samples with differing information. For some purposes, even the different universes for the childbearing variables can have an impact. The importance of varying information, however, depends considerably on the complexity of household structures across countries. Links within primary families in most countries are quite strong.
The codes for PARRULE are described below. PARRULE is a two-digit code: the first digit identifies the basic rule under which a link occurred; the second digit identifies issues that affect the accuracy and the comparability of PARRULE across samples. A lower-numbered rule (greater than 0) generally indicates a more certain link than a higher numbered rule, but this is not always true. We limit the links made under Rules 4 and 5 -- involving unspecified relatives and non-relatives -- to children under age 16, because the information for linking can be particularly tenuous.
Rule 0 = No parent. No parent of this person is present in the household
Rule 1 = Head and spouse. This rule covers four basic cases, and occurs when the person linked is 10 to 69 years younger than the probable parent. The allowable relationship pairings are as follows:
Head, Sibling, or Sibling/sibling-in-law -- linked to -- Parent, Parent/parent-in-law, or Parent/grandparent
Spouse/partner, Sibling-in-law, or Sibling/sibling-in-law -- linked to -- Parent-in-law
Child/child-in-law -- linked to -- Head or Spouse
Child/child-in-law/grandchild -- linked to -- Head or Spouse
Stepchild/child-in-law -- linked to -- Head or Spouse
Spouse/partner, Sibling-in-law, or Sibling/sibling-in-law -- linked to -- Parent-in-law/Stepparent
In some instances, the linkages falling under this rule were more ambiguous and required special treatment. The second digit identifies links made under these circumstances.
Many samples do not distinguish parents from parents-in-law. In these samples, the head is allowed to link to the first eligible parent/parent-in-law; but the head's spouse cannot link to a parent. Less commonly, parents, grandparents, and other "ascending line relatives" are combined into one category; and are treated as parents for pointer creation. While the pointers can be used to identify the presence of multiple-generation families, researchers may prefer to restrict analyses of intergenerational families to samples with appropriately explicit relationship information.
A large number of samples do not clearly distinguish between children and children-in-law. In these instances, we allow only the first eligible child in a marriage between two children/children-in-law to link to the household head and spouse. A small number of samples do not distinguish between children, children-in-law, and grandchildren. A child/grandchild will link under Rule 1 to the head and spouse only when the age differences falls between 10 to 44 years, while a child/grandchild is allowed to link to another child/grandchild when this same age requirement is met (under Rule 2). Despite the ambiguity in the relationships to household head, we believe these procedures are usually successful at identifying children or grandchildren.
In Haiti 1982 and 2003, the same relationship codes are used to identify both step- and in-law relationships: stepchildren and children-in-law are identified jointly, as are stepparents and parents-in-law. A stepchild/child-in-law is allowed to link to the head and spouse under Rule 1 unless they are married (linked by SPLOC) to a child of the household head. Parents-in-law/stepparents are treated like parents-in-law for the purposes of pointer creation.
When the head of household is in a polygamous union (identified either through marital status or through the presence of multiple spouses), we place additional restrictions on links between children of the head and potential mother. The intent is to prevent improbable links and to minimize the chances that all children will link to one mother where there are multiple candidates. A potential mother must be 10-54 years older than the child, and links occur first to spouses who have not yet exceeded their child cap. If, after allowing links in excess of reported childbearing, still no link has been made, the child links to the head and to his first wife (the wife he points to in SPLOC). Spouse and child parental pointers recognize only polygamous unions involving the household head; other polygamous unions are uncommon and frequently not identifiable.
- Code 11 = Unambiguous relationship pairing.
- Code 12 = Same as rule 11, except that there is some ambiguity in relationships in the household:
Parents are not distinguished from parents-in-law;
Children are not distinguished from grandchildren;
Children are not distinguished from children-in-law, and a child/child-in-law is linked to another via SPLOC (i.e., they are married)
In samples with information on a potential mother's children ever born (CHBORN) or surviving (CHSURV), we exceed the empirical "child cap" only after all other potential matches are made. In samples without childbearing data (and for unmarried men in all samples), we use the constructed child cap to distribute grandchildren among multiple potential parents. In samples with no childbearing information or instances when we exceeded the "cap," our confidence in the links is reduced. The second-digit of rule 2 identifies these situations.
- Code 21 = Link does not exceed the potential parent's empirical child cap (married men get cap from spouse or partner).
- Code 22 = Link occurs to a potential parent without CHBORN information, but does not exceed the parent's constructed child cap.
- Code 23 = Link exceeds empirical or constructed child cap.
The following pairings fall under Rule 3:
Nephew-in-law/niece-in-law -- linked to -- Sibling-in-law or Sibling/Sibling-in-Law;
Grandchild or great-grandchild -- linked to -- Grandchild;
Cousin -- linked to -- Aunt/uncle
- Code 31 = Link does not exceed the potential parent's empirical child cap (married men borrow cap from spouse or partner).
- Code 32 = Link occurs to a parent without childbearing information, but does not exceed that parent's constructed child cap.
- Code 33 = Link exceeds empirical or constructed child cap
Relationship pairings allowed under rule 4 are:
Other relative or Other relative/Non-relative -- link to -- Child (allowed only in samples without grandchild relationship)
Other relative or Other relative/Non-relative -- link to -- Unmarried partner
Other relative -- link to -- Other relative
Other relative/Non-relative -- link to Other relative/Non-relative
Other relative or Other relative/Non-relative -- link to -- Grandchild
Other relative or Other relative/Non-relative -- link to -- Sibling or Sibling-in-law
Other relative or Other relative/Non-relative -- link to -- Unmarried Partner
- Code 41 = Link does not exceed the potential parent's empirical child cap (married men borrow cap from spouse or partner).
