3.3.4 Cognitive Social Structures (CSS)

Cognitive social structures are not complex networks in the same sense as the examples described above, but it borrows some of the same data structure characteristics that stem from multiple representations of the same simple graph. Essentially the CSS approach asks all members of a study population to report on their perceptions of relationships among all members of that population (Krackhardt 1987). So, if N is the number of nodes in the population, you end up with N NxN adjacency matrices. As can be seen in Figure 3.8, this means that each person {k} provides data about the presence of ties for each {i,j} pair. For example, students in a class may all be asked to identify friendship groups within their classroom. Analyses can then aggregate across these multiple reports, to use collective perceptions of network position to predict outcomes of interest like aggression or school performance (Neal and Neal 2017).

Figure 3.8: Cognitive Social Structure Data Format. These ‘stacked’ adjacency matrices are used to represent the each report (Xijk) about each tie in a CSS. As with other adjacency matrices, i represents the sender of a tie, and j the receiver, while the added dimension k indicates which node provided the report. I.e., X231 indicates 1’s report of the tie from 2 to 3.
Figure 3.8: Cognitive Social Structure Data Format. These ‘stacked’ adjacency matrices are used to represent the each report (Xijk) about each tie in a CSS. As with other adjacency matrices, i represents the sender of a tie, and j the receiver, while the added dimension k indicates which node provided the report. I.e., X231 indicates 1’s report of the tie from 2 to 3.

While Chapter 2 described how to sample and measure networks, this chapter described the principles behind the many platforms within which researchers can accomplish those tasks. More so than how sampling, measurement, and the BSP, the practices laid out in this chapter are subject to rapid changes. As was briefly indicated in the examples using tablet, API, or other digital methods for data collection, technological innovations are likely to continue to be a site for continued development in the foreseeable future. While the practices may continue to change, we must continue to keep in mind how these implementations address the broad principles of social network data collection. The next chapter turns to examining the unique ethical considerations facing social networks research.