In-law Classification Procedures for 2008 American Community Survey (ACS) and 2008 Puerto Rican Community Survey (PRCS)
In the 2000-2007 ACS and 2005-2007 PRCS questionnaires, the Census Bureau combined the householder's parents-in-law, siblings-in-law, and children-in-law into a general in-law category. For consistency with other years and greater accuracy in constructing family inter-relationship variables, the IPUMS broke this general category back into its three components using a scheme described here. Starting with the 2008 ACS and 2008 PRCS questionnaires, the Census Bureau provided separate categories for parents-in-law and children-in-law, with the result that most siblings-in-law are likely to be contained in the residual "other relative" category. This required a different scheme, which uses four pieces of information:
- the other relative's age
- the household head's age
- the household head's spouse's age
- the presence in the household of married siblings of the head
Generally speaking, people who are relatively close in age to the head and/or the head's spouse are classified as siblings-in-law rather than other relatives. The permissible age differences are relaxed somewhat when there are married siblings of the head in the household, which suggests that siblings-in-law are likely to be present.
More specifically, "other relatives" are classified as siblings-in-law if they meet at least one of the following conditions:
- They are no more than 10 years younger or older than the householder's spouse (if present)
They are married, the householder has at least one married sibling in the household, and at least one of the following is true:
- They are no more than 12 years younger than the householder and no more than 17 years older than the householder
- They are no more than 12 years younger than the householder's spouse (if present) and no more than 17 years older than the householder's spouse (if present)
- They are linked as the spouse of the householder's sibling via SPLOC
Assessing the error rates
Applying this classification scheme to the 2000 5% PUMS (which contains full detail on the in-law categories), 221,064 (31.9 percent) of the 693,751 siblings-in-law are misclassified as other relatives, and 194,999 (7.5 percent) of the 2,582,964 other relatives are misclassified as siblings-in-law. In total, then, 415,863 (12.7 percent) of the 3,276,715 combined siblings-in-law and other relatives would have been misclassified. In contrast, not using this classification scheme would have resulted in 693,751 errors, since all siblings-in-law would have been misclassified as other relatives. Thus the scheme reduces the total number of errors by 277,888, or 40 percent.
|Actual code from 2000 Census|
|Relationship under classification scheme||Sibling-in-law||472,087||194,999||667,086|
Note: Weighted data from 2000 5% PUMS, using the 1990 definition of households.
Additionally, the classification scheme accurately reproduces the overall distribution of siblings-in-law and other relatives in the 2000 5% PUMS:
|Actual relationship in 2000 5% PUMS||Relationship under classification scheme||Relationship without classification scheme|
|Total siblings-in-law/other relatives||100.0%||100.0%||100.0%|
Although the classification scheme preserves the cross-temporal comparability of the sibling-in-law and other relative relationship categories, users who want to combine the two categories can easily recode IPUMS-USA RELATE codes of 8 to 10. However, they should realize that this code will not perfectly represent the data as it would have existed without the classification scheme, as family interrelationship variables incorporate the relationships as refined by the classification scheme.