Scenario 5: Family Structure Because human infants require the long term care and nurturing of adults before they can fend for themselves in often hostile environments, the family in some identifiable form is universal. In the United States, the family has generally been understood to consist of an adult male with a female partner and immature offspring as the nuclear family. At times this nuclear family becomes extended to include other relatives and also non-relatives living in a common household. The decennial censuses and more recently the American Community Survey (ACS) provide a historical record of the composition of American households. The census data compiled by the University of Minnesota Integrated Public Use Microdata Series (IPUMS) project allow us to look at that historical record over the span from 1850 through 2007 with ACS data added annually. Data on the household structure as given by the variable HHTYPE are available for 1900-2007 with the exception of 1920.
1: Use the IPACS08 data set with PDQ-Explore to determine the percentage composition of households from 1900 through 2007. Question 1: What are the distributions for the types of American households over the period 1900-2007? >=1900 hhtype Percent by row hhtype Type of household in the housing unit. when the data were collected. (>=1900 selects persons in the 1900 though 2000 decennial census samples and the 2001 through 2007 ACS samples.) Results and Comment: Taking the results from this query as presented, several significant changes over time are apparent. The percentage of Married couple family households has decreased steadily from approximately 70% in 1900 to 40% in 2007. Relatively large increases are seen in the Male householder living alone and Female householder living alone categories. Making sense of this table requires familiarity with census definitions of household, householder, and family. The N/A category includes both vacant housing units and group quarters. The householder, not living alone and householder, living alone categories refer to non-family households where the householder has no relatives living within the housing unit. These are distinguished from the Male householder, without wife and Female householder, without husband categories where the householder is living with relatives, typically children, as a family. The GQ (Group Quarters) variable may be used to drop the vacant and group quarters housing units from the results by selecting only housing units that contain households.
Also note that the ACS s from 2001 and on compress the time scale to single intervals rather than the 10 intervals for the earlier decennial censuses. The data from 1850 through 2000 are from the decennial censuses while 2000-2007 are from the ACS. Some s have more than one data set available. The default weight for PDQ- Explore queries using these data implicitly selects the first data set available for each (datanum=1). Each of these is generally the largest sample available for the given. 2: Use the IPACS08 data set with PDQ-Explore to determine the percentage composition of households from 1900 through 2007. Select only housing units containing households to drop housing units that are vacant or group quarters from the analysis. Question 2: What are the distributions of household types in American housing units over the period 1900-2007 after dropping vacant and group quarters units? >=1900 & (gq=1 gq=2 gq>=5) Year hhtype Percent by row hhtype type of household in the housing unit. Gq identifies vacant housing units, households, and group quarters when data were collected. (>=1970 selects persons in the 1900 though 2007 samples.) Results and Comment: Selecting categories 1, 2, and 5 on GQ drops from the tabulation vacant housing units and group quarters. Expand the variable GQ under the IPACS09 and Housing headings to see the code categories for the variable. For a full description of the variable, go to the IPUMS USA website at www.ipums.org. The IPUMS project provides extensive on-line documentation for their archives of census data.
3: Let s look at some of these changes by race and ethnicity. For example, using the same Universe/Selection as used in the previous example, tabulate RACEHISP by HHTYPE=3 (Female householder, no husband present) for s 1900 through 2007. Question 3: What proportion of American households are made up of women living as families with no husbands present? >=1900 & (gq=1 gq=2 gq>=5) hhtype=3 racehisp Percent by row hhtype type of household in the housing unit. gq identifies vacant housing units, households, and group quarters racehisp PDQ custom item recode that identifies five race and Hispanic categories. when data were collected. (>=1900 selects persons in the 1900 though 2007 samples.) Results and Comment: The increase in the percentage of single-female headed families over the s is very apparent and especially noteworthy is the dramatic increase in the proportion of Non-Hispanic Black households that are families headed by a single female. RACEHISP is a recode of the original RACE and HISPAN variables that creates four non-hispanic race categories (White, Black, Asian, and Other non-hispanics) and one Hispanic category. Selecting the FOREACH option under the Axis 3 Swap Dimension tab expands the results and makes it easier to scroll across the set of RACEHISP tables. Note again that the ACS s from 2001 and on compress the time scale to single intervals rather than the 10 intervals for the earlier decennial censuses. These decennial census and ACS files are hierarchical files that nest person records within housing records. When queries reference both housing and person data, the records are implicitly flattened with the housing data considered to be appended to the person record. PDQ-Explore assigns the default housing or person weight based upon the variables included in the query. The default can be overridden by specifying an alternative weight to be used.
Some PSC-Explore functions (specifically SUM, MIN, MAX, and PICK) create a result at a given hierarchy level by scanning over the records nested below that hierarchy. For example, MAX(AGE) will return the age of the oldest person in the housing unit and tag it as a housing level variable. 4: Rather than looking at the households in terms of the householder, we can take a different perspective and look at the changes from the perspective of children within the households. Let s select children (age<=18) over the s, look at the household types in which they reside, and again view the results by race and ethnicity. Let s initially contrast children living with a single female parent against those living in all other household types. Question 4: What proportion of children live in households headed by women with no husband? Look at this by and by raceethnicity. >=1900 & age<18 hhtype=3 racehisp Percent by row age age of the person in s. hhtype type of household in the housing unit. racehisp PDQ custom item recode that identifies five race and Hispanic categories. when data were collected. (>=1900 selects persons in the 1900 though 2007 samples.) The increase in the percentage of non-hispanic Black children living in a singleparent, female-headed household is dramatic with much of the increase coming in the 1950-1990 s. The relative increases for the other race/ethnic groups are also significant with the increases starting perhaps a decade or two later. Experiment with using the Axis drop-down menus to view the table from different perspectives. For example, set Axis 01: to FOREACH, leave Axis 02: hhtype=3 as COLUMN, and set Axis 03: racehisp to ROW. Note that the FOREACH setting makes it easier to scroll up and down through a series of tables.
