Examining the Impact of Data Collection Interventions on Data Quality, Cost, and the Risk of Nonresponse Bias

• Activity Number: 437
• Type: Invited
• Date/Time: Wednesday, August 6, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Survey Research Methods Section
• Author(s): Benjamin M. Reist*+ and Stephanie M. Coffey
• Companies: U.S. Census Bureau and U.S. Census Bureau
• Keywords: Adaptive Design ; Paradata ; Propensity Modeling ; R-Indicators ; Nonresponse Bias ; NSCG

Abstract

The Census Bureau is incorporating adaptive design techniques in multiple surveys and the Decennial Census in an effort to maintain or increase data quality while controlling costs. Model-based use of paradata including contact history, interviewer observations, and cost are beginning to guide data collection decisions. This talk mentions the wide-ranging adaptive design efforts underway at the Census Bureau, and then provides a case study of the National Survey of College Graduates (NSCG).

The 2013 NSCG included a mode switching study that aimed to increase data quality by creating a more balanced respondent population. By leveraging NSCG frame information and survey paradata, we calculated R-indicators throughout data collection to monitor and evaluate representativeness, and intervened by tailoring contact strategies and switching modes during data collection. We analyze whether the mode switching experiment improved data quality while allowing for cost control. We report on R-indicators, subgroup response rates, and cost estimates. Finally, we examine whether creating a more balanced respondent population in the NSCG actually has the potential to reduce nonresponse bias.