Codes and Frequencies
Most IPUMS data transformations are performed using variable harmonization tables that specify how each value in the source data is recoded. Some variables also require programming logic in addition to the harmonization table. The harmonization documents for this variable are:
- Complex/technical variable: contact email@example.com for details.
SPRULE explains the criteria by which the IPUMS-International variable SPLOC linked the person to his/her probable spouse.
IPUMS-International establishes spouse-spouse links according to five basic rules, and SPRULE gives the number of the rule that applied to the link in question. A sixth rule identifies sample-specific linking procedures only imposed in selected instances.
The design of the interrelationship variables is described in this paper on IPUMSI family linking methodology.
Comparability — General
SPRULE 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 sample information page).
SPRULE works the same general way for all samples. Both partners are required to be age 12 and older and have a marital status of married, consensual union, or unknown. The only exception to the marital status requirement is for a household head and their spouse or unmarried partner (as indicated by RELATE), who may be linked regardless of reported marital status. Marital status is determined from MARST and, for some samples, CONSENS.
Many samples identify more than one type of "married" status, including being in a consensual union. SPRULE and SPLOC do not require an exact match of marital status codes, because we determined that few if any of our datasets consistently supported such distinctions. However, for the weaker links (Rules 3 and 4), exact matches happen first, before pairings of mismatched marital codes are allowed, and most sample-specific links (Rule 6) require matching marital status.
Possible pairings of spouses depend on the persons' relationships to the household head. A small number of samples have relationship information for subfamilies within households. For these samples, SPLOC uses relationship to the head of subfamily (SUBFREL) instead of RELATE. Only persons within the same subfamily can link to one another. The samples for Brazil, Greece 2011, Hungary, Ireland (except 1996 and historical samples), Mexico 1970, Poland 1978 and 2002, Senegal 1988, and Togo 1970 identify subfamilies in this way.
The 1971-1987 German samples provide no information on relationship to household head. The 1971 and 1981 samples, however, report subfamily membership. Although Germany 1987 has no subfamily information, the number of complex households is small and we allow only conservative links.
The 1985 and 1990 Turkish censuses combined children-in-law and grandchildren into a single relationship category. We have distinguished between these two categories based on the observation that children-in-law are always ever-married while grandchildren are rarely married. Consequently, grandchildren do not receive spouse pointers in these samples.
In polygamous unions (where the man is identified as polygamous), each female spouse identifies the husband's person number in SPLOC. The husband in polygamous unions points to the first female spouse that links to him, which will generally be the most proximate in the data. In samples in which only men are identified as polygamous (Cameroon 1976, Guinea 1983 and 1996, Egypt 1986-2006, Iraq 1997, Mali 1987 and 1998, Rwanda 1991, Sierra Leone 2004, South Africa 2001 and 2007, Togo 1970, and Uganda 2002), multiple women can link to a polygamous man as long as the women are in a marital union of some kind. Finally, some samples do not identify polygamous unions, although polygamy was widely practiced. We allow multiple female spouses to link to heads in those samples: Bangladesh, Botswana, Ethiopia, Ghana 2000, Iran, Jordan, Kenya 1969, Malawi, Morocco, Mozambique, Nigeria 2006, 2007 and 2009, Palestine, Pakistan, South Africa 1996 and 2011, South Sudan, Sudan, Tanzania, Uganda 1991, and Zambia 2010. Polygamous unions not involving the head and spouse cannot be identified in these samples, but such unions are relatively rare.
The implementation of the links depends to a degree on the relative positions of the persons within the household. For the basic rules (1 to 4) adjacent persons are linked first. In most samples the person order has meaning, but the power of the adjacency test varies across data sets.
We emphasize consistency across samples in the algorithms that determine SPLOC and SPRULE. We use the same age constraints and relationship pairings across samples. Those pairings consistently allow specified relatives (e.g., sibling) to link to residual categories such as "other relative" even in those samples that had a specific "in-law" category that seemed the logical complement (e.g., sibling-in-law). Careful analysis of the data convinced us that many more good links than bad were created by allowing such weaker pairings, even in samples with seemingly reliable data.
