For a web-based survey of college students collecting information about sexual behavior and drug use, where direct identifiers are not collected but IP addresses may be present, risk of harm should be evaluated by which factors?

For a web-based survey of college students collecting information about sexual behavior and drug use, where direct identifiers are not collected but IP addresses may be present, risk of harm should be evaluated by which factors?

• A. Data quality and response rate.
• B. Investigator credentials.
• C. Sample size and duration.
• D. Magnitude and probability.

Answer: (D)

The key idea is evaluating confidentiality risk by two dimensions: how bad the harm could be (magnitude) and how likely it is to happen (probability). In a web-based survey on sensitive topics like sexual behavior and drug use, even if you don’t collect direct identifiers, IP addresses can sometimes be used to identify someone or be linked to other data. That means the potential harm if confidentiality is breached may be substantial, and the chance of a breach or reidentification still exists. So you assess both how severe the consequences would be for a participant and how likely it is that those consequences could occur, given how data are stored, transmitted, and protected. That dual focus—how big the harm could be and how probable it is—best captures the risk of harm in this scenario. The other factors listed—data quality and response rate, investigator credentials, or sample size and duration—pertain to study quality, oversight, or scope more than to the actual likelihood and severity of potential harm to participants.

The key idea is evaluating confidentiality risk by two dimensions: how bad the harm could be (magnitude) and how likely it is to happen (probability). In a web-based survey on sensitive topics like sexual behavior and drug use, even if you don’t collect direct identifiers, IP addresses can sometimes be used to identify someone or be linked to other data. That means the potential harm if confidentiality is breached may be substantial, and the chance of a breach or reidentification still exists. So you assess both how severe the consequences would be for a participant and how likely it is that those consequences could occur, given how data are stored, transmitted, and protected. That dual focus—how big the harm could be and how probable it is—best captures the risk of harm in this scenario. The other factors listed—data quality and response rate, investigator credentials, or sample size and duration—pertain to study quality, oversight, or scope more than to the actual likelihood and severity of potential harm to participants.