From a Classification to Another

In Brazil, the statistic variable for skin color is collected through declarations. Interviewers can offer different alternatives among which the respondents choose a word. The most used is the classification of the Brazilian Institute of Geography and Statistics. This classification is sometimes criticized. Thus several researchers studied the distribution of skin color declarations toward various classifications. Their results underlined there is a stability of these declarations. So they conclude classifications are mainly equivalent. However, this stability hides reclassification processes. Indeed, there are two sources of changing words between two sets of alternatives. First, the term initially chosen is not proposed anymore. Second, the respondent can find another term more suitable, even if the term initially chosen is still proposed. Moreover, sometimes the new term was already proposed in the initial classification. I propose to focus on the individuals who are reclassifying themselves in order to apprehend if such a decision is correlated to given attributes and / or characteristics. Whithout rejecting a global stability of declarations, I conclude that skin color reclassifications exist from one classification to another. Personnal attributes and characteristics influence these reclassifications. This can lead first to nuance the stability of skin color declarations. Second, it has to be considered in quantitative applications because this can induce variability in results when using a classification or another. Presented during RC25 session "Hegemonies in Classification Processes"