This paper examines alternative forms of match bias arising from earnings imputation. Wage equation parameters are estimated based on mixed samples of workers who do and do not report earnings, the latter group being assigned earnings of donors who share some but not all the attributes of the...

We report results from a validation study of Nielsen Homescan data. We use data from a large grocery chain to match thousands of individual transactions that were recorded by both the retailer (at the store) and the Nielsen Homescan panelist (at home). First, we report how often shopping trips...

We use Michigan Panel Study of Income Dynamics data and data from a 2008 telephone survey of adults conducted by Westat for the Princeton Data Improvement Initiative (PDII) to explore the importance and feasibility of adding retrospective questions about actual work experience to cross-sectional...

Recent research suggests that rates of extreme poverty, commonly defined as living on less than $2/person/day, are high and rising in the United States. We re-examine the rate of extreme poverty by linking 2011 data from the Survey of Income and Program Participation and Current Population...

When race is not directly observed, regulators and analysts commonly predict it using algorithms based on last name and address. In small business lending--where regulators assess fair lending law compliance using the Bayesian Improved Surname Geocoding (BISG) algorithm--we document large...

Proxies for race are commonly used in settings where race cannot be observed directly. In the context of small business lending, we examine the standard race prediction algorithm (BISG), which regulators use to assess compliance with fair lending laws. The algorithm relies on an individual’s...

The Census Tree is the largest-ever database of record links among the historical U.S. censuses, with over 700 million links for people living in the United States between 1850 and 1940. These high-quality links allow researchers in the social sciences and other disciplines to construct a...

Synthetic microdata - data retaining the structure of original microdata while replacing original values with modeled values for the sake of privacy - presents an opportunity to increase access to useful microdata for data users while meeting the privacy and confidentiality requirements for data...

Differences in answers in Internet and traditional surveys can be due to selection, mode, or context effects. We exploit unique experimental data to analyze mode and context effects controlling for arbitrary selection. The Health and Retirement Study (HRS) surveys a random sample of the US 50+...

Earnings nonresponse in the Current Population Survey is roughly 30% in the monthly surveys and 20% in the annual March survey. Even if nonresponse is random, severe bias attaches to wage equation coefficient estimates on attributes not matched in the earnings imputation hot deck. If nonresponse...