Data Cart

Your data extract

0 variables
0 samples
View Cart
RELATE
Relationship to household head
Return to Demographic variables list

Description

RELATE describes an individual's relationship to the head of household or householder. Beginning in 1880, data on household relationship was asked of every person. The general relationship code is reasonably comparable across years. The detailed code makes distinctions that cannot be made in all years.

The relationship codes are divided into two categories: relatives (codes 1-10) and non-relatives (codes 11-13). In general, the codes for relatives are self-explanatory. The non-relative codes are divided into three groups: "Partner, Friend, Visitor," roughly described as persons who do not pay or work for their accommodations (unless they share ownership); "Other Non-Relatives," including those persons paying or working for accommodations; and "Institutional Inmates." See the comparability discussion for further information about the coding scheme.

RELATE is not available for 1850-1870, but the IPUMS variable IMPREL produces similar results. As a convenience, the extract system is set up so that users may include RELATE in extracts of the 1850-1870 samples. In those years, RELATE contains the information that is documented in the IMPREL variable description.

Codes and Frequencies




Can't find the category you are looking for? Try the Detailed codes

Comparability

1. Group Quarters

The general code is comparable across all years. Users should note, however, that there are some fundamental differences between the early period (before 1940) and the later period (1940-2010, the ACS, and the PRCS). Group quarters residence is a primary distinction in the relationship variable for the later period, but before 1940 and in the 1940 100% dataset "relationship to head" was recorded regardless of group quarters status. Persons classified as related to the head (codes 1 through 10) in the early period would have been classified in the "Other non-relative" category based on their group quarters status in the later years.

We decided not to impose consistency between the two periods because it would have resulted in the loss of too much information in the early period. Users may recode RELATE for greater comparability across the two periods, but we recommend caution; a recode may lose considerable information, changing what were considered ordinary households in the early years into group quarters. To impose the 1970 group quarters definition on all years, create a new general RELATE category "14"--'Non-institutional group quarters members'--and classify all group quarters members (see GQ) from RELATE categories "1" through "12" into the new category. The meaning of the general codes for RELATE would remain the same, except that categories "1" through "12" would be strictly household members, and categories "13" and "14" would contain group quarters members (separated into institutional and non-institutional members).

In addition to the sharp differences between the pre-1940 and 1940-present periods, the definition of group quarters has changed in other years as well; see GQ for more information.

Note: The ACS and PRCS sampling frame did not include group quarters before 2006, and the 2000-2005 samples therefore lack the relationship categories that would be associated with persons in such living arrangements.

2. Category differences

Beginning in 1960, the census questionnaire provided a changing number of check boxes for specific relationships, followed by an open-ended write-in category. We believe the check boxes are likely to yield more reliable information. Users are encouraged to examine the relevant census forms to see which categories were written on the forms in particular censuses from 1960 to 2010. Prior to 1960 there were no pre-specified categories; all responses were open-ended.

Some of the following categories posed unique problems due to changing definitions and categories across census years:

Head/householder:

Beginning in 1980 the census questionnaire no longer referred to a "head of household", specifying instead the designation of "person one"--the first person listed on the census form. This reference person could be any household member in whose name the property was owned or rented. If no such person was present, any adult could be selected. Prior to 1970, enumerators were instructed to record the male as the head of household for all married couples, regardless of the couple's designation of a head.

Spouse:

In all years, the "spouse" and the householder must be opposite sexes, according to the Census Bureau's policies. See the "Partner/friend" discussion, below, for details about how same-sex "spouse" responses were recoded.

Child:

Step-children are not identified from 1960 to 1980, or in the 2000-2007 ACS and the 2005-2007 PRCS. The 1990 census form had a check-box for step-child, but still did not distinguish between natural-born and adopted children. The 2000 and 2010 census forms and the 2008 ACS/PRCS form specified separate categories for natural-born, step-, and adopted children. In 1990, step-children were reported as "step-children" even if adopted by the householder, while in 2000, an adopted step-child was to be reported as adopted. Natural-born, step-, and adopted children are all considered "own children" by the IPUMS, regardless of age or marital status. Foster children are considered non-relatives (see below).

