Reducing Nonresponse and Data Linkage Consent Bias in Large-Scale Panel Surveys

Area of Science:

• Survey Methodology
• Biostatistics
• Data Science

Background:

• Panel surveys face selection bias due to unit nonresponse over time.
• Inability to link administrative records also introduces selection bias.
• Both biases threaten the validity of panel study conclusions.

Purpose of the Study:

• To discuss methods for reducing selection bias in panel studies.
• To emphasize techniques for mitigating bias in the US Health and Retirement Study.
• To enhance the reliability of longitudinal survey data.

Main Methods:

• Review of recently proposed statistical methods.
• Focus on techniques addressing unit nonresponse.
• Examination of strategies for improving administrative record linkage.

Main Results:

• Identified methods can effectively reduce selection bias.
• Specific techniques are highlighted for the US Health and Retirement Study.
• Improved data validity is achievable through bias reduction.

Conclusions:

• Addressing selection bias is crucial for accurate panel survey research.
• Proposed methods offer practical solutions for nonresponse and linkage issues.
• Mitigating bias strengthens the evidence base from longitudinal studies.