Social networking tools can be a great way to visualise and find relationships in your data. It also can be a good way to present information to your readers, Jennifer LaFleur looks at the tools available to help you do this.
There are several programs that do social networking, here we look at UCINET and the drawing program NetDraw.
get your data ready
The first spreadsheet is the one we used at ProPublica to do our online social network of bank executives who were connected the Treasury Secretary Henry Paulson. But what we got was names and companies:
But UCINet requires the data to be in a matrix like this:
So we need to use Excel’s pivot table tool to create a matrix. To do this, in the spreadsheet, highlight the columns containing the Last names and the companies, including the titles at the top. The go to ‘insert pivot table’
You’ll get a blank spreadsheet with a list a list of fields to the right:
What we want is names down the side (row labels) and companies along the top (column labels). Drag ‘lname’ to the box under ‘row labels’ and drag ‘company’ to the box under ‘column labels. We still need it to fill in values, so drag ‘company’ to the box under ‘values’.
What you get is ones where the ‘fullname’ and ‘company’ match, but blanks everywhere else. UCINET needs those blanks to be zeros, so we need to change a setting in Excel. Under ‘pivot table options’, under ‘format’, check the box that says ‘for empty cells show’ and type 0 in the box next to it. Excel should fill all your blanks with zeros.
Now we can our data in UCINET. Copy the headers and the data items from your pivot table – not the totals.
Open UCINET and go to ‘ data’ ‘spreadsheets’ ‘matrix’. Click in the upper left corner of the grid and paste your spreadsheet:
Then choose ‘file’ and ‘save as’.
Now is when the fun begins, we’re going to flip over to another tool called Netdraw. Click on the network icon at the top of UCINET.
Once you get to Netdraw, go to ‘file’ ‘UCINET dadaset’ ’2-mode network’.
You basically will work with two types of networks: people to people (Network) and people to organisation (2-mode network).
You can click this button to redraw the network if it doesn’t seem clear at first.
There are many tools in both UCINET and NETDRAW that experts in network analysis use regularly. Below are just a few.
cleaning up your network.
If you’re not interested in those entities with no connections or with separate connections, you can get rid of them two ways.
This button removes “isolates” or individual nodes that are not connected to anything else.
This button removes “pendants” or nodes connected by only one line.
If you want to restore anything you delete, on the main menu, go to ‘transform’ ‘restore/delete’ ‘restore all deletions’.
measures of centrality
There are several measures that will tell you more about your data.
Get there by going to ‘analysis’ ‘centrality measures’.
You can add all of the measures to your data or you can redesign your network based on one of the measures.
closeness
How fast a person can get into contact with the rest of the network
betweenness
Shows to what degree a person is a link between two subgroups in the network.
degree
Shows how many other people in the network an individual is linked to. The higher the number, the more central the person.
and now for the bad news…
While this is a great tool for visualising relationships, it doesn’t have great output options. In most cases, I’ve had to have graphics folks recreate data based on what I did in Netdraw and UCINET.
For more information, go to www.ire.org/sna
social network analysis – Jennifer LaFleur, ProPublica – July 2009