In-laws:

In 2000, the two categories "parent-in-law" and "son-in-law/daughter-in-law" were added to the census form. Some "parents-in-law" belong to households where the head has never married--an impossible combination. These are probably step-parents of the head: since there was no "step-parent" option on the questionnaire, respondents likely selected "parent-in-law" as the next-best choice.
The 2000-2007 ACS and 2005-2007 PRCS questionnaires provided a single category combining all types of in-laws: child, sibling, and parent. We distinguished among these types using a classification scheme described here. When applying this scheme to the 2000 5% PUMS data, in which full detail on in-laws is available, 3.9 percent of in-laws are misclassified. Users who do not want to use the detailed in-law classifications can recode children-in-law, parents-in-law, and siblings-in-law into a general in-law category.
Starting with the 2008 ACS and PRCS, the Census Bureau provided detail on parents-in-law and children-in-law, but not siblings-in-law, who were likely classified by respondents as "other relatives." As in the 2000-2007 period, we have attempted to pull siblings-in-law out of this "other relatives" category using a different classification scheme described here. Applying this scheme to the 2000 5% PUMS results in 13.1 percent of siblings-in-law and other relatives being misclassified. Users who do not want to use the refined classification can recode siblings-in-law as other relatives.
Finally, the 2006-2008 3-year ACS/PRCS contains a blend of these two schemes: the 2006 and 2007 data in this file is classified as in the single-year files for those years, but the 2008 data collapses parents-in-law and children-in-law into a general in-law category while lumping siblings-in-law in with "other relatives". We use the 2008 classification procedures to pull out siblings-in-law, then classify all members of the "in-law" category who are younger than the householder as children-in-law and all those who are at least as old as the householder as parents-in-law. The error rates here are negligible.

Partner/friend:

This category changes considerably in meaning across census years. Before 1960, the category "partner" refers to a non-relative who shares the home and expenses with the head, including responses such as co-head and business partner. In 1960 and 1970, the 'Partner/friend' value is a residual category for all non-relatives in the household who are not employees or lodgers. In 1980, 'Partner/roommate' refers to a non-relative who lives with the householder and shares expenses. In 1990, 2000, and the ACS, this category is split into 'Housemate or roommate' and 'Unmarried partner'--the latter of which is supposed to distinguish those with a "close personal relationship" with the householder (see "spouse," above).
Unmarried Partner:
Relationships between unmarried, same-sex partners cannot be consistently identified before 2000. The "unmarried partner" category is not comparable between 1990 and 2000. In 1990, same-sex spouse reports were recoded in one of two ways: one of the respondents' sexes was recoded or the "spouse's" relationship was recoded into another relationship. In 2000, the policy was changed; in this census, all same-sex spouse responses were recoded as "unmarried partner" responses.

Other non-relatives:

The group quarters distinction makes this category particularly unwieldy, so in order to simplify the detail codes we separated this category into Pre-1940 system and 1940-2000 system. In general, the Pre-1940 System does not make any distinctions between households and group quarters, but the 1940-2000 system separates households and group quarters. Any use of the detail codes across several census years will require particular attention to this category and possibly some recoding. The following groups represent most of the people classified as "Other non-relatives:"
Roomers/boarders/lodgers and foster children:
Before 1940 roomers and in the 1940 100% dataset, boarders, lodgers, and foster children are classified individually when possible. In 1940-1990, they are all classified together while distinguishing between those in rooming houses (group quarters, '1293') and those in households. In 1940 and 1950, however, boarders and lodgers in group quarters are in the 'Other non-inmates 1940-1950' (1284) category with students and military. In 2000, "foster child" was a separate category from "roomer/boarder".
Students, military, and members of religious orders:
The student (1222) category in the Pre-1940 system generally include people enumerated in secondary boarding schools and some elementary and secondary students enumerated in orphanages and asylums. Since 1940, people in college dormitories and military installations are classified as group quarters members under 'Military' and 'College dormitory,' except in 1940 and 1950 when they are part of the category 'Other non-inmates 1940-1950' (1284). The Pre-1940 system also contains a category for members of religious institutions.
Employees:
In the Pre-1940 system (including 1940 100%), we only retained the distinctions between the most common employee relationships. Most of these categories are domestic employees (servants, housekeepers, maids, cooks, nurses, and other probable domestics) who worked for the family. "Other employees" includes unspecified employees, farm workers, institutional employees, and other employees hired for business purposes. We recommend using the occupation variables (OCC and OCC1950) to determine the specific occupation of employees. In the 1940-2000 system the classification of employees is not consistent. All employees in households are classified "125-" except in 1990, when they are included in the 'Other non-relatives' (1260). Group quarters employees are only identified in 1940 and 1950, and for the other years employees are part of the 'Other non-inmates' categories (1270, 1294, and 1295).
Other non-relatives:
This classification in the early years (1230) includes any unspecified non-relatives, and is rather consistent for these years. In the 1940-2000 system this category (1260) is not as consistent; as mentioned above, in 1990 this includes employees in households, but for 1940, 1950, and 1980 employees are classified elsewhere. Also, the 1960 and 1970 censuses classified any non-relative other than lodgers or employees in the 'Partner/friend' category (1110).
Other non-inmates:
There are four 'Other non-inmate' categories with subtly different definitions. 1990 has the broadest definition; 'Other non-inmates 1990' (1270) includes people enumerated in military installations, college dormitories, rooming houses, and employees and heads of group quarters. In 1940 and 1950 people living in college dormitories, military installations and rooming houses are included in 'Other non-inmates 1940-1950' (1284), but employees, heads, and relatives of staff of group quarters are classified elsewhere. 1960 and 1970 have the narrowest definition, classifying military installation, college dormitory, and rooming house residents and non-inmates in institutions in separate categories, leaving only employees in 'Other non-inmates 1960-1970' (1295). 1980 has a definition similar to 1960/70, but 'Other non-inmates 1980' (1294) includes non-inmates in institutions, in addition to employees.

