A generalized three-way data paradigm for cognitive social networks is introduced. The paradigm requires each actor to report on a given network relation, for even ordered pair embedded in the network. In this way, each actor 'describes' the entire network from an ego-based perspective. The richness of information obtained makes it possible to compare individual perceptions with each other and with respect to a global aggregate view of the structure. Correspondence analysis is discussed as a descriptive multidimensional scaling method to achieve these goals. It is also possible to investigate each actor's perception of the social structure in terms of structural properties like centrality, reciprocity, and transitivity. Tests are developed for the null hypotheses that centrality ranks, reciprocity, and transitivity are at chance levels, controlling for tie density. Some examples of current statistical models for three-way data an also discussed.