In some samples, however, we were compelled to customize the links beyond what could be implemented more generally. The biggest issue concerned large numbers of married children who were not linked because their apparent spouse was also identified as a "child" rather than a child-in-law or other-relative. Our solution is generally to allow child-to-child links if the census form did not have a specific category for child-in-law. However, some censuses had instructions that specified or strongly implied that children-in-law were to be included with "other relatives." In those samples we only allow child-to-child links if comparison with other samples for that country suggested that the census instructions were ineffective. The sample-specific links are identified as Rule 6 in the data and are implemented after all other pairings have been made. To further limit the possibility of bad links, the spouses are usually required to have identical marital statuses.
The codes for SPRULE are as follows. A lower-numbered rule (greater than 0) takes precedence over a higher numbered rule.
0 = No spouse of this person was present in household.
1 = A married/in-union woman and a married/in-union man were linked because they were adjacent in the dataset and both persons' relationship to the household head (see RELATE) justified a link. A woman will link to an adjacent preceding man before linking to an adjacent subsequent man in the household. Marital statuses can differ between spouses as long as both persons are some category of "married." Possible relationship pairings for rule 1 include:
Head to Unmarried partner
Parent to Parent
Parent-in-law to Parent-in-law
Parent-in-law/Stepparent to Parent-in-law/Stepparent
Child to Child-in-law
Child to Stepchild/Child-in-law
Sibling to Sibling-in-law
Aunt to Uncle
Grandparent to Grandparent
(All relationship pairings are not available in every sample.)
2 = Same as rule 1, but the married/in-union woman and married/in-union man were not adjacent. A woman will link to the most proximate preceding man before linking to the most proximate subsequent man in the household.
3 = A married/in-union woman and a married/in-union man were linked because they were adjacent in the dataset and both persons' relationship to the household head (RELATE) suggested a possible pairing and the female was no more than 20 years older or 35 years younger than the male. A woman will link to an adjacent preceding man before linking to an adjacent subsequent man in the household. The rule is executed first for persons with identical marital status codes before being performed again to allow persons with mismatched "married" codes to link. The possible relationship pairings are as follows:
Nonrelative to Nonrelative
Other-relative-or-nonrelative to Other-relative-or-nonrelative (RELATE code 6000)
Child to Other relative
Child to Other-relative-or-nonrelative
Sibling to Other relative
Grandchild to Grandchild
Grandchild to Other relative
Nephew/niece to Nephew/niece
Cousin to Cousin
Unknown to Unknown
Head to Unknown
Child to Unknown
(All relationship pairings are not available in every sample.)
4 = Same as rule 3, but the married/in-union woman and married/in-union man were not adjacent in the data. A woman will link to the most proximate preceding man before linking to the most proximate subsequent man in the household.
5 = Other consensual union links. This rule identifies consensual union links between relationship or marital status pairings not allowed in the general rules above. It is implemented after the general rules. The following pairings are allowed:
Any relative to Non-relative (both indicate consensual union)
Head to Spouse (one indicates consensual union and the other is single)
6 = Sample-specific links. This rule is implemented last, and addresses situations arising in selected samples. The most common sample-specific rules implemented at present concern situations in which children are allowed to link to children, as discussed above. Other examples include samples where examination of the data indicate that a non-specific category, such as "Other relative" or "Other non-relative", was used for the spouses or partners of more specific categories, such as "Child", "Sibling", or even "Head". In most cases, marital statuses must match exactly. The samples that allow child-to-child links are:
Papua New Guinea 1990-2011
Sierra Leone 2004
South Africa 1996
- All persons
- 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
- 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
- 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
- Germany: 1971, 1981, 1987
- 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
- Jamaica: 1982, 1991, 2001
- Jordan: 2004
- Kenya: 1969, 1989, 1999, 2009
- Kyrgyz Republic: 1999, 2009
- Laos: 2005
- 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, 2010, 2015
- 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
- Philippines: 1990, 1995, 2000, 2010
- Poland: 1978, 2002
- Portugal: 1981, 1991, 2001, 2011
- Puerto Rico: 1970, 1980, 1990, 2000, 2005, 2010
- Romania: 1977, 1992, 2002, 2011
- Russia: 2002, 2010
- Rwanda: 1991, 2002, 2012
- Saint Lucia: 1980, 1991
- Senegal: 1988, 2002, 2013
- Sierra Leone: 2004
- 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
- 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
- Uruguay: 1963, 1975, 1985, 1996, 2006, 2011
- Venezuela: 1971, 1981, 1990, 2001
- Vietnam: 1989, 1999, 2009
- Zambia: 1990, 2000, 2010
- Zimbabwe: 2012