3. Note on coding in the 1910 and 1920 Puerto Rican Census samples

Entries of "hijo/hija de crianza" were coded as foster child, and entries of "socios/business partners" were coded as non-relative in the 1910 and 1920 Puerto Rican Census samples.

Universe

  • All persons.

Availability

United States
  • 2022: All samples
  • 2021: All samples
  • 2020: All samples
  • 2019: All samples
  • 2018: All samples
  • 2017: All samples
  • 2016: All samples
  • 2015: All samples
  • 2014: All samples
  • 2013: All samples
  • 2012: All samples
  • 2011: All samples
  • 2010: All samples
  • 2009: All samples
  • 2008: All samples
  • 2007: All samples
  • 2006: All samples
  • 2005: All samples
  • 2004: All samples
  • 2003: All samples
  • 2002: All samples
  • 2001: All samples
  • 2000: All samples
  • 1990: All samples
  • 1980: All samples
  • 1970: All samples
  • 1960: All samples
  • 1950: All samples
  • 1940: All samples
  • 1930: All samples
  • 1920: All samples
  • 1910: All samples
  • 1900: All samples
  • 1880: All samples
  • 1870: All samples
  • 1860: All samples
  • 1850: All samples
Puerto Rico
  • 2022: All samples
  • 2021: All samples
  • 2020: All samples
  • 2019: All samples
  • 2018: All samples
  • 2017: All samples
  • 2016: All samples
  • 2015: All samples
  • 2014: All samples
  • 2013: All samples
  • 2012: All samples
  • 2011: All samples
  • 2010: All samples
  • 2009: All samples
  • 2008: All samples
  • 2007: All samples
  • 2006: All samples
  • 2005: All samples
  • 2000: All samples
  • 1990: All samples
  • 1980: All samples
  • 1970: All samples
  • 1930: All samples
  • 1920: All samples
  • 1910: All samples

Flags

QRELATE  QRELATE2 

Editing Procedure

RELATE (Relationship to reference person) and MARST (Marital status)
ACS Years: 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015
ACS editing procedure:
Note: There is an important change in 2013 in how the ACS addressed same-sex couples. The detailed editing and allocation process is described below. To briefly summarize: prior to 2013, RELATE and MARST would be edited to not include same-sex married couples. In 2013, same-sex married couples were included in the ACS and these editing rules were removed. Beginning in 2014, two parents and/or parents-in-law of the same sex were also allowed.
Note: The editing procedure for relationship to the reference person (RELATE) and marital status (MARST) in part depend on the original order of people in the household. This information is not available in public use data.
If a person is living in a 1-person household and their marital status (MARST) is missing, it will be allocated from another person in a 1-person household of the same sex (SEX) and similar age (AGE). When this happens, the flag variable will indicate the value of MARSTwas allocated (QMARST).
If there is someone in the household who is older than 14 (AGE), is the spouse of the reference person (RELATE), is a different sex than the reference person (only applies prior to 2013), but the reference person's marital status is missing or something other than "Married", the reference person's marital status will be changed to "Married." When this happens, the flag variable will indicate the value of MARST was allocated (QMARST).
If there is someone in the household who is older than 14 (AGE) and is the unmarried partner or the same-sex spouse of the reference person (only applies prior to 2013), and the reference person's marital status is "Married" or missing, the reference person's marital status will be allocated. The allocated value will be drawn from another reference person who is the same sex, similar age, and also living in a household with an unmarried partner who has a marital status of widowed, divorced, separated, or never married. When this happens, the flag variable will indicate the value of MARST was allocated (QMARST).
Only applies prior to 2013: For people who are listed as the reference person's spouse and are 15 or older, if the spouse is the same sex as the reference person the RELATE value will be replaced with "Unmarried partner." If they are listed as "Married" marital status will be allocated from another person who is the same sex, similar age, and an unmarried partner who has a marital status of widowed, divorced, separated, or never married.
If a person under 15 is listed as a "Housemate/roommate," RELATE will be replaced with "Roomer/boarder" or "Other non-relative" (alternating between the two values). When this happens, the flag variable will indicate the value of RELATE was allocated (QRELATE).
If a person over 18 (in 2008 and later: over 21) or who is not "never married" is listed as a "Foster child," RELATE will be replaced with "Roomer/boarder" or "Other non-relative" (alternating between the two values). When this happens, the flag variable will indicate the value of RELATE was allocated (QRELATE).
There are numerous situations that will cause a person's RELATE value to be allocated.