- Code 42 = Link occurs to a parent without childbearing information, but does not exceed the parent's constructed child cap.
- Code 51 = Link is within limits of empirical child cap (married men borrow cap from spouse or partner).
- Code 52 = Link occurs to a parent without childbearing information, but does not exceed the constructed child cap.
Comparability — Argentina [top]
No pointers are available for the 2010 sample because marital status is not currently available.
Relationship codes are largely consistent over time. Parents-in-law are combined with parents in every year.
Childbearing information is available in each year, but the age universe and the top-codes vary.
Comparability — Armenia [top]
Relationship codes are detailed, identifying grandchildren and in-laws.
Comparability — Austria [top]
Relationship information became more detailed over time, with the addition of grandchildren and children-in-law categories in 1981. For all years 1971-2001, parents are combined with grandparents in an "ancestor of the head" category. Childbearing information is available in all those years.
No pointers are constructed for 2011, which lacks household relationship information.
Comparability — Bangladesh [top]
Children-in-law and parents are never identified. In 1991 and 2001 they are combined included with a combined other relative/non-relative category; in 2011 they are included with other relatives. Grandchildren are separately identified only in 2011.
Polygamous marriages are inferred for household heads in Bangladesh based on the presence of multiple spouses.
None of the samples provide fertility data.
Comparability — Belarus [top]
Relationship codes are detailed, identifying grandchildren and in-laws. Childbearing data are available.
The census collected the person number of every individual's mother (and in some cases father). See the unharmonized variables.
Comparability — Bolivia [top]
Grandchildren are separately identified in 1976 and 2012, but are included with other relatives in 1992 and 2001. Children-in-law are identified in each sample, while parents are combined with parents-in-law. Childbearing data are available. In 2012, Other relatives (code 4900) are allowed to link to Spouses and Heads under rule 4, and Non-relatives (code 5900) are allowed to link to Spouses under rule 5, because the data do not explicitly identify stepchildren, and a close examination indicated that many of these pairings involved probable stepchildren of the head.
Comparability — Botswana [top]
The Botswana samples include but do not identify persons in polygamous marriages. Polygamous status is inferred for household heads based on the presence of multiple spouses, and this information is used in the construction of pointers.
Fertility data are available in all years.
Comparability — Brazil [top]
Brazil reports information on subfamilies in all years except 2010. No links can be made across subfamilies in 1960 and 1970, but heads can be linked to parents in other subfamilies in 1980, 1991, and 2000.
The level of detail for relationship varies considerably over time in Brazil, with the 1991 and 2010 censuses offering the greatest detail. Among the differences, parents-in-law and children-in-law are explicitly identified only in 1991 and 2010.
The universe of CHBORN changes slightly over time, but it is available in all samples.
The 1991 sample contains a pointer variable indicating the person number of each individual's mother within the subfamily. See the unharmonized variables.
Comparability — Burkina Faso [top]
The 1985 sample Burkina Faso does not lacked information on relationship to household head. Pointers could be created only for the 1996 and 2006 sample.
Polygamous marriages are identified as a category of marital status in each Burkina Faso sample, and heads are allowed to link to multiple wives.
Relationship categories are largely consistent over time with following exceptions: grandchildren are separately identified only in 2006, while parents are separately identified only in 1996. Parents-in-law are never identified.
Fertility data are available in all years, with small differences in the age universe.
Comparability — Cambodia [top]
The intercensal population surveys of 2004 and 2013 have more detailed relationship information than the census samples of 1998, 2008, and 2019. The intercensal surveys identify children-in-law, siblings, siblings-in-law, nephews, and nieces, whereas in the census samples, these categories are all grouped with "Other relative, not elsewhere classified." The intercensal surveys also distinguish parents-in-law from parents.
Childbearing data were collected in all samples.
Comparability — Cameroon [top]
The 1976 and 1987 samples provide very limited information on relationship to head. The 2005 sample provides very detailed codes. Grandchildren, children-in-law, and parent-are separately identified only in 2005, along with other family relationships.
Polygamous marriages are identified as a category of marital status in each Cameroon sample, and heads are allowed to link to multiple wives.
Fertility data are available only in 2005.
Comparability — Canada [top]
Pointers can only be constructed for the 2011 samples; data for other years are not organized into households. The 2011 sample lacks childbearing data.
Comparability — Chile [top]
The 1960 Chilean sample consists of person records that are not grouped into households. Pointers could be created for 1970-2002 only.
Grandchildren are identified in all samples. Parents-in-law are separately identified in 2002, and are combined with parents in all other years. Children-in-law are first identified in 1982.
Childbearing data are available in every sample.
Comparability — China [top]
Relationship codes are consistent over time in China. All censuses separately identify grandchildren, but the pre-2000 samples combine children and parents with in-laws.
Information on childbearing, CHBORN AND CHSURV, was asked only of women age 15 to 64.
Comparability — Colombia [top]
The 1964 Colombian sample consists of persons not grouped into households. Pointers could be created only for 1973-2005.
Relationship codes are largely consistent over time. Parents-in-law can never be distinguished from parents, while children and children-in-law are separately identified in all years except 1973.
Childbearing information is available in all years.
Comparability — Costa Rica [top]
The 1963 Costa Rican sample consists of person records not grouped into households, thus pointers are not constructed for that year.
