SPSS Text Analysis for Surveys™ (v. 2.0)

Product Overview SPSS Text Analysis for SurveysTM(or STAFS) partly automates the process of categorizing or coding responses to openended survey questions. Several methods for automating this process have been developed by software vendors. Common approaches are based on statistical algorithms, neural networks, and other techniques that include specific rulebased processes. The down side to these techniques is that they require considerable expertise to use, and the underlying technologies are unavailable to most users. Perhaps more important, the accuracy of analyzing text using these particular methods is fairly low. Automated linguistics-based solutions to text analysis, the second most popular approach, considers both grammatical structures and meaning. This method relies on naturallanguage processing (NLP) or computationallinguistics. Linguistics-based text mining typically results in more reliable and useful results. STAFS is based on linguistic technologies, although it offers a separate statistics-based method as well. In a nutshell, this product allows the user to import text responses, extract terms or concepts, create categories and assign responses to these categories, and export results either as dichotomies or as categories for use in quantitative analysis. So how does STAFS do all these wonderful things? I spent over an hour with a very helpful product manager at SPSS, who gave REVIEWED BY JOHN DEATON