Data, Textual

Textual data refer to systematically collected material consisting of written, printed, or electronically published words, typically either purposefully written or transcribed from speech. Text collected for use as data typically reflects a conscious research purpose, motivated by a design aimed at yielding insight on some feature of the social or political world. This entry outlines the purpose, issues, and challenges involved in selecting, preparing, and analyzing textual data.

The process of textual data analysis follows several main steps. First, a text and its author are identified that will directly inform the research question at hand, for instance, legislative speeches if one is interested in lawmakers' policy agenda. Second, this text must be processed, typically involving numerous decisions about how to convert, store, edit, and combine texts. Next, the processed text is analyzed, a step that may take any of a wide variety of forms. Humans may read the text to classify it, possibly after dividing texts into smaller units; computerized tools may analyze the text to count frequencies of preidentified words or phrases; statistical methods may be applied to specific words, word patterns, or sentences to construct scales, clusters, or to identify other patterns. Finally, the results of the analysis are summarized, interpreted, and reported to inform the research question that motivated the analysis of the textual data.

The steps in this process are not dissimilar to those found in any empirical research design, but the systematic use of textual data may present special challenges in the stages related to identifying the data to be collected and preparing the data for analysis. In the modern era of near-ubiquitous electronic textual content, this problem stems not from a lack of availability of textual data or tools for working with them but, rather, from the opposite: The staggering quantity and variety of textual data create real challenges in selecting which texts to analyze. Furthermore, the technical challenges of working with different formats for recording electronic texts, as well as converting these into formats or “data sets” that suit the purpose of analysis, are often far from trivial. Despite such challenges, however, textual data remain one of the most promising and one of the least explored sources of systematic information about the political and social world.