So now we've got some insight into how we can describe collaboration networks in terms of the five main building blocks that help us to understand network size, network strength, range, centrality, and density. Also how we can map collaboration networks inside an organization that you're interested in. How can you actually draw those collaboration networks, figure out who is collaborating with whom. In this fourth segment we're going to look at how can we evaluate our collaboration networks. So we've got our network map, we know some things about it. But is it good or is it bad? Are people doing what they should be doing in terms of collaborating or not? So evaluating collaboration networks is where the rubber really starts hitting the road in terms of really starting to get some value out of all these tools and techniques that we can use to analyze collaboration. So here is our 15 person team again, new product development team. And now we've really got their network map. We've got all the data, that we need and we're going to start thinking about, okay what can we really take away from this, what insights can we generate from our analysis? So there's two main questions that we can look at here and there are others too, but these are probably two of the most important ways in which we can use network tools and techniques to evaluate collaboration inside organizations. The first is to look at how do collaboration patterns vary inside the organization and across the different people in the network? And then the second is how do collaboration patterns matter for important outcomes? And so these maybe outcomes that are at the individual level, group level, organizational level, maybe it's a unit of a different kind, maybe it's a team, but how do collaboration patterns matter for outcomes that we care about? So we're going to take each of these two questions in turn. And start off with, how do collaboration patterns vary, okay? So here is where we apply those five building blocks that we introduced back in the second segment; network size, strength, range, density, centrality and there are other network measures too, but these are five of the most important metrics that we can use. And I'm just going to focus on one right now for simplicity, let's say network size okay? So if we look at network size and we're going to map, what we want to do is to use this building block of network size and we want to compare across each of the individuals in the network how big is their network, right? And I'm not going to do every single individual in the network, there's 15 of them, I'm just going to pick 5 people. So we have Lee, John, Paul, Helen and Julia. Now network size, in this particular network that we've been looking at, is actually a little bit complicated because as you've seen there are one way arrows and there's two way arrows. There's network size in terms of what we call inbound size which is the number of people who seek information from person A. So we can count the number of people who seek information from Lee, from John, from Paul, right, from Helen and from Julia. There's also network size in terms of the number of people that Lee goes to, to get information, so your network may be different if you count the number of people that you go to, or the number of people who come to you. Those are two different aspects of the network size that are both embedded in this particular diagram. So if we think about the in-bound size of the network, Lee has three people who come to him for information. And the out-bound size, he goes to two people to get information that he needs. So even just looking at these simple numbers across these five people, we start to ask ourselves some questions, right? And this is really what network methods allow you to do. They don't give you all the answers, but they allow you to ask pointed and important questions about the network and who's collaborating with whom. What we can see, for example, very strikingly is Paul. We know that Paul is an important player. Here we see that 9 people go to Paul for information that they need. And Paul only goes to 3 people himself right? So Paul is doing a lot of giving there and not a lot of taking [LAUGH] right? On the other hand John only has 1 person who comes to him for information and he goes to 2 people. So he's inbalanced in the other direction. Helen and Julie are pretty balanced they got the same number of people coming to them as they go to. So again this is not to say that this is good or bad necessarily but when we look at these sort of very simple descriptive statistics based in this case just on that one building block of network size. We can look across individuals, and we can say, well, Paul is doing something that looks pretty different from John, and even different from Helen and Julia. We can look across individuals, and we can also look over time, and understand how their networks are changing over time, right? So are the number of network ties, the outbound ties changing over time, are there numbers of inbound ties changing over time. If we do, for example, a network survey at different points in time, after one year, after two years. So if we're trying to change the networks, are those changes really taking place? So why do these kinds of statistics matter, what can they help us to do? Well, if you think about this from a people analytics perspective, there's lots of things in terms of managing people, that being able to see these kinds of numbers, and analyze these kinds of statistics, can really help us with, right? They help us with things like, just some examples here, performance assessment. So if we're evaluating Paul, we probably should be evaluating Paul not just on how he does his own work but on how much he's helping people. That's a very important part of what he's spending his time doing. And if we're not evaluating him on that, he's not getting any credit for that. And that relates to the pay and promotion,s