This week we take on two different topics, although in a deep sense they are ultimately one thing. The first is about visualization of a pattern of correlations (or other proximities) using multidimensional scaling. We also cover correspondence analysis and network diagrams. The second is about constructing scales and indices -- scrunching a bunch of measurements into one. I think of scales as a particular kind of index in which all of the component items "point" toward the same thing. Indices in general can incorporate unrelated components. For example, we could construct an index of national well-being that combines economic indicators, technological level, health statistics, and so on.
Note that in this class we will introduce UCINET, so please have it downloaded and installed.
Primary Readings
Swenson. Building an additive scale [html]
Borgatti
Trochim. On Scaling. Read through the various links.
Datasets
Tutorial
What we did in class today [docx]