A new stopping rule for surveys.

Area of Science:

• Survey Methodology
• Statistical Inference
• Data Collection Optimization

Background:

• Non-response significantly impacts survey accuracy and necessitates strategies to mitigate bias.
• Current survey designs often inflate sample sizes to achieve target response rates, increasing costs.
• Existing stopping rules for data collection may not fully leverage available respondent and non-respondent data.

Purpose of the Study:

• To develop and evaluate a novel rule for determining when to cease data collection in sample surveys.
• To compare the performance of the new stopping rule against existing methods, such as those proposed by Rao, Glickman, and Glynn (RGG).
• To provide a data-driven approach for optimizing survey data collection efficiency and reducing costs.

Main Methods:

• Development of a new stopping rule integrating complete interview data and covariates from non-responders.
• Utilizing simulations to compare the proposed rule with the RGG rules.
• Implementation and assessment of the new rule using data from a real-world survey.

Main Results:

• The new stopping rule demonstrates potential for more efficient data collection by considering respondent data and non-respondent characteristics.
• Simulations indicate that the proposed rule can effectively identify when additional data collection is unlikely to substantially alter survey estimates.
• Comparative analysis suggests the new rule offers an alternative to traditional response rate-based stopping criteria.

Conclusions:

• The developed stopping rule offers a statistically grounded method to optimize survey data collection.
• This approach can lead to more efficient resource allocation in surveys affected by non-response.
• Further application and validation in diverse survey contexts are warranted.