The 1973 census is considerably less detailed than later samples; grandchildren, parents, and children-in-law are combined with other relatives in 1973. Later samples separately identify these relationships, though some combine parents and parents-in-law.
Childbearing information was asked in all censuses.
Comparability — Cuba [top]
Relationship codes are generally detailed, identifying grandchildren and children-in-law, but combining parents with parents-in-law.
Consensual unions are identified in marital status, but no fertility information is available.
Comparability — Dominican Republic [top]
Pointers could not be constructed for the 1960 and 1970 samples which are not organized into households.
Grandchildren and children-in-law are identified in all years. Parent and parents-in-law are reported in a combined category in 1981 and are categorized separately in subsequent years.
Fertility data are available in all years.
Comparability — Ecuador [top]
The 1962 Ecuadorian sample consists of person records not grouped into households. Pointers could be created only for years 1974-2010.
The 1974 census separately identified parents and parents-in-law; later censuses combined these categories. Grandchildren are separately identified in all years.
Childbearing information is available in all censuses.
Comparability — Egypt [top]
The Egyptian censuses identify men in polygamous marriages. Pointers for the children of the head take into account the presence of multiple wives.
Parents are combined with other relatives in 1986 and with grandparents in 1996.
No childbearing information is available.
Comparability — El Salvador [top]
Relationship codes are detailed in both samples, although parents are combined with parents-in-law in 1992. Fertility questions were asked of females ages 12 and older in each year.
Comparability — Ethiopia [top]
Parents are combined with in-laws in all years and grandchildren are not separately identified prior to 2007.
Fertility data are available in all years.
Comparability — Fiji [top]
The 1966 relationship codes did not provide enough detail to construct pointers. Pointers are available for the years 1976-2014.
From 1976 onwards, relationship codes are generally consistent over time. Children-in-law are identified in each year, as is a combined parent/parent-in-law category. Grandchildren are also identified each year.
Fertility data are available only in 1966, 1976, and 2007.
The 2007 sample collected mother's position in the household. This information is available in this unharmonized source variable. In addition, the census office constructed subfamilies for a subset of household members (see SUBFREL, SUBFNUM).
Comparability — France [top]
Relationship codes in France are distinct from other countries, and changed substantially over time. Grandchildren are combined with children and children-in-law in 1962-1975. In 1982-2011, grandchildren are separately identified from children, but children remained combined with children-in-law. Parents can never be distinguished from parents-in-law.
Women's childbearing histories are not available in any year.
Comparability — Ghana [top]
Polygamous households are included in the 2000 Ghanaian census, but these marriages are not explicitly identified. We inferred polygamy based on the presence of multiple spouses and used this information to construct the pointers. The 2010 sample contained no households with multiple spouses.
Grandchildren and children-in-law are identified in Ghana, but parents are combined with parents-in-law.
Childbearing information is available in all years.
Comparability — Greece [top]
Relationship detail in Greece is greatest in 2001, when unmarried partners and in-laws are first identified. Grandchildren are identified in each year. Parents and parents-in-law are combined in 1971-1991, while children-in-law are never separately identified.
Childbearing information was collected in 1991 and 2001, but is not available in earlier censuses.
The 2001 sample identifies subfamilies but provides no information on the relationships within subfamilies. The pointers link individuals across subfamilies only when the pairing is head or sibling to parent. Within subfamilies, pointers were created based on reported relationship to the household head; there was no information on relationships within subfamilies. The earlier years do not contain subfamily information.
Comparability — Guatemala [top]
Parents are combined with parents-in-law in all samples except 2002. Children-in-law are not distinguished from "Other relatives" in the 1964 and 1973 samples.
Comparability — Guinea [top]
Polygamous marriages are identified explicitly for men in Guinea. Pointers take into account the presence of multiple wives when identifying potential mothers for the children of the household head.
Relationship codes are more detailed in 1996 than in 1983. In 1983, grandchildren and parents are included with other relatives. In 1996, grandchildren are separately identified, while parents are combined with parents-in-law. Children-in-law are never separately identified.
Childbearing information is available only in 1996.
Comparability — Haiti [top]
Relationship detail increases over time. In 1971, parents-in-law and children-in-law are not identified. In 1982 and 2003, they are identified but combined with stepparents and step-children. Parents-in-law/stepparents are treated like parents-in-law during pointer construction. Stepchildren/children-in-law are allowed to link to heads and spouses, like stepchildren, so long as they aren't already married by SPLOC to a child of the household head. Grandchildren, however, are identified in every year.
Childbearing data are available in 1982 and 2003.
Comparability — Hungary [top]
The Hungarian samples all contain subfamily information, but provide varying details on relationships between the primary family (containing the household head) and other subfamilies. In the 1980 and 1990 samples it was possible to link household heads to their parents enumerated in separate subfamilies. In 1970 and 2001, the relationship between members of subfamilies to the primary family is unknown, and links across subfamilies are not possible.
Relationship categories vary over time. The 1970 relationship variable does not identify parents, while the 1980 and 1990 samples identify parents and grandparents in a combined category. In 2001, the household head's parents can be separately identified only if they are enumerated in the primary family. Consequently, the 1970 and 2001 pointers cannot link a parent enumerated in a different subfamily from their child.
Childbearing information is available in all years for women. Information for men was collected in 1990 and 2001, but is not used in the construction of pointer variables.
