Online Program Home
My Program

Abstract Details

Activity Number: 654 - Evaluating and Reducing Nonsampling Errors in Surveys
Type: Contributed
Date/Time: Thursday, August 1, 2019 : 10:30 AM to 12:20 PM
Sponsor: Government Statistics Section
Abstract #304192 Presentation 1 Presentation 2 Presentation 3
Title: Impact of Certified Mail on Nonresponse Rates
Author(s): Redouane Betrouni* and Peter Schilling and Dedrick Owens and Bac Tran
Companies: U.S. Census Bureau and U.S. Census Bureau and U.S. Census Bureau and U.S. Census Bureau
Keywords: Experimental design; Classification regression trees; Bayesian inference; Nonresponse; nonsampling errors
Abstract:

The Annual Survey of Public Employment and Payroll (ASPEP) is conducted by the U.S. Census Bureau to collect data on federal, state, and local governments’ civilian employees and their gross payrolls. Quality of survey estimates can decrease in the presence of low response rate due to the effects of nonresponse bias. We conducted an experiment in the 2017 Census of Governments: Employment Component (the government census-year version of ASPEP) to measure the relative nonresponse rates of units receiving certified mail follow-up reminders versus regular telephone mail follow-up. The study design sought to eliminate factors that might cloud the contrasting impact of competing treatments. In this research, supervised machine learning methods were used to design our experiment and data analysis was conducted using both Bayesian statistical inference and traditional methods. The research concludes that using certified mail follow-up reduces nonresponse relative to regular mail follow-up. We recommend certified mail follow-up as a cost and quality effective option when Telephone Follow-up cost per unit exceeds a predetermined critical value.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program