Principal component analysis of narrative corpus: A mixed-method design to obliterate the pitfalls of vignette-based analysis of personal narratives

Abstract

Social scientists engaged in qualitative research rely heavily on the narratives shared by their respondents to attribute meaning to a phenomenon. The authors believe that the prevailing practice of fragmenting long narratives into short vignettes during the analysis of narratives deprives the researchers from capturing the full meaning of the narrative, further leading to inferior analysis. The present study describes various pitfalls arising out of vignette based analyses of narratives and suggests an alternative mixed method. The proposed design combines the qualitative rigor of thematic analysis with the graphical explanatory power of nonlinear principal component analysis. The proposed design preserves the respondent stories in their entirety during classifying the codes to themes and while interpreting each theme in detail.