Comparability — Indonesia [top]
The 2000 relationship codes are limited, identifying only heads, spouses, children, and a combined category of other relatives and non-relatives. Grandchildren and children-in-law are identified in all other years. Parents are combined with parents-in-law in 1980, 1985, 1990, 2005, and 2010, and separately identified in 1971, 1976, and 1995.
Childbearing data are available in every year except 1985. In 1991, fertility questions were asked only of ever-married women ages 10-54. In the remaining samples, fertility questions were asked of all ever-married women.
Comparability — Iran [top]
The 2006 sample identifies men in polygamous marriages; polygamy is inferred by presence of multiple spouses in 2011. The samples provide detailed relationship information. Childbearing data are available only for ever-married women.
Comparability — Iraq [top]
Polygamous marriages are identified explicitly for men in Iraq. Pointers take into account the presence of multiple wives when identifying potential mothers for the children of the household head.
Grandchildren are separately identified, but parents are combined with parents-in-law.
Childbearing information was collected only for ever-married women.
Comparability — Ireland [top]
With the exception of 1996 and 2011, the Irish samples contain subfamily information. In 1971-1986, only subfamily relationships are reported and it is not possible to link across subfamilies. In these samples, the household head will only link to a parent if the parent is enumerated in the same subfamily, and only in 1986 can a household head living with a spouse or child be enumerated in the same family as their own parent.
Beginning in 1991, information on relationship to household head is available but the detail varies considerably across years. In these later samples, it was possible to link household heads to their children and parents enumerated in separate subfamilies.
In 1971, currently married women are asked to report the number of children born alive to the "present marriage". In 2011, fertility data were collected of all women who were present in the household at the time of the census. No childbearing data are available in intervening years.
Comparability — Israel [top]
Relationship information is consistent between censuses, except that parents and parents-in-law are combined in 1983, and separately identified in 1970 and 1995.
The universe for CHBORN was restricted to only ever-married females in 1970 and 1983, but extended to all females in 1995. The 1995 CHBORN variable is interval-coded for women of parities 3 and higher. As a result, the parental pointers in 1995 may underestimate the number of children who should link to higher parity mothers.
Comparability — Italy [top]
Relationship codes are detailed, identifying grandchildren and in-laws. Childbearing information was not collected.
Grandchildren and parents-in-law are not identified in the 2011-2013 labor force surveys.
Comparability — Jamaica [top]
Relationship categories are generally consistent over time. Common-law partnerships are common in Jamaica; unmarried partners are combined with spouses in 1982 and 1991, and separately identified in 2001. Stepchildren and siblings are identified only in 2001.
Although childbearing data are available in each census year, the age universe becomes increasingly restrictive: ages 14+ in 1982, 14-64 in 1991, and 14-49 in 2001.
Comparability — Jordan [top]
Although grandchildren are separately identified, children-in-law and parents-in-law are included with other relatives.
Childbearing information was not collected.
Comparability — Kenya [top]
The 1979 Kenyan sample contains individuals not grouped into households. Pointers are available for the 1969 and 1989-2009 samples.
All Kenyan samples include polygamous households, which can usually be identified by marital status, MARST, or polygamous status, POLYGAM. In 1969, polygamous status was inferred for household heads based on the presence of multiple wives. Polygamous households receive special treatment under pointer rules.
In Kenyan samples, children-in-law, and parents-in-law are combined with other relatives, as are grandchildren in all years except 1969 and 2009.
All censuses collected consistent data on women's childbearing histories.
Comparability — Kyrgyzstan [top]
Relationship codes are detailed and consistent over time, identifying grandchildren and in-laws. Childbearing information is available.
Comparability — Laos [top]
Grandchildren and children-in-law are only identified separately in the 2015 sample.
In all samples, fertility data were only collected for women aged 15-49.
Comparability — Liberia [top]
Pointers are not available in 1974, because the sample is not organized into households.
The 2008 census explicitly identifies polygamous marriages for both men and women. Grandchildren and parents are identified, while in-laws are included with other relatives.
Fertility data are available only for women of reproductive age.
Comparability — Malawi [top]
The Malawi samples include but do not identify persons in polygamous marriages. Polygamous status is inferred for household heads based on the presence of multiple spouses, and this information is used in the construction of pointers.
Relationship categories are generally consistent over time and grandchildren are combined with other relatives in all years. Fertility data are available in all years.
Comparability — Malaysia [top]
The 1970 sample combines grandchildren with other relatives and parents with grandparents, and does not identify any in-laws. The later samples consistently identify these more detailed categories, although parents and parents-in-law are always combined.
Childbearing information is collected only for married women in 1970 and 1980, and not at all in 1991 or 2000.
Comparability — Mali [top]
Relationship detail increases over time. In 1987 and 1998, Grandchildren and in-laws are grouped with other relatives. In 2009, grandchildren and children-in-law are separately identified, while parents-in-law are included with other relatives.
The 1987 and 1998 samples include fertility data for all women ages 12 and older; in 2009, fertility questions were asked only to women of reproductive age.
Comparability — Mexico [top]
The 1960 Mexican sample consists of persons not grouped into households, while the 2005 census did not collect information on marital status. Consequently, pointers could be created only for years 1970-2000 and 2010-2020. Pointers were also created for the 2005-2020 labor force surveys.
The 1970 census differs substantially from later years. Subfamily information was collected in 1970, and links are never allowed across subfamily boundaries.
The detail of relationships codes also varies considerably across years. The 1970 and 1995 samples do not identify parents or children-in-law, combining them instead with other relatives. The 1990, 2000, and 2010-2020 relationship codes are notably more detailed, as are the codes in the Mexican labor force surveys.