Look at children living in other household types. To view the household type categories, expand the HHTYPE entry under HOUSING and IPACS08 in the PDQ- Explore workspace window. For additional information on household types, go to the www.ipums.org website. 5: Family structure changes over time in response to a variety of internal and external factors. For example, the change from a predominantly agrarian society to one where less than 1% of the population lives on farms has had some impact on the value of children as laborers in a family. On the other hand, difficult economic times, unemployment, or high costs of housing may result in children remaining at home for longer periods of time. We can easily look at the proportion of young adult children living at home over the s. Question 5: What proportion of 23-30 -old children live at home? age>=23 & age<=30 child racehisp Percent by row age age of the person in s. child PDQ custom item recode that identifies a person as the child or child-in-law of the household head. racehisp PDQ custom item recode that identifies five race and Hispanic categories. This distinguishes women from men. when data were collected. This query covers the s from 1850-2007. When have the percentages of 23-30 children living at home peaked? Do they differ by race and ethnicity? What might account for the variations by time and by race? Would the results vary by state or urban area? Are these children largely students living at home? Try this for other age ranges?
6: Another way to look at young adult children who live at home is to see if we can find traits that distinguish them from those that do not live at home. We can select a along with an age range to consider and then set the column to the CHILD recode. A number of characteristics can be tabulated by row to see if we can find obvious differences between the two groups of young adults. One might expect that education, school enrollment, gender, having a job, cost of housing, or where a person lives might be factors. Question 6: What characteristics distinguish young adult children who live at home from those who do not? =2007 & age>=23 & age<=30 metaread Child Percent by column or by row age age of the person in s. child PDQ custom item recode that identifies a person as the child or child-in-law of the household head. metaread metropolitan area where housing unit is located. when data were collected. (=2007 selects persons in the 2007 ACS sample.) This query covers the s from 1850-2007. When have the percentages of 23-30 children living at home peaked? Do they differ by race and ethnicity? What might account for the variations by time and by race? Would the results vary by state or urban area? The resulting rows or columns can be sorted by clicking on a row or column heading. The first click gives a descending order, the second an ascending order, and the third restores the original order. Do not mix row and column sorts; i.e., once sorting by a column, continue sorting by column until the original order is restored before sorting by row. Which metropolitan areas have the highest percentages of 23-30 -olds living at home as children of the heads.
7: Let s look at children living in poverty. The poverty variable in the PUMS and ACS data expresses the individual s family income relative to the poverty level with the codes ranging from 0 through 501. A poverty score of 100 indicates family income exactly at the poverty level while 501 indicates family income more than five times the poverty level. The poverty code is based on family income and the family composition. Although each child is given a poverty score, the score is the same for all members of a family, but not for all members of the household. Question 7: What are the poverty levels for children in the U.S. for racialethnic groups? =2007 & age<18 poverty/100 racehisp Percent by column or by row age age of the person in s. racehisp PDQ custom item recode that identifies five race and Hispanic categories. poverty ratio of family income to poverty level for each individual. when the data were collected. (=2007 selects persons in the 2007 ACS sample.) This query is only for 2007. Has childhood poverty increased or decreased over the s? Look at these results by area (state or metro areas).
8: The racial composition of families in the U.S. has also changed over the s. The PUMS and ACS data allow us to explore those changes. Note how the percentage of mixed-race marriages has increased over the s. Question 8: What are the percentages over the s of married household heads who have a spouse of a different race? relate=1 & marst=1 racehisp <> pick(relate=2,racehisp) Percent by column or by row age age of the person in s. marstt marital status of the person. pick PDQ function that returns the value of the associated variable for the first match on the condition RACEHISP for the spouse of head (RELATE=2) is returned. racehisp item recode that identifies five race and Hispanic categories. relate relationship of the person to the head of household. when data were collected. This query covers the s from 1850-2007. When have the percentages of mixed-race marriages peaked? Do they differ by state, race and ethnicity, or other traits? The PICK function is used here to get the RACEHISP code for the person who is the spouse of the head (RELATE=2), which is then compared to the RACEHISP code for the head (RELATE=1). Only married, spouse present heads are selected (MARST=1).
9: The age composition of families in the U.S. has also changed over the s. A simple indicator would be the percentage of households that have one or more persons age 80 and older in the household. Question 9: Across the s, what are the percentages of households that have one or more persons 80 s of age or older in the housing unit? <All Records> max(age)>=80 Percent by column or by row age T age of the person in s. max PDQ function that returns the maximum value over the records in the hierarchy for the given argument age in this case. when data were collected. This query covers the s from 1850-2007. When have the percentages of households containing one or more octogenarians or older peaked. Select a single and look at this by state or metropolitan area. What other indicators of aging American households could be used? The MAX function is used here to get the age of the oldest person residing in the household. The function scans across the person records associated with a housing unit and returns the maximum value found for the argument, AGE in this case. References IPUMS Documentation: http://www.usa.ipums.org/usa PDQ-Explore Handout: Filling in the Blanks Michael Rosenfeld, 2007. The Age of Independence: Interracial Unions, Same- Sex Unions and the Changing American Family. Cambridge, Harvard University Press. (See http://news.stanford.edu/news/2007/june13/rosenfeld-061307.html for an informative review.)