Full Citation
Title: An Assessment of the National Representativeness of Children in the 2007 Panel Study of Income Dynamics
Citation Type: Working Paper
Publication Year: 2012
ISBN:
ISSN:
DOI:
NSFID:
PMCID:
PMID:
Abstract: There are many ways that nationally-representative samples drawn for panel surveys can become non-representative over time. For example, selective attrition and non-response, immigration, and shortcomings in the initial sample may lead to a divergence of the panel sample from the corresponding population it seeks to represent. Although initial cross-sectional weights together with attrition weights in later waves may correct for some sources of non-representativeness, they do not correct for all possible sources. Furthermore, a periodic reassessment of sample representativeness is important for understanding the cumulative effects of attrition and immigration and determining whether additional measures are necessary to correct for declining representativeness and coverage.In this paper, we assess the national representativeness of the 2007 sample of children in the Panel Study of Income Dynamics (PSID)a total of approximately 7,100 children aged 017 years of age. We assessed the national coverage for PSID children using microdata from the American Community Survey (ACS) and a new statistical technique called generalized boosted regression models. The ACS provides a gold-standard comparison based on its extremely high (98%) response rate, excellent data quality and completeness, and large sample sizes (approximately 700,000 children aged 017 years in 2007). We constructed a reasonably consistent set of covariates across PSID and ACS to describe children on the basis of their age, race/ethnicity, sex, poverty status, geographic region, and having a foreign-born parent. The generalized boosted models provided flexible, non-parametric estimates for assessing the relationship between our dependent variable (an indicator of whether an observation in the pooled ACS-PSID sample came from PSID) on covariates and their interactions. We also used the ACS sample to identify cases that are logically excluded from PSID (e.g., children whose parents are both post-1997 immigrants to the U.S.), and assess the size and composition of this group.Our results reveal that PSID child sample provides good representation of the corresponding population with coverage of approximately 97% of the U.S. population of children in 1997 and reasonable balance for most groups, although with some exceptions.
User Submitted?: No
Authors: Duffy, Denise; Sastry, Narayan
Series Title:
Publication Number: 12-01
Institution: PSID
Pages:
Publisher Location:
Data Collections: IPUMS USA
Topics: Other
Countries: