FAQ's:

What is the difference between network similarity and “C” that I have read about?

What is the Coherence measure and when is it appropriate?

How is the correlation of proximity data accomplished?

Are there site licenses?

What are the known bugs?

Can you give more help on installing the software?

The following are no longer issues in the new windows version of the software.

How can I save the networks for printing or for including in other documents?

How can I run PCKNOT from any folder (directory)?

What if I have problems launching the GUI?

How can I save the networks for printing or for including in other documents?

This is not a problem in the new windows version. You can save network pictures in a variety of formats. For the old version, the best method is to use a screen capture program to place the contents of the screen in a file with the appropriate format. There is a good screen capture program that will save the network layout screens in files of various formats. Check out SnagIt at this url: http://www.techsmith.com/

How can I run PCKNOT from any folder (directory)?

In the new version, you select the working folder after starting the program so it is not a problem. In the old version, you must include the path to the PCKNOT software in the PATH environment variable:

Under Start, Select "Control Panel"

Open "System"

On the "Advanced" tab select "Environment Variables"

Edit the Path variable by adding a semicolon and the path to PCKNOT

For some reason, some machines won't properly launch PCKNOT unless you start it from the command prompt. To do this follow the instructions below.

What if I have problems launching the GUI?

For the old version, you can try starting KNOT from a command line prompt:

Under Start, select Run and enter “command”

In the window that appears, enter:

cd \pcknot (assuming you have the program files in the folder: C:\pcknot) then enter:

pcknot

This should start the GUI in a full screen.

What is the difference between network similarity and “C” that I have read about?

Goldsmith and his colleagues originally developed a measure of network similarity that they called C (for Closeness). This measure involved computing the similarity of the neighborhood for each node and then averaging this measure across nodes. Unfortunately, the probability distribution for this measure has not been discovered. So the similarity measure used in the KNOT software was developed. This measure is the ratio of the number of links shared by two networks over the number of links found in either of the two networks (the cardinality of the intersection of the links in the two networks over the cardinality of the union of the links in the two networks). This measure follows the hypergeometric probability distribution. The KNOT software computes values from this distribution to provide information concerning the similarity expected by chance.

What is the Coherence measure and when is it appropriate?

The measure is based on a kind of transitivity assumption, i.e., if two concepts have similar relationships with other concepts, then the two concepts should be similar to one another. Now we know that transitivity does not necessarily hold for all sets of concepts, but failure of transitivity is the exception while transitivity is the rule. The coherence measure then computes an indirect measure of similarity by correlating the ratings given for each item in a pair with all of the other concepts -- if we have 5 concepts (ABCDE) then for the pair AB we would correlate the ratings:

AC BC

AD BD

AE BE

which gives the indirect similarity of AB - the extent to which A and B have similar relationships with other concepts. If we do this for all pairs, we can construct a half-matrix of indirect similarities. The coherence is the correlation between these indirect measures and the original ratings given for each pair of items. A more consistent set of ratings will yield a higher coherence. One data set came from the study of physics expertise where the coherence increased monotonically with level of expertise. Extremely low coherence (less than .20) may indicate that the rater did not take the task seriously.

Correlation of Proximity Data

The correlation between proximity data files is accomplished by simply correlating the corresponding proximity data entries in the two files. Because missing (or infinite, out of range) data in either member of a pair of proximities leads to dropping the pair from the correlation, the measure may not be very good for data with a large proportion of missing proximities.

Pearson correlation does not require ratio scale data, and it is common to use the measure on rating data as we often have with proximity data.

Are there site licenses?

The following fees are for site licenses for KNOT software for Pathfinder Network analysis. Documentation is available within the software.

1 to 2 users : $300

3 to 5 users : $600

6 to 10 users : $1,000

11 to 20 users : $1,500

21 or more users : $2,000

20% discount on software for educational institutions

Can you give more help on installing the software?

The new windows version of the software does not usually present any difficult installation problems. If you encounter the message: could not find file ‘mclmcrrt77.dll’ it can be due to a faulty installation. Installing again can fix the problem. In a reinstallation, select the "repair" option.

If reinstallation fails you can try:
1. Open My Computer
2. Double Click: Local Disk (C:)
3. Right click: autoexec.bat (or autoexec), then select "edit"
4. add the following line as the last line of the file:
PATH = %PATH%;C:\Program Files\MATLAB\MATLAB Component Runtime\v77\runtime\win32
5. Restart your computer

If you encounter an error with "VDIST" in it, try installing Microsoft Visual C++ before installing Pathfinder (Get Microsoft Visual C++ here)

The old version can present some dificulties as follows:

PCKNOT software was originally designed to run under DOS. It creates its own Windows and handles all of the user interactions directly. It uses and creates several different files. Although it was designed for DOS, it runs just fine under Windows 3.11, NT, 95, 98, 2000, and XP.. To work properly however, a few DOS level things need to be in place. (1) The directory containing the PCKNOT files must be in the PATH variable in the DOS environment variables, and (2) A mouse driver must be installed in DOS.

In all probability you will not need to worry about the mouse (#2 above). Windows will have taken care of this.

Here are the steps involved in installing the software in Windows:

1. open My Computer (on the desktop)
2. open (C:)
3. Under the File menu item select "New" then "Folder"
4. Select the "New Folder" which was created and change the name to "PCKNOT"
5. copy the "KNOTFILE.EXE" file into that folder
6. Using the Start Menu, select RUN
7. enter c:\pcknot\knotfile and click ok
8. open the pcknot folder
9. right click the pcknot.bat item then select send to .. desktop (create shortcut)
now you will be able to start pcknot from your desktop
10. using notepad, open the file, c:\autoexec.bat
11. add the following line as the last line of the file:
PATH %PATH%;C:\PCKNOT
12. save the file
13. you should now be able to use pcknot

Known Bugs:

The command line layout display program (only in the old version). When you save the network from the display screen, it changes the order of some of the links, reversing their directionality. Essentially, this layout program is only appropriate for nondirected networks. The GUI display is correct, and it handles directed links correctly. If you have nonsymmetric data, just be careful not to save networks viewed by the command line display. If you rely on the GUI display version, you will fine.

The Coherence measure is not correct for asymmetrical (or nonsymmetrical) data. In fact, we cannot think of a good way to compute coherence for such data - at least using the methods we use for symmetrical data. The computation performed only considers the out (from A to B ) data to compute coherence. This is clearly wrong, but we do not know what a correct method would be for asymmetrical data. With nonsymmetrical data, you could consider producing a symmetric version of your data files for computing coherence (e.g., you could take the maximum similarity for AB and BA to represent the AB pair). The software can do this for you.