Childbearing information was not collected in 1995, but is available in all other years, including labor force surveys.
Comparability — Mongolia [top]
Relationship codes are detailed, identifying grandchildren and in-laws.
Women's childbearing histories were collected in 1989 only.
Comparability — Morocco [top]
Polygamous marriages are inferred for household heads in the Moroccan censuses based on the presence of multiple wives.
Relationship codes are generally quite detailed. Parents-in-law, however, are included with other relatives in all censuses, while grandchildren are combined with other relatives in the 1982 census.
Childbearing are available in 1982 for ever-married women age 50 and under, and for all married women in 1994 and 2004.
Comparability — Mozambique [top]
Neither sample identifies parents-in-law. Polygamous unions are identified by presence of multiple spouses of household head.
Both samples have childbearing information, but the age range for respondents is limited in 1997.
Comparability — Nepal [top]
The Nepali census collected information on polygamous men and women. Because the number of women with multiple husbands is very small, the spouse and parent pointers account only for polygamous men.
The census combined other relatives and non-relatives together into an "other relative or non-relative" category. For the purposes of pointer creation, this category is treated as "other relatives."
Fertility data are available only for ever-married women of reproductive age (15-49).
Comparability — Nicaragua [top]
Relationship codes become more detailed over time. The 1971 and 1995 samples include a combined category of parents/parent-in-law. The 1971 census did not identify children-in-law, including them with other relatives. Fertility data are available in all years with no marital status or age restrictions.
Comparability — Pakistan [top]
The 1981 Pakistan sample consists of person records that are not grouped into households. Pointers could be created for 1973 and 1998 only.
In 1973, 15 percent of households were "headless" (no person was identified as the household head). Many of these are household fragments that include persons identified as spouses and children, who are linked by MOMLOC and POPLOC. It is uncertain how many of these children in fact reside with both parents.
Relationship detail declines considerably over time. The 1998 sample identifies only the five basic categories: head, spouse, child, other relatives, and non-relatives, while the 1973 sample identifies grandchildren, children-in-law, and a combined category of parents/parents-in-law.
Both samples include but do not explicitly identify polygamous households. Polygamous marriages are inferred for household heads in 1973 and 1998 based on the presence of multiple spouses, and this information is used in the construction of parent pointers.
Fertility data are available only in 1973 and only for ever-married women.
Comparability — Palestine [top]
The Palestinian censuses contain polygamous households. Polygamous marriages, however, are not identified, and can be inferred only for household heads with multiple wives.
Relationships to household head are consistent over time and detailed, except that parents-in-law are combined with other relatives in both samples.
Childbearing data were asked only of ever-married women, and only for children born to them during their married life.
Comparability — Panama [top]
Relationship detail increased significantly over time. The 1960 census identifies only heads, spouse, and children. Grandchildren can be identified in years 1970-2010; parents are combined with parents-in-law in 1970, and can be distinguished in later years; children-in-law are first identified in 1980.
Childbearing data are available for all women 1970-2010, and for indigenous women in 1960.
Comparability — Paraguay [top]
The samples differ in the detail they provide on household relationships. The 1962 and 1972 samples do not separately identify parents or grandchildren.
All censuses provide women's childbearing histories.
Comparability — Peru [top]
Relationship categories are consistent over time, identifying grandchildren and children-in-law, but combining parents with parents-in-law.
Childbearing information is available in all years.
Comparability — Philippines [top]
Relationship codes are detailed and consistent over time. Parents-in-law are included with other relatives.
Childbearing information was asked only in 1990 and only of ever-married females age 15 to 49.
Comparability — Poland [top]
The 1988 and 2011 samples lack household relationship information, preventing the construction of pointers. In 1978 and 2002, subfamily relationships are available, but the 1978 sample has very limited information.
Comparability — Portugal [top]
The 1991 census provides more relationship detail than earlier censuses, identifying parents-in-law, stepchildren, grandparents, and siblings for the first time. Grandchildren and children-in-law are identified in each year.
None of the censuses provide women's childbearing histories.
Comparability — Puerto Rico [top]
Relationship codes are generally consistent over time. Grandchildren and parents are separately identified in each year, while in-laws are identified in all years except 2005. The 1990 census is the first to separately identify unmarried partners. In previous years, partners are combined with other non-relatives into a "partner/friend" or "partner/roommate" category. Persons in consensual unions, however, can be identified by marital status in the 1970-1990 samples.
Childbearing information is available for all women from 1970 to 1990. The 2000-2020 samples do not contain such information.
Comparability — Romania [top]
The 2011 sample lacks household relationship detail available in other years. It does not identify parents or grandchildren. All samples contain childbearing information.
Comparability — Rwanda [top]
The Rwandan samples identify polygamous marriages and the Rwandan pointers account for the presence of multiple wives.
Relationship codes changed slightly between 1991 and 2002. Most significantly, the 1991 census lacks an "other relative" code and it is unclear if unspecified other relatives are grouped with non-relatives. Grandchildren and parents are identified in both years, while parents-in-law and children-in-law not separately identified.
Childbearing information is available and consistent across samples.
Comparability — Saint Lucia [top]
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. Cohabitation is very common in Saint Lucia, and in both samples, cohabiting partners of the head are explicitly included with spouses in the relationship-to-head variable. Additional information on consensual unions was collected in both samples, but the 1980 question was asked only for women of reproductive age and is not used in the construction of spouse and parent pointers.
Fertility data were collected only for women not attending school full-time. In 1991, data collection was further limited to women ages 15-65.
