Fig. 5: Minimizing jingle–jangle fallacies.

a, The number of fallacies as a function of a number of clusters proposed by a specific clustering algorithm (hierarchical clustering using Ward linkage (HC ward)). The plot shows the trade-off between minimizing jingle and jangle fallacies as a function of the number of clusters and highlights both the most parsimonious solution that minimizes the number of clusters while reducing fallacies and the optimal solution that considers the minimization of fallacies only. The dashed line indicates the number fallacies produced by the original IPIP label assignments. b, The number of clusters and fallacies across various clustering algorithms. For each algorithm, the most parsimonious solutions are determined as those with the fewest fallacies using no more than 100 clusters and constructs (the arbitrary criterion for the maximum number of constructs used in the example). The optimal solutions are those with the fewest fallacies. Kmeans and Mclust refer to k-means and Gaussian mixture clustering, respectively. The numbers in parentheses indicate the number of fallacies produced by the parsimonious and optimal solutions, respectively. c, A radial plot indicating a possible mapping of 459 scales to 68 labels based on the parsimonious solution shown in a. Solid and dashed lines reflect assignments of optimal (top-five rank) and suboptimal similarity, respectively.