I hope you're beginning to get a feel of how to detect visually when a model is visually and analytically effective. One of the keys to getting to this point is to test multiple methods before deciding on a final choice. For example, we might have felt perfectly good about using the OpenOrd graph with our school data if we hadn't experimented with other algorithms. For this data, we would certainly select another model, because we were able to see the results from each approach.
Equally important is to consider how we are trying to frame the data and ultimately our graphical output. We should always ask ourselves a few questions before settling on a final choice:
Am I trying to provide an overview of the entire network, or is my goal to focus on a specific node and its relationships?
What is more critical to my display – showing the sheer number of connections between nodes or their intensity (frequency)?
Are there groups in my data that should be treated as...