Comparability — Senegal [top]
The relationship codes are not comparable across censuses. The 2002 sample collected relationship to household head for persons, identifying grandchildren, parents, and siblings. In-laws are not identified in 2002.
The 1988 sample contains subfamily information. Detailed information on relationship to household head is available only for members of the primary family. Persons in subfamilies (for instance married children of the head, or mother and siblings of the head) are identified only as "other relative" or "other relative or non-relative". Only Rule 1 links are allowed in the primary family, because all other parent-child relationships should be reported in subfamilies.
Links across subfamilies are allowed only when the subfamily head is most likely the child of the household head. Specifically, potential parents must be the household head or spouse, and potential mothers must be 15-44 years older than the subfamily head, while potential fathers must be 20-44 years older than the subfamily head. Links within subfamilies fall under Rule 4.
Polygamous marriages are identified for both men and women in each sample, but information on children ever born and children surviving are available only for the 2002 sample.
Comparability — Sierra Leone [top]
The 2004 and 2015 samples identify men and women in polygamous marriages and includes data on childbearing for all females ages 10 and older. In both samples, relationship categories are detailed, except that parents-in-law are combined with other relatives.
In the 2015 sample, there is a much lower prevalence of grandchildren than in the 2004 sample, and analysis of potential parental links indicated that many grandchildren may have been classified as "Other" (harmonized as "Other relative, not elsewhere classified" in RELATE). Despite this suggestive evidence, we opted not to allow direct parental links between biological children of the household head and "Other relative" children to avoid artificially inflating parental linkage rates.
Comparability — Slovenia [top]
Grandchildren and children-in-law are combined with other relatives, while parents are combined with parents-in-law.
Childbearing information is available.
Comparability — South Africa [top]
The South African samples contain polygamous households, but only the 2000 and 2007 samples identify polygamous marriages. In 1996, we inferred polygamous status based on the presence of multiple spouses of the head. Polygamous households receive special treatment under pointer rules.
Relationship codes changed substantially over time. The 2001-2011 samples provide substantially more detail, including information on adopted and step-children. Children-in-law and parents-in-law are combined with other relatives in 1996, and separately identified later samples. Grandchildren are identified in all years.
Childbearing data were asked in both censuses, but restricted in 2001-2011 to reproductive age women.
The 2001 and 2007 South African censuses collected the person number of strictly biological co-resident mothers and fathers. See the unharmonized variables for the sample. Unlike the IPUMS pointers, these do not recognize social parents.
Comparability — South Sudan [top]
The census provides detailed relationship to household head categories: grandchildren, children-in-law, and combined categories of parents/parents-in-law and siblings/siblings-in-law are identified.
Polygamous unions are not explicitly identified in by marital status, but are inferred for household heads based on the presence of multiple wives. This information is then used to construct pointers without polygamous households.
Childbearing data are available only women ages 12-54.
Comparability — Spain [top]
Pointers could not be constructed for the 1981 Spanish census because the data are not organized into households. Parental pointers are constructed for 1991 and 2001. In 2011 parental location information comes directly from the census form; it is not constructed by IPUMS.
Relationship categories changed between 1991 and 2001. In contrast to earlier years, the 2001 census does not identify grandchildren. Unmarried partners, however, are identified for the first time in 2001.
Childbearing information is available in 1991, but not in 2001.
The 2001 census identifies subfamilies, but provides no information on the relationships within subfamilies. IPUMS links individuals across subfamilies only when the pairing is head or sibling to parent. Within subfamilies, pointers were created based on reported relationship to the household head. The subfamily information was not collected as a census question, but was inferred by the statistical office. The 1991 sample does not include subfamily information.
Both the 1991 and 2001 samples have pointer variables among their original unharmonized variables, but they differ in important ways. The 1991 census collected the person number of the parent (mother or father) listed first in the enumeration. The 2001 mother and father pointers were constructed by the National Institute of Statistics using family names and household order.
The Spain labor force surveys also have family pointer variables among their original unharmonized variables.
Comparability — Sudan [top]
Polygamous unions are not identified in Sudan, but are inferred for household heads based on the presence of multiple wives.
The Sudan census provides detailed information on relationship to household head. Children-in-law are separately identified, while the census reports combined categories for parents/parents-in-law and siblings/siblings-in-law.
Childbearing data are available only for ever-married women ages 12-54.
Comparability — Switzerland [top]
The 2011 sample lacks pointer variables because it is a sample of individuals not organized into households. In other Switzerland samples, 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.
Fertility data are available only in 2000.
Comparability — Tanzania [top]
Polygamous unions are not identified in Tanzania. We inferred polygamy based on the presence of multiple wives and used this information to construct pointers for children of the household head.
Relationship category detail increased between the 1988 and 2002-2012. Parents and grandchildren are not distinguished from other relatives in 1988, while no sample identifies in-laws.
Childbearing data are available in all years.
Comparability — Thailand [top]
Relationship detail changes over time. Biological children are distinguished from step and adoptive children in all years except 2000, while children-in-law are identified in all years. The treatment of parents and parents-in-law also varies over time: in 1970, only parents are separately identified; in 1980 and 1990 parents are combined with parents-in-law; 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.
Childbearing questions are asked of ever-married women.
