Data analysis (and the a,i,s,c buttons)

Analysis of Data:

☼ Analysis>blocks and cutpoints will give you a quick and dirty way to see cut points- these are brokers.

☼ Analysis>k-cores will force the data into levels of connectedness where each k-core has more connections within its core than it does to rest of network. This is a way to get at clustering.

☼ Analysis> centrality self-explanatory. Values will show up in nodes pane.

☼ Analysis> structural holes. Constraint is the key measure of brokerage. Low means you are in more of a structural hole. Isolates are undefined.

☼ In node pane, if relevant, you can set the values to >, <, or = a given value. You can change views of data with “a,i,c,s”. These stand for all, inverse, step, and cumulative.

Views of Data

Views of Data:

☼ You can change node size, shape, and color.

☼ You can change the on-screen layout in the layout menu, or with the lighting bolt icon which places most central nodes in middle using an algorithm,

☼ You can manually change nodes by right clicking on any one node. Other node choices (like size by attrib) will override your manual choices.

☼ Layout>Emily’s Chart allows you to play with “slices” of the net by various combinations of attributes and to control whether you only look at egos or egos and alters. If you close the window, Emily’s chart will cut some nodes, so use with caution (save as a new file).

☼ Layout> egonets allows you to see individual or multiple egonets.

☼ Layout> Move, recenter, resize are pretty self-explanatory.

Links for Netdraw Help

Zak asked me for links for tutorials. Such things exist, but they may vary a lot in how useful they are.

Analytic tech has some learning handouts on its web page.

There is a Netdraw manual.

There is also an online textbook for Netdraw and another program called UCINET. It may be useful.

Handling Data

Handling Data

☼ Make your OWN copy of the BIG networks.

☼ I added a subfolder called “BIG Nets for final paper.” It includes a file from the Fall of 06 class for comparative purposes.

☼ You can save anything on screen as its own graphic for pasting into papers, or as a .vna file for further analysis. Say you want only the network of everyone in a dorm. You can select those in the node pane. Then save as.

☼ Any analysis you do will get added to your node attribute file.

☼ You can cut and paste from the node attribute file into excel for easier data analysis. If you do so, the safest thing is to copy the WHOLE file and do not change the order of rows EVER. When you paste back, put the cursor in the upper left cell, but not so the cursor is blinking (click once in other words), and hit ctrl-v.

☼ The restore deleted nodes and lines will always bring you back from the brink. Save new versions of file for different purposes.

How to add a node in Netdraw AFTER VNA format

1) Open file
2) Click on the icon on far right that has a red dot. Now, in the map area, a click adds a node.
3) Repeat as necessary for number of nodes.
4) Open Node Attrib Editor [NAE].
5) Add names under ID.
6) IN NAE save to file and update graph.
7) Save as A NEW FILE NAME.VNA
8) Go to File>new and then file>open to open update vna file.

Questions and script

The primary question we agreed on is:

"Who are the people you feel close to at Bucknell (up to 15)? "

In addition, we are adding this attribute data
dorm
sports (none)
major (undecided)
hometown (city, state)
year (fresh, soph, jnr, snr)

When you interview someone, try this script:

"Hello. I am collecting information for a class research project about the social structure of Bucknell. Your participation is voluntary. Any information you give me will be kept in confidence. May I ask you question?

Who are the pople at Bucknell you feel closest to?"

(Stop them at 15)


For each one, ask if they know Dorm, Sports, Major, Hometown, Year. If they are not sure, leave it blank.

Procedures for files and questions

Here are specifics we agreed on:

1) You are each constructing your ego network based on this question:
Who are the people you feel close to at Bucknell (up to 15)?

In addition, you should connect any one you are connected to who you are reasonably sure would be on each other’s ego network generated from the same question.

2) Once your ego network is in Netdraw, you can add attribute information. Be sure your node attribute editor is set up with columns in this order. If you don't it will make it impossible to combine egonetworks:

ID
Dorm
Sports
Major
Hometown

3) To SAVE this data, you now have to save it as a VNA file (this saves attribute and tie data).

Here is a simple version of the procedure.

