An alternative to comparing among multiple (self-)reports of the same relational data, in some cases it is possible to compare to externally verifiable sources of related (if not identical) types of information (e.g., can a respondent accurately recall the timing and duration of phone calls they placed to friends yesterday?). As mentioned in the section on passive data collection strategies in Chapter 3, the use of proximity sensor data can provide useful indicators of behavioral interaction data between members of a population (Salathé et al. 2010). These types of behavioral interactions provide a useful substrate over which models can then develop accurate accounts of disease flow through a population, or to design more efficient strategies for intervention (Salathé and Jones 2010). Others have suggested that such data can serve as a potential alternative strategy for encoding finely grained interaction ties, and aggregating those into approximations of the relational ties others are interested in (Eagle, Pentland, and Lazer 2009). However, while some aspects of relational network structure are jointly captured across differing data collection strategies (especially dyadic level information), others are not (including some of the features of global network structure, adams 2010). In short, we have some confidence in people’s ability to accurately report their own relationships.