Q re: ASEC record types and structure

When using IPUMSR to create ASEC samples, do all the records have the same internal structure, or are there two or more record types containing different fields or arranging the data differently? If the latter, is there some field that identifies the record type? Where do I look for the field structure for the record type or types?

IPUMS CPS data extracts can be created as rectangular person-level extracts, hierarchical extracts, or longitudinal person-level extracts. These three data structures differ in how the data are formatted within the file.

In a rectangular person-level extract, there is only one record type—a person record. Each row or observation represents a single person, and each column is a variable.

In a hierarchical extract, there are two record types—household records and person records. A household record is a row representing a household. The columns are variables, and only household-level variables will be populated for the household records. Nested beneath each household record are the person records of each person in the household. Household-level and person-level variables will be populated for each person record. The variable RECTYPE reports whether each record, or row, is a person or household record.

In a longitudinal extract, there are only person records, and each variable is observed twice for each person. For example, instead of having EDUC in an extract, you will have EDUC_1 (which reports the person’s EDUC value in time period 1) and EDUC_2 (which reports the person’s EDUC value a year later, in time period 2).

To change your extract type, go to my data, and click revise next to the extract. Then click change next to structure. Hierarchical extracts can only be downloaded as .dat files, which come with syntax files to read into SAS, SPSS, R, or Stata.