Mixing quantitative and qualitative content analysis: triangulation at work

The paper discusses an original path model for triangulating the results of three types of content analysis: (i) the analysis of the co-occurrence of words; (ii) the substitution model of quantitative content analysis and (iii) qualitative coding. The model refers to the "within method" as well as the "between methods" of triangulation. It shows how this model helps assess the reliability and validity of content analysis. In the former case, the text is used as a unit of analysis; in the latter-the variable (theme, category or qualitative code). The model is tested on two sets of transcripts of semi-structured (N = 64) and unstructured (N = 43) interviews carried out independently by two research teams. Outcomes of the test show consistent patterns in both cases. Some directions for further explorations are discussed, including latent qualitative analysis.