Resumen de Analyzing likert-type data using penalized non-linear principal components analysis
We consider a survey on animal ethics and sustainability consisting of various Likert-type items. Although this kind of (ordinal) data often occurs in the social sciences, in case of principal components analysis (PCA) those data are either treated as numeric implying linear relationships between the variables at hand, or nonlinear PCA is applied where the obtained coecients are sometimes hard to interpret. We therefore revisit penalized nonlinear PCA for ordinal variables as an intermediate between the mentioned methods used so far. The new approach o ers both better interpretability as well as better performance on validation data.
