Assessing response reliability of health interview surveys using reinterviews

Area of Science:

• Health Services Research
• Survey Methodology
• Public Health

Background:

• Household surveys are crucial for assessing community health status and healthcare payment factors.
• While sampling error effects are understood, non-sampling errors often exceed them and are rarely assessed.
• Non-sampling errors can significantly influence the interpretation of survey findings and policy decisions.

Purpose of the Study:

• To report and analyze non-sampling errors in a household survey conducted in Sierra Leone.
• To identify specific question types that are more susceptible to non-sampling errors.
• To provide insights for improving future survey designs and data reliability.

Main Methods:

• The study employed a re-interview methodology to assess non-sampling errors.
• Original interview responses were compared with those from subsequent reinterviews.
• Data analysis focused on quantifying discrepancies and identifying patterns in non-sampling errors.

Main Results:

• Non-sampling errors were found to be present in the household survey data.
• Certain question types demonstrated a higher susceptibility to non-sampling errors compared to others.
• The magnitude of non-sampling errors may significantly affect the validity of survey-derived community parameters.

Conclusions:

• Non-sampling errors are a critical consideration in the interpretation of household survey data.
• Survey designers and policymakers should be aware of and actively assess non-sampling errors.
• Methodological improvements focusing on question design are needed to mitigate these errors and enhance data quality for health policy.