MultiNet

MultiNet is a data analysis package that can be used for ordinary data (in which you have a file that has one line of data for each case) and for network data (in which there are two files -- the "index" or "node" file describes the individuals and the "link" file describes the connections between individuals).

 


New Version

MultiNet 2.11 (2000 version, updated Dec 29, 1999) is now available.


Click here for more information on the 2000 version.


MultiNet runs in DOS, Windows 3.1, Windows NT 4.0, and OS/2. You can also run MultiNet in Windows 95 in "real DOS". To run the program in DOS, you must turn EMM386.EXE (and probably other extended memory managers) off.

Click here to see what makes MultiNet different from other packages.

Click here to see information about the format of the data MultiNet uses.

Click on the little pictures to see more details.

.

little picture

Click on the above picture to see what it is like to use MultiNet to perform an analysis.

For ordinary analysis, the program does univariate (mean, mode, median, standard deviation, etc.) and bivariate statistics (crosstabs, chi-square, ANOVA, correlation, and regression).

little picture

For network analysis, it does all of the above, but it allows you to combine individuals and connections between individuals into a single analytic model.

little picture

The program does two kinds of network analysis. The first is one that looks at ego-centric data -- data in which the links of each node in the sample is described, but in which there is no expectation that the nodes listed as contacts are also in the sample. The second kind of network analysis looks at ordinary network data, in which a set of nodes and the contacts they have with one another are described by the data.

little picture

With the second type of data, it is possible to analyze the structure of the network with eigen methods that allow, among other things, clusterings, colorings, and other forms of structural equivalence to be determined. The results of eigen analyses may be used to create new index variables which may be used in further analysis, creating a range of hybrid types of analysis.

little picture

The program is menu-driven. It has context-sensitive on-line help. It always shows a graphical representation of the results of whatever it is doing. You can switch between graphics and ordinary text by pressing a single key. It produces color PostScript or b/w Epson-FX output files (for graphics) or ordinary ASCII files (for text). It has a wide range of data transformation and recoding capabilities that greatly simplify many of the tasks one encounters in the course of analysis.

little picture

 

 


Click here to download a copy of MultiNet for 2000. (c) 2000, William D. Richards.

Some features of the version of MultiNet you may have downloaded or received on a disk in 1999 or earlier will not work past the end of 1999. You should download the program again by clicking above (MultiNet for 2000).

Click here to download a copy of unzipg.exe, which you may need to unzip the MultiNet package.

 

In the works...

  • sparse matrix methods for eigen decomposition will increase the size of networks that can be examined with these methods;
  • ability to save ASCII files that contain any new variables created by recoding, partitioning, or defining;
  • ability to import space-, tab-, or comma-delimited files as well as the fixed-format approach now used;
  • ability to import UCINET or KRACKPLOT files;
  • the dot matrix output option will be replaced with .bmp or .gif format for graphic output;
  • ability to perform multidimensional scaling methods on data other than adjacency matrices;
  • multivariate ANOVA and regression analysis;
  • graph theoretic measures
  • more
  • Go back.