Measurement error in the Current Population Survey: a nonparametric look

Area of Science:

• Economics
• Survey Methodology
• Statistical Analysis

Background:

• Accurate income reporting is crucial for economic research and policy.
• Previous studies have highlighted potential discrepancies in survey data.
• Understanding measurement error sources is key to improving data quality.

Purpose of the Study:

• To analyze income reporting errors using linked survey and administrative data.
• To investigate the relationship between measurement error and earnings levels.
• To compare measurement error in cross-sectional versus panel survey designs.

Main Methods:

• Utilized an exact match file linking the 1978 March Current Population Survey (CPS) with Social Security Administration (SSA) administrative records.
• Employed nonparametric statistical methods to analyze income reporting errors.
• Differentiated between cross-sectional and panel data analysis.

Main Results:

• Identified higher measurement error in cross-sectional samples than in panel data.
• Found that the negative relationship between measurement error and earnings is primarily due to overreporting by low earners.
• Observed no significant relationship between median response errors and earnings.

Conclusions:

• Cross-sectional survey designs may introduce more income reporting error than panel designs.
• Low-income individuals are more prone to overreporting earnings in surveys.
• Further research is needed to fully understand the nuances of survey response error in income data.