A vector space method to quantify agreement in qualitative data.
Information Storage and Retrieval
Natural Language Processing
Pattern Recognition, Automated
Terminology as Topic
Interrater agreement in qualitative research is rarely quantified. We present a new method for assessing interrater agreement in the coding of focus group transcripts, based on vector space methods. We also demonstrate similarities between this vector method and two previously published interrater agreement methods. Using these methods, we showed that interrater agreement for the qualitative data was quite low, attributable in part to the subjective nature of the codes and in part to the very large number of possible codes. These methods of assessing interrater agreement have the potential to be useful in determining and improving reliability of qualitative codings.