Inductive Codebook Development for Content Analysis: Combining Automated and Manual Methods

Abstract

At the core of every content analysis is a codebook of relevant categories, frequently developed qualitatively based on a small sample of texts. Currently text mining methods enable us to explore an almost unlimited number of texts in an efficient, fast, and comprehensible manner. In this article, we suggest a procedure for codebook development using these methods to inductively derive coding categories from a large text corpus for content analysis. These methods are combined with qualitative, manual content analysis. First, we derive thematic main categories from a text corpus via text mining. In a next step, we then manually validate these categories and add sub-categories via qualitative content analysis. The method is exemplified with a codebook that was developed for the analysis of the citizen dialog on the "Quality of Life in Germany" [Gut leben in Deutschland], an open-ended questionnaire initiated by the German government to gather citizens' opinions on important aspects of quality of life.