Full Citation

Title: Sample Designs and Sampling Errors in the Integrated Public Use Microdata Series

Citation Type: Journal Article

Publication Year: 1995

ISBN:

ISSN:

DOI:

NSFID:

PMCID:

PMID:

Abstract: Discusses the methods by which the effects of sampling errors that occur in Public Use Microdata Samples (PUMS) are reduced. PUMS are random samples taken from the decennial censuses since 1850 that can be used to determine American social trends over time. Because the samples are entire household units, they are referred to as cluster samples, which sometimes produce a relatively high incidence of error. This problem is solved by providing design factors for variables such as age, sex, and marital status. Design factors are based on the standard error for each cluster, which in turn is based on standard deviation. These design factors allow for more accurate use of the PUMS, as does the technique of stratification. This involves separating the cluster populations into strata by certain key characteristics, such as household size, home ownership, and race. Stratification has the beneficial effect of further reducing the standard error in cluster samples. An appendix discusses the process by which the PUMS cluster samples for each census were created.

User Submitted?: No

Authors: Ruggles, Steven

Periodical (Full): Historical Methods

Issue: 1

Volume: 28

Pages: 40-46

Data Collections: IPUMS USA

Topics: Methodology and Data Collection

Countries: