An original methodology for the analysis and interpretation of word-count based methods: Multiple factor analysis for contingency tables complemented by consensual words

• We propose a new methodology for analyzing word-count based methods.
• The method preserves the description performed by all the panelists.
• A tool is proposed to identify consensual words to interpret a product map.

In sensory analysis, results from word-count based methods are customary analyzed through correspondence analysis applied to the global table products × words summing the citations of a same word given by all the panelists. This approach assumes that a same word mentioned by different panelists corresponds to a similar perception, which is not always verified. To solve this problem, we propose a new methodology based on multiple factor analysis for contingency tables. This methodology offers a mean configuration of the products taking into account all the individual words but spots these that are consensual to ease the interpretation. The consensual words have the same meaning for most of the consumers as far as they describe the same products. A test, based on resampling techniques, allows for assessing the significance of the consensus. A real example shows how this methodology eases the interpretation of the word-count based methods by solving problems arising from the large diversity of vocabulary and the different meanings possibly associated to a same word.