Estimation tools for reducing the impact of sampling and nonresponse errors in dual-frame RDD telephone surveys

Area of Science:

• Survey Methodology
• Statistics
• Public Health Research

Background:

• Dual-frame random-digit-dial (RDD) telephone surveys utilize both landline and cell phone frames.
• Estimators for population totals in these surveys are susceptible to both sampling and nonsampling errors.
• Achieving an optimal balance between landline and cell phone samples can be challenging.

Purpose of the Study:

• To introduce alternative estimators for population totals in dual-frame RDD telephone surveys.
• To develop an application of shrinkage estimation to reduce sampling variability.
• To demonstrate the impact of differential nonresponse on survey weighting.

Main Methods:

• Application of shrinkage estimation to mitigate sampling variability.
• Analysis of differential nonresponse mechanisms based on telephone status.
• Utilizing data from the National Immunization Survey-Child for illustration.

Main Results:

• Developed alternative estimators for dual-frame RDD surveys.
• Showcased the utility of shrinkage estimation for reducing sampling variability.
• Illustrated the effects of differential nonresponse on survey weighting.

Conclusions:

• Alternative estimators, including shrinkage estimation, can improve population total estimates in dual-frame surveys.
• Understanding and accounting for differential nonresponse is crucial for accurate survey weighting.
• The methods are validated using real-world survey data.