Developing an Optimal Approach to Account for Late-Filed Returns in Population Estimates

• Activity Number: 568
• Type: Topic Contributed
• Date/Time: Wednesday, August 4, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Cynthia Belmonte*+ and Charles D. Day and Paul Arnsberger and Brian Raub
• Companies: IRS and IRS and IRS and IRS
• Keywords: non-response bias ; population estimates ; post-stratification ; Bernoulli sampling

Abstract

Estimates for populations of interest for Statistics of Income (SOI) programs are produced by drawing stratified, random Bernoulli samples of tax and information returns as they are filed, over predetermined sampling periods that often span multiple years. While this methodology results in the inclusion of the majority of targeted returns, a small number of returns for each study are filed beyond the data collection period, potentially introducing non-response bias into the population estimates. For a given sampling period, the paper will analyze historical filing patterns to develop an approach for accounting for late-filed returns. This research will assess the weight adjustment approach currently used in SOI's estate tax study and will provide a basis for application of a similar approach in each of the exempt organizations and private foundations studies.