Comparability — Togo [top]
Subfamilies are only identified in the 1970 sample. In that sample, links are not allowed across subfamilies, except for subfamily heads with a defined relationship to the household head. Grandchildren are not identified in the 1960 sample. The 1960 sample lacks a category for siblings of the head, so "Nephew/nieces" are allowed to link to "Other relative" parents. Close examination also prompted us to allow "Other relatives" to link to "Nephew/niece" and "Cousin" parents in the 1960 sample. Close examination of the 2010 sample indicated that children and grandchildren of the household head were sometimes miscoded as "Other relatives." For this reason, "Other relatives" were allowed to link to Heads and Spouses.
Comparability — Trinidad and Tobago [top]
There is considerable variation in the relationship and childbearing information with which to construct pointers across the Trinidad and Tobago samples. Pointers are not constructed in 1990 because the sample lacks marital status information.
Comparability — Turkey [top]
The 1985 and 1995 Turkish censuses included a combined category of grandchildren and children-in-law. We separately identified these relationship categories using marital status. The 1985 and 1995 censuses combined parents and parents-in-law into a single relationship category. The 2000 sample has more detailed relationship codes, but does not distinguish between unclassified categories of relatives and non-relatives.
Childbearing data were collected from ever-married women in all years.
Comparability — Uganda [top]
Both Uganda samples include polygamous households. In 2002, polygamous marriages are identified in the marital status variable. In 1991, polygamous marriages must be inferred for household heads from the presence of multiple spouses.
The 2002 census provides more relationship detail than the 1991 sample, identifying stepchildren, parents-in-law and siblings-in-law. Grandchildren and children-in-law are never identified.
Childbearing information is available in both samples, but is limited to reproductive age women in 2002.
Comparability — United Kingdom [top]
The 2001 UK sample is a sample of persons not organized into households, and the 1961 and 1971 samples lack reliable information on relationship-to-head. Pointers are constructed only for the 1991 sample.
Relationship codes in 1991 are detailed, identifying grandchildren and in-laws. Childbearing information, however, was not collected.
Comparability — United States [top]
Relationship codes are generally consistent over time. Grandchildren, parents, and in-laws are separately identified in each year. The 1990 census is the first to separately identify unmarried partners. In previous years, partners are combined other non-relatives into a "partner/friend" or "partner/roommate" category.
Childbearing information is available for all women from 1970 to 1990 and for ever-married women in 1960. The 2000-2020 samples do not contain such information.
Comparability — Uruguay [top]
Children-in-law are identified in all years except 1963. Parent and parents-in-law are reported in a combined category in all censuses except 2006 and 2011. Grandchildren are separately identified in only the 1996, 2006, and 2011 censuses.
Fertility data are available in all years except 1963.
Comparability — Venezuela [top]
The 1971 census is considerably less detailed than later samples, combining grandchildren, parents and children-in-law with other relatives. Later samples separately identify these relationships, although parents are combined with parents-in-law.
All years except 1981 provide childbearing information.
Comparability — Vietnam [top]
The 1999 and 2009 censuses combined other relatives and non-relatives together into an "other relative or non-relative" category. For the purposes of pointer creation, this category is treated as "other relatives." Grandchildren are separately identified in 1989, 2009, and 2019, and are included in the "other relative or non-relative" category in 1999. Children-in-law are combined with children in 1989 and with "other relatives or non-relatives" in all other years. Parents-in-law are combined with parents in all years.
In both years, CHBORN was asked only of women of reproductive age: 15 to 49.
Comparability — Zambia [top]
The 2000 census includes some households with multiple families. Because no information is available on relationships between members of different families, pointers are constructed only within families.