  • A person under 15 is listed as a "Spouse"
  • Beginning in 2010: A person listed as "Spouse" who is 50 or more years old/younger than the reference person.
  • A "Son/daughter" is not 12 to 69 years younger than a male reference person or 12 to 50 years younger than a female reference person
  • In 2008 and later: A son- or daughter-in-law who is less than 15 years old.
  • A "Brother/sister" is more than 35 years older or younger than the reference person
  • A "Father/mother" is less than 12 years older than the reference person
  • In 2008 and later: A "Parent-in-law" is less than 12 years older than the reference person
  • Prior to 2014: A "Father/mother" if there is more than one person with the same sex listed as a "Mother/father" (e.g., a household has two fathers of the reference person). In 2014 and later, this occurs if there are more than two parents of the same sex.
  • In 2008 to 2013: A "Parent-in-law" if there are more than one person with the same sex listed as a "Parent in law" (e.g., the reference person has two fathers-in-law) In 2014 and later, this occurs if there are more than two parent-in-laws of the same sex.

  • A "Grandchild" is less than 30 years younger than the reference person
  • An "Unmarried partner" or "Spouse" if there is another person in the household with this RELATE value (the person who comes first in the household record retains their RELATE value; the RELATE of the person who comes later in the household record will be allocated)
  • An "Unmarried partner" when the reference person married, when the unmarried partner is married, or when the unmarried partner is under 15
  • Beginning in 2010: An "Unmarried partner" who is 50 or more years younger/older than the reference person
  • If a person's RELATE value is missing, both this person and the reference person are married, and this person and the reference person are different sex, then RELATE will be replaced with Husband/Wife. In 2010 and later, this step also requires the person being no more than 50 years younger/older than the reference person. In all other cases, a missing RELATE value will be allocated.

When RELATE is allocated, it will be drawn from another person who:

  • Is the same sex as the person whose RELATE is being allocated
  • Is in the same broad age range (15 or 15+)
  • Has a similar age difference between themselves and the household reference person
  • Lives in a household with a reference person that is the same sex and marital status as the reference person in the household of the person whose RELATE is being allocated.


The above allocations will be looped through at least twice. After these allocations, final corrections are made for relative age:

  • A "Son/daughter" with an age that is less than 12 years younger than the reference person, will be replaced with "Brother/sister" or "Other relative" (alternating between the two values).
  • A "Son/daughter" with an age that is more than 50 years younger than the reference person, will be replaced with "Grandchild" or "Other relative" (alternating between the two values).
  • Prior to 2008: "Brother/sister" with an age that is more than 35 years older or younger than the reference person, will be replaced with "In-law" or "Other relative" (alternating between the two values).
  • In 2008 and later: "Brother/sister" with an age that is more than 35 years older or younger than the reference person, will be replaced with "Other relative"
  • A "Grandchild" with an age that is less than 30 years younger than the reference person will be replaced with "Brother/sister" or "Other relative" (alternating between the two values).

The flag variable will indicate the value of RELATE was allocated (QRELATE).
After these edits, MARST will be changed if necessary to be consistent with RELATE. If the spouse of the reference person is not listed as "Married," MARST will be replaced with "Married." Anyone who is under 15 and is not the reference person or the reference person's spouse will be listed as "Never married." Any "Foster child" will be listed as "Never married."
In 2008 and later: If a son- or daughter-in-law is "Never married," MARST will be allocated from another son- or daughter-in-law with the same sex and a similar age.
If someone has a missing value of MARST it will be allocated if the person is 15 or older and whose value of RELATE is son/daughter, brother/sister, father/mother, grandchild, in-law, other relative, room/boarder, housemate/roommate, nonrelative, or unmarried partner. The allocated value will be from another person who is the same sex, has the same value of RELATE, and is a similar age. If an unmarried partner (RELATE) has a value of "Married" for MARST it will be allocated from another unmarried partner who is the same sex and a similar age.
The flag variable will indicate when the value MARST is allocated (QMARST)
Internal ACS variable : REL and MAR