[Potential biases in epidemiological studies using respondent-driven sampling method: a comparison between its...

Area of Science:

• Epidemiology
• Social Sciences
• Statistical Methods

Background:

• Respondent-driven sampling (RDS) is a key method for studying populations lacking sampling frames.
• Face-to-face RDS has limitations; the online version (WebRDS) offers potential advantages but raises concerns.
• Understanding biases in different RDS modalities is crucial for accurate population estimates.

Purpose of the Study:

• To contrast the application and potential biases of face-to-face RDS and WebRDS.
• To evaluate the trade-offs between speed, cost, and data quality in each RDS format.
• To identify specific concerns related to WebRDS, such as network size definition and recruitment clarity.

Main Methods:

• Comparative analysis of face-to-face and WebRDS methodologies.
• Examination of potential sources of bias inherent in each sampling approach.
• Review of assumptions required for unbiased estimation in both RDS formats.

Main Results:

• Both face-to-face RDS and WebRDS can yield unbiased estimates under certain conditions.
• WebRDS offers advantages in speed and cost but may introduce biases due to less direct participant guidance.
• Concerns regarding network size definition and peer recruitment clarity are heightened in WebRDS.

Conclusions:

• Careful consideration of potential biases is essential for both RDS modalities to meet necessary assumptions.
• Further research into analytical approaches tailored to the specific biases of each modality is warranted.
• Ensuring accurate application of RDS principles is critical for reliable population estimates.