MultiNet is a Windows-based computer program designed for exploratory data analysis of social and other networks. MultiNet is highly interactive and always provides both textual and visual representations of results. The visualizations are innovative in the use of colour and interaction, and some are unique to MultiNet. MultiNet was designed from the beginning to handle large amounts of data, and uses compact data formats, special storage schemes, and calculation methods that are highly efficient in terms of both space and time. MultiNet was also designed to handle large numbers of variables, both attribute (node) and network (link); it allows easy construction of new variables of either type by means of various operations on existing ones. Hybrid variables are easily constructed: node variables derived fiom networks; link variables derived fiom attributes. These capabilities provide crucial links among other parts of the program. The application of spectral methods to large, sparse networks is both the theoretical and practical centre of the research and development that has gone into MultiNet. Spectral methods provide analytic visualizations of network data: pictures that not only provide understanding, but that provide numerical values that can be used in further analysis. The results of the spectral methods, as well as other attribute and network data, are used together with simple, standard statistical methods such as cross-tabulations, analysis of variance and correlations for testing hypotheses about relationships among the data. MultiNet provides unique methods that allow attributes and networks to be freely mixed in such analyses, and presents results in both textual and interactive visualizations that include two or three discrete or continuous variables. The largest part of this thesis consists of descriptions of the seven main MultiNet program modules. Supplementary sections describe the theoretical background for spectral analysis and provide specific examples of spectral analysis, including a peer-reviewed, published paper that uses most of the parts of MultiNet together. In addition, a separate CDROM provides a working version ofthe program, electronic documentation, sample datasets, software aids and videos showing how the program is used.
The author has placed restrictions on the PDF copy of this thesis. The PDF is not printable nor copyable. If you would like the SFU Library to attempt to contact the author to get permission to print a copy, please email your request to firstname.lastname@example.org.