tetano
Editor, Senior Moderator
BMC Med Res Methodol
. 2025 Oct 22;25(1):234.
doi: 10.1186/s12874-025-02702-2. Evaluating the accuracy of survey data: a case study of COVID-19 vaccination rates in Germany
Karolina von Glasenapp[SUP] 1 [/SUP]
Affiliations
Background: Surveys are an important source of timely and comprehensive population health data and play a crucial role in public health research and policymaking, as shown during the COVID-19 pandemic. However, the reliability of survey data depends on their accuracy, which is often difficult to assess due to the limited availability of benchmark data. This study evaluates the accuracy of survey estimates of the COVID-19 vaccination rate in Germany and examines the impact of survey design and adjustment weights on accuracy.
Methods: I compared survey estimates of the COVID-19 vaccination rate from multiple surveys conducted in Germany between 2021 and 2022 against administrative data on the vaccination rate as an external benchmark. Accuracy was assessed by calculating absolute and relative deviations between survey estimates and the administrative data. Further, I analyzed accuracy differences based on survey design, focusing on sampling procedure and survey mode, and I examined whether adjustment weights improved estimation accuracy.
Results: The accuracy of survey estimates varied over time, with early surveys underestimating the vaccination rate and later surveys tending to overestimate it. Probability-based mixed-mode or personal interview surveys yielded more accurate estimates than did other survey designs. While some nonprobability web surveys performed well, their accuracy varied considerably. The application of adjustment weights generally improved estimation accuracy, suggesting that this weighting technique effectively addressed some sources of survey error.
Conclusions: Survey-based estimates of COVID-19 vaccination rates should be interpreted with caution due to widespread inaccuracies, particularly overestimation. Probability mixed-mode and personal interview surveys produced more accurate estimates, underscoring the importance of robust survey methodologies. The fact that I found that adjustment weights enhanced accuracy highlights their value in survey research. These findings provide valuable insights for public health researchers and survey methodologists in designing and interpreting health-related survey data.
Keywords: Accuracy; Adjustment weights; COVID-19 vaccination rate; Data quality; Survey design.
. 2025 Oct 22;25(1):234.
doi: 10.1186/s12874-025-02702-2. Evaluating the accuracy of survey data: a case study of COVID-19 vaccination rates in Germany
Karolina von Glasenapp[SUP] 1 [/SUP]
Affiliations
- PMID: 41126070
- DOI: 10.1186/s12874-025-02702-2
Background: Surveys are an important source of timely and comprehensive population health data and play a crucial role in public health research and policymaking, as shown during the COVID-19 pandemic. However, the reliability of survey data depends on their accuracy, which is often difficult to assess due to the limited availability of benchmark data. This study evaluates the accuracy of survey estimates of the COVID-19 vaccination rate in Germany and examines the impact of survey design and adjustment weights on accuracy.
Methods: I compared survey estimates of the COVID-19 vaccination rate from multiple surveys conducted in Germany between 2021 and 2022 against administrative data on the vaccination rate as an external benchmark. Accuracy was assessed by calculating absolute and relative deviations between survey estimates and the administrative data. Further, I analyzed accuracy differences based on survey design, focusing on sampling procedure and survey mode, and I examined whether adjustment weights improved estimation accuracy.
Results: The accuracy of survey estimates varied over time, with early surveys underestimating the vaccination rate and later surveys tending to overestimate it. Probability-based mixed-mode or personal interview surveys yielded more accurate estimates than did other survey designs. While some nonprobability web surveys performed well, their accuracy varied considerably. The application of adjustment weights generally improved estimation accuracy, suggesting that this weighting technique effectively addressed some sources of survey error.
Conclusions: Survey-based estimates of COVID-19 vaccination rates should be interpreted with caution due to widespread inaccuracies, particularly overestimation. Probability mixed-mode and personal interview surveys produced more accurate estimates, underscoring the importance of robust survey methodologies. The fact that I found that adjustment weights enhanced accuracy highlights their value in survey research. These findings provide valuable insights for public health researchers and survey methodologists in designing and interpreting health-related survey data.
Keywords: Accuracy; Adjustment weights; COVID-19 vaccination rate; Data quality; Survey design.