Beyond descriptive numbers, networks are often presented in a plot - as we have already shown. Network plots display nodes and ties, but they leave a lot of space for creativity as well. A plot is not a neutral representation of data but rather an artefact for communication (we return to this theme in detail in Chapter 10.1; see also (van Es et al., 2018)). There are several changeable properties of nodes and ties (see Table 4.1) that illustrate the high customisability. Beyond these properties of nodes and ties, the layout of the items in the network can vary and communicate a perspective of the data.
There are many ways to lay out items for the network. Figure 4.6 illustrates how the same data are shown differently depending on the chosen layout algorithm. This illustrates the opportunities to use network analysis to tell different narratives of the data. Others have conducted formal studies on this topic. For example, McGrath et al. (1997); Blythe et al. (1996) show that the layout algorithm impacts how people interpret how prominent and central the nodes are, which nodes act as bridges and how many groups the network has. On the other hand, for other kinds of questions, such as âhow long is the shortest path between nodes?â, Purchase (1998) did not observe differences between layout algorithms. These results indicate that there may be substantive differences among what people see in the network. McGrath et al. (1997) summarises this well:
Interestingly, the 'best' spatial arrangement for a social network may often depend on the information that the arrangement is intended to convey. In this study, the arrangement which led to the most accurate perceptions of the number of groups was different from the one that led to the most accurate perception of relative prominence.This means that researchers must be careful when thinking about what the visualisation communicates and how that corresponds into the theoretical ideas and empirical findings one has. During the explorative stages of the research, one should use different visualisations to ensure insights are not caused by artefacts.
Various force-directed layout approaches are used in network analysis to understand how networks look. In a force-directed layout, connected nodes attract each other, whereas non-connected nodes move further away from others (see Figure 4.6b). Thus, this layout algorithm helps to group clusters of nodes. For intuition on this process, think of nodes as magnets or springs.4.2Just like magnets or springs, nodes can pull other nodes closer or push them farther. Magnets have strength and springs have tensions that affect how they behave. Similarly, force-directed layouts also have several parameters that influence how the network plot appears. The degree of repulsion or attraction can be varied. Depending on the chosen algorithm, further options for fine-tuning the visual representation are available. Therefore, careful consideration is required when doing or interpreting a network plot.