are we not only accessing but also rewarding this kind of behavior. because if we're not rewarding it, it's going to stop, so we may want to change the way that we pay people, the way that we access people, to take into account some of these inbalances that we're seeing, or the patterns we're seeing in the network. Do people have the right roles and responsibilities. We might be thinking about, well Paul actually is a very senior person who shouldn't be spending all his time giving other people information. He has more valuable things to do with his time, maybe there is somebody less senior who could be doing this better. So it helps us to think about whether people have the right roles or responsibilities and it might help us also with things like training and mentoring. So if you look at the people who really seem to need a lot of help from other people, maybe those are people who should be targeted for particular kinds of training or mentoring kinds of programs. Or maybe they're just junior in the organization, they're finding their way around and this will play itself out over a couple of years. So we have to really understand why, when we see a network pattern, it's happening. But the first step is to see the pattern and be able to analyze and understand it. So that's the first question. How do collaboration patterns vary? The other important set of questions that we can use or look at when we're evaluating networks. Really important here is how do collaboration patterns matter for important outcomes that we care about for our employees, or our teams, or our groups, or our units, or the organization as a whole? So what is it these collaboration patterns actually affect? Again, say we've got these five building blocks, and maybe we're just going to look at network size for a minute but what we really want to do now is to map that to individual outcomes, again they could be a group or organizational outcomes, but let's say individual outcomes. And there's lots of individual outcomes we might care about in terms of managing people inside organizations, but let's just focus on performance, right? Performance is a very important individual outcome, so the question that we then construct to answer when we have these kind of network data are, how do the attributes of our collaboration network affect individual performance? And here this gets very complicated. There's lots of possibilities, lots of ways to look at and analyze these kinds of patterns. So just to give you a very simple example. We can look essentially at correlations between attributes of the network, so say it's network size, and individual performance. And so again we have these two kinds of network size in this example. So we have inbound ties, the number of people who seek information from you. And maybe you find that there's a positive correlation between the number of people that seek information from you and how well you perform. And maybe you might see that, because people who tend to perform well, other people come to them for advice. But outbound ties, maybe you see a negative correlation with performance, the people who tend to go to other people, go to a lot of other people for information that they need for their work. Maybe they tend to perform less well. And that's not causal. It's not because they go to other people that they perform less well, but it's usually probably because they're junior or they don't know the area so well. So again, we might see a result but we then have to ask ourselves why do we see that result, before we can actually take action. But being able to evaluate how our network building blocks are associated with performance outcomes or outcomes that we care about, is a critical step in being able to get to that intervention. So what we're doing here is a correlational analysis. Or when you get more sophisticated, you're going to be thinking about a multivariate analysis where you can control for some of the variables, and look at the effects of others more specifically in terms of how they matter for outcomes. You're trying to identify relationships between the network variables and outcomes. And again, what kinds of implications does this have for managing employees? Well all sorts, right? Depending on what your outcomes are that you're trying to look at. Again, it can matter for performance assessment, for roles and responsibilities, pay and promotion. Who you want to train and mentor. Maybe it matters for job rotations and career development. You want to get people opportunities to grow their networks, if that really you find that, that is really associated with better performance. Maybe even retention. If you find the people who a lot of people come to for advice, maybe your outcome that you're going to measure is whether people feel burned out, they're more likely to leave the organization, and so this is going to have big implications for retention. So depending on which network variables you look at, and even more importantly, what outcome variables you look at, you're going to be able to come up with action implications that are going to be very important for the kinds of people, managing people implications that we care about, when we're thinking about people analytics. One thing to be very cautious about is that there is no one best collaboration network for every organization and every situation. So again it's not the bigger networks are always better, it's not the denser or sparser networks are always better. You really have to be able to understand how does a particular network configuration map to outcomes in a particular situation. So to understand what's best for your organization and your situation, you need to collect and analyze the data. You can't just say, well we have this network configuration, that's good, that's bad. We actually need to map that and we can connect it to outcomes that we care about to really know whether or not it's good or bad and what we might be able to do about it.