1) Download and open visone and netdraw (If analytictech downloads UCINET, open it and click on the colored box icon to get Netdraw).
2) Draw your network and add names.
3) Export the file as a .dl file. Go to file>export. In the window, make sure you have selected .dl files.
4) In netdraw, go to fie>open>UICENT dl text file>Network 1-mode. Click on the file browse button (three dots). Find your exported file.
5) To edit attributes, go to the menu transform>node attribute editor. A small spreadsheet should become visible. You have to insert columns from menu. You can cut and past a whole column to get them in the right order. It is a very crude spreadsheet.
6) After editing, be sure to go to the Node Attrib Editor’ menu and choose update graph and save to file. DO this before you exit or you will LOOSE your data.
7) When done, soave your file as a ‘vna file. From the file menu, choose save data as and pick .vna.

Remember, once you export from visone, you can not send a file back to visone for editing.

Make sure you save ties and attributes as the VNA file format or you will lose work.

Spring 2007

Hello Spring 07 students.

This blog is live. Please email me questions so we can post answers here.

Good online reosurces

Here is a good resource for netdraw.

http://faculty.ucr.edu/~hanneman/nettext/C4_netdraw.html

Check out locaction locaction, location

How do I get a legend for node size or color by attribute?

A quick and dirty way is to open the properties menu. Select the property you want to change by attribute (say dorm). While the dialog window is up, press ALT + PrINT SCREEN. The image is now on your clipboard. Open a word file. From edit menu, select paste. You should have an image of the window from netdraw.

How do I use K-Cores in Netdraw?

Peter-

Here is what it means:


A k-core is a maximal group of actors, all of whom are connected to some number (k) of other members of the group. To be included in a k-plex, an actor must be tied to all but k other actors in the group. The k-core approach is more relaxed, allowing actors to join the group if they are connected to k members, regardless of how many other members they may not be connected to. By varying the value of k (that is, how many members of the group do you have to be connected to), different pictures can emerge. K-cores can be (and usually are) more inclusive than k-plexes. And, as k becomes smaller, group sizes will increase.

NetDraw includes a tool for identifying and coloring a graph according to its K-cores. The UCINET algorithm for identifying K-cores is located at Network>Regions>K-Core.

In our example data, if we require that each member of a group have ties to 3 other members (a 3-core), a rather large central group of actors is identified {1,2,3,4,5,7,10}. Each of the seven members of this core has ties to at least three others. If we relax the criterion to require only two ties, actors 8 and 9 are added to the group (and 6 remains an isolate). If we require only one tie (really, the same thing as a component), all actors are connected.

The k-core definition is intuitively appealing for some applications. If an actor has ties to a sufficient number of members of a group, they may feel tied to that group -- even if they don't know many, or even most members. It may be that identity depends on connection, rather than on immersion in a sub-group.


This from: http://faculty.ucr.edu/~hanneman/nettext/C11_Cliques.html#kcore

In our large network a k-core with a high value means those people in a kcore of 5 are attached to at least five other people in a cluster whoa re also attached to at least five of the same cluster. Its looser than a clique definition which is very strict and hard to see at cliques of size 4,5,6. There can be different K-cores of 4,5,6 and so on that may or may not be connected.

What this means precisely will be up to you to interpret. In our data, the K core of six is zach evans an is basketball friends. K-cores of 2 may not be that interesting individually as there are many.

Does this help?

How do I use the ego stack file?

How do I use the ego stack file?

You need to use netdraw. Open the egostack file. Its extension will be .##h. It takes a minute to load.

You can decrease the size of the labels using the menu button with anumber in it and up and down arrows next to it. You can use the other arrows to play with node size. This should help.

Also, you can go to the layout menu and select egonets. You get a new window. You can hit clear and then start to assemble linked egonets by checking the box of a particular individual.

Links

The purpose of this blog is to provide one stop technicalhelp for my six degrees calss.

Two sites for help.

www.visone.info There is a link for documentation.

http://www.analytictech.com/Netdraw/netdraw.htm There is a link for the manual.

You cna email me questions and I will post them here.