Usuário:Abdo/MMMNS/RDS/en
De Stoa
< Usuário:Abdo | MMMNS
Defining the Network
Choices of link:
- The elegibility criteria (current)
- Question: of those people that you know, how many would you consider inviting to the study?
- Question: of those people that you know, how many would you think would invite you to the study?
RDS calculations are affected mainly by the effect of the definition of relationship over the degree of individuals.
Definitions of relationships improve RDS calculations if they:
- are more precise
- are more reciprocal
- correspond to what people actually do with their coupons
- Precision
- True
- Proportional to reciprocity
- Suffers from error in guessing
- Reciprocity
- True (low transmission error hypothesis)
- Unknown
- Unknown
- Correspondence
- Poor
- Improves if reciprocity is high
- Improves if guessing is precise
Method
Exchange definitions to compare the results in each scenario, and see if there's a favorite by comparing them with other, non-RDS, estimates.
As the non-RDS estimate might be unreliable, do that in several locations, with different sources of non-RDS data.
Conclusion
Ask both questions, if not possible, keep #2.
The main advantage of #2 over #3 is that it does not involve guessing, and it becomes the best if there's strong reciprocity overall.
Also consider that asking both questions might confuse the subject.