Contrast Analysis of Semantic Similarity (CASS)
with simple text-analytic tools for social scientists
INTRODUCTION
In the social sciences, it is useful to understand the relative similarities of concepts that are embedded in a particular text (from a particular group or a particular person). For example, in trying to estimate conservative bias in FoxNews, one might estimate its tendency to associate conservative concepts (conservative, republican) and good concepts (good, positive, etc.), compared to conservative and bad concepts. The output would indicate conservative favoritism. This comparison could be further refined by taking into account important "baseline" information about the valences associated with liberal, namely liberal and good in comparison to liberal and bad. The following set of comparisons results:
[(conservative & good) - (conservative & bad)] - [(liberal & good) - (liberal & bad)]
The output from this equation would allow one to determine the relative bias in the news station. Technically, it is comparing four higher-order concepts gleaned from semantic space--models (e.g., BEAGLE) that group semantically similar words and allow one to run CASS easily. Such analyses could be applied to a wide range of topics.
In its computation, it is quite similar to other methods, such as the Implicit Association Test. Like the IAT, CASS could be used to study self-concept, self-esteem, attitudes, and stereotypes.
To start your own CASS studies or to try it out using some sample text that we supply, please read the documentation here.
RESEARCH and SOFTWARE DEVELOPMENT
CASStools.org is based on a paper (under review) by Nick Holtzman, John Paul Schott, Mike N. Jones (creator of BEAGLE), Dave Balota, and Tal Yarkoni. The software was written and developed by Tal and Nick, so questions concerning this site should be directed to them.
Last updated 16 June 2010