Related Experiment Videos
Assessing response bias in random-digit dialling surveys: the telephone-prefix method
P Diehr1, T D Koepsell, A Cheadle
1Department of Biostatistics, University of Washington, Seattle 98195.
Statistics in Medicine
|June 15, 1992
Summary
This study developed a method to identify survey non-response bias by analyzing phone number prefixes. The health behaviors survey data may under-represent older adults and college-educated individuals.
Area of Science:
- Public Health
- Survey Methodology
- Statistics
Background:
- Understanding survey non-respondent characteristics is crucial for assessing generalizability to the target population.
- A health behaviors survey achieved a 54% overall response rate, necessitating an investigation into potential biases.
Purpose of the Study:
- To develop and evaluate a novel method for identifying non-response bias in survey data.
- To assess potential demographic biases in a random-digit dialing health behaviors survey.
Main Methods:
- Grouped attempted phone numbers by prefix to analyze the association between response rates and prefix characteristics.
- Employed a simulation study to validate the method's ability to detect non-response biases under specific assumptions.
Main Results:
- The analysis indicated that the survey data likely under-represent older individuals and those with a college education.
- No significant biases were detected in reported health behaviors, potentially due to unmet assumptions of the method.
Conclusions:
- The developed prefix-based method shows promise for identifying non-response bias in survey research.
- Further research is needed to refine the method and its underlying assumptions for broader applicability.