Polygamous marriages can be inferred based on the presence of multiple spouses only in 2010. In previous years, polygamous marriages cannot be identified.
The 1990 sample does not identify any categories of relatives other than spouse and child. The 2000 and 2010 censuses identify grandchildren, parents, and children-in-law. Parents-in-law are identified in 2010 only.
Fertility data are available in all years.
Universe
- All persons
Availability
- Argentina: 1970, 1980, 1991, 2001
- Armenia: 2001, 2011
- Austria: 1971, 1981, 1991, 2001
- Bangladesh: 1991, 2001, 2011
- Belarus: 1999, 2009
- Benin: 1979, 1992, 2002, 2013
- Bolivia: 1976, 1992, 2001, 2012
- Botswana: 1981, 1991, 2001, 2011
- Brazil: 1960, 1970, 1980, 1991, 2000, 2010
- Burkina Faso: 1996, 2006
- Cambodia: 1998, 2004, 2008, 2013, 2019
- Cameroon: 1976, 1987, 2005
- Canada: 2011
- Chile: 1970, 1982, 1992, 2002
- China: 1982, 1990, 2000
- Colombia: 1973, 1985, 1993, 2005
- Costa Rica: 1973, 1984, 2000, 2011
- Cuba: 2002, 2012
- Côte d'Ivoire: 1988, 1998
- Denmark: 1845, 1880, 1885
- Dominican Republic: 1981, 2002, 2010
- Ecuador: 1974, 1982, 1990, 2001, 2010
- Egypt: 1986, 1996, 2006
- El Salvador: 1992, 2007
- Ethiopia: 1984, 1994, 2007
- Fiji: 1976, 1986, 1996, 2007, 2014
- France: 1962, 1968, 1975, 1982, 1990, 1999, 2006, 2011
- Ghana: 2000, 2010
- Greece: 1971, 1981, 1991, 2001
- Guatemala: 1964, 1973, 1981, 1994, 2002
- Guinea: 1983, 1996, 2014
- Haiti: 1971, 1982, 2003
- Honduras: 1974, 1988, 2001
- Hungary: 1970, 1980, 1990, 2001, 2011
- India: 1983, 1987, 1993, 1999, 2004, 2009
- Indonesia: 1971, 1976, 1980, 1985, 1990, 1995, 2000, 2005, 2010
- Iran: 2006, 2011
- Iraq: 1997
- Ireland: 1901, 1911, 1971, 1979, 1981, 1986, 1991, 1996, 2002, 2006, 2011, 2016
- Israel: 1972, 1983, 1995
- Italy: 2001, 2011, 2011Q1, 2012Q1, 2013Q1, 2014Q1, 2015Q1, 2016Q1, 2017Q1, 2018Q1, 2019Q1, 2020Q1
- Jamaica: 1982, 1991, 2001
- Jordan: 2004
- Kenya: 1969, 1989, 1999, 2009
- Kyrgyzstan: 1999, 2009
- Laos: 1995, 2005, 2015
- Lesotho: 1996, 2006
- Liberia: 2008
- Malawi: 1987, 1998, 2008
- Malaysia: 1970, 1980, 1991, 2000
- Mali: 1987, 1998, 2009
- Mauritius: 1990, 2000, 2011
- Mexico: 1970, 1990, 1995, 2000, 2005Q1, 2005Q2, 2005Q3, 2005Q4, 2006Q1, 2006Q2, 2006Q3, 2006Q4, 2007Q1, 2007Q2, 2007Q3, 2007Q4, 2008Q1, 2008Q2, 2008Q3, 2008Q4, 2009Q1, 2009Q2, 2009Q3, 2009Q4, 2010, 2010Q1, 2010Q2, 2010Q3, 2010Q4, 2011Q1, 2011Q2, 2011Q3, 2011Q4, 2012Q1, 2012Q2, 2012Q3, 2012Q4, 2013Q1, 2013Q2, 2013Q3, 2013Q4, 2014Q1, 2014Q2, 2014Q3, 2014Q4, 2015, 2015Q1, 2015Q2, 2015Q3, 2015Q4, 2016Q1, 2016Q2, 2016Q3, 2016Q4, 2017Q1, 2017Q2, 2017Q3, 2017Q4, 2018Q1, 2018Q2, 2018Q3, 2018Q4, 2019Q1, 2019Q2, 2019Q3, 2019Q4, 2020, 2020Q1, 2020Q3
- Mongolia: 1989, 2000
- Morocco: 1982, 1994, 2004, 2014
- Mozambique: 1997, 2007
- Myanmar: 2014
- Nepal: 2001, 2011
- Nicaragua: 1971, 1995, 2005
- Nigeria: 2006, 2007, 2008, 2009, 2010
- Pakistan: 1973, 1998
- Palestine: 1997, 2007, 2017
- Panama: 1960, 1970, 1980, 1990, 2000, 2010
- Papua New Guinea: 1980, 1990, 2000
- Paraguay: 1962, 1972, 1982, 1992, 2002
- Peru: 1993, 2007, 2017
- Philippines: 1990, 1995, 2000, 2010
- Poland: 1978, 2002
- Portugal: 1981, 1991, 2001, 2011
- Puerto Rico: 1970, 1980, 1990, 2000, 2005, 2010, 2015, 2020
- Romania: 1977, 1992, 2002, 2011
- Russia: 2002, 2010
- Rwanda: 1991, 2002, 2012
- Saint Lucia: 1980, 1991
- Senegal: 1988, 2002, 2013
- Sierra Leone: 2004, 2015
- Slovenia: 2002
- South Africa: 1996, 2001, 2007, 2011, 2016
- South Sudan: 2008
- Spain: 1991, 2001, 2005Q1, 2005Q2, 2005Q3, 2005Q4, 2006Q1, 2006Q2, 2006Q3, 2006Q4, 2007Q1, 2007Q2, 2007Q3, 2007Q4, 2008Q1, 2008Q2, 2008Q3, 2008Q4, 2009Q1, 2009Q2, 2009Q3, 2009Q4, 2010Q1, 2010Q2, 2010Q3, 2010Q4, 2011Q1, 2011Q2, 2011Q3, 2011Q4, 2012Q1, 2012Q2, 2012Q3, 2012Q4, 2013Q1, 2013Q2, 2013Q3, 2013Q4, 2014Q1, 2014Q2, 2014Q3, 2014Q4, 2015Q1, 2015Q2, 2015Q3, 2015Q4, 2016Q1, 2016Q2, 2016Q3, 2016Q4, 2017Q1, 2017Q2, 2017Q3, 2017Q4, 2018Q1, 2018Q2, 2018Q3, 2018Q4, 2019Q1, 2019Q2, 2019Q3, 2019Q4, 2020Q1, 2020Q2, 2020Q3, 2020Q4
- Sudan: 2008
- Suriname: 2012
- Switzerland: 1970, 1980, 1990, 2000
- Tanzania: 1988, 2002, 2012
- Thailand: 1970, 1980, 1990, 2000
- Togo: 1960, 1970, 2010
- Trinidad and Tobago: 1970, 1980, 2000, 2011
- Turkey: 1985, 1990, 2000
- Uganda: 1991, 2002, 2014
- United Kingdom: 1991
- United States: 1960, 1970, 1980, 1990, 2000, 2005, 2010, 2015, 2020
- Uruguay: 1963, 1975, 1985, 1996, 2006, 2011
- Venezuela: 1971, 1981, 1990, 2001
- Vietnam: 1989, 1999, 2009, 2019
- Zambia: 1990, 2000, 2010
- Zimbabwe: 2012
