Very helpful short overview of data theory and dimensional analysis techniques for the social sciences. Rather than detailing how to perform each, Jacoby gives some overarching frameworks, describes how each method fits in and how it relates to other methods, and gives some heads-ups and gotchas for working with each methodology. It also has incredibly helpful diagrams. This would be a useful book to read completely through before planning a multivariate data analysis project (e.g., on psychological item-response data).

A lay overview of this book's topic: We often collect data on rows representing items and columns representing variables, with each cell value containing the value for that item on that variable. Sample data could be a set of regular shapes on which we collect a set of variables. When we have 2 variables (like area and perimeter), then we call the collected data "2-dimensional" -- the data could be visualized and understood using a 2-D graphic in which each axis represents a different variable and a marker appears at the position of each item. When have have 3 variables (like area, perimeter, and number of sides), we can still do the same visualization and conceptualization; however, now we have to work in 3-D space. Most of the time when we collect data, however, we collect on many more variables than just 2 or 3 -- this book describes approaches to deal with the higher dimensionality of common data.

This book describes, compares, and contrasts the following techniques for working with and scaling data to make it useful for particular research questions:
-- data types (nominal, ordinal, interval, ratio)a
-- assigning meaningful values to non-interval/ratio data
-- cumulative scales (e.g., Guttman scaling analysis)
-- factor analysis
-- paired comparisons
-- psychophysical magnitude scaling
-- multidimensional scaling (classical, asymmetric, weighted)
-- profile distances
-- multidimensional unfolding
-- alternating least squares, optimal scaling approach (ALSOS) ( )