Benchmarking the Double-Entry Intraclass Correlation as an Index of the Similarity between Correlation Profiles
Year: 2026
Type: preprint
Abstract: To infer whether the test scores subjected to analyses tap into the same construct (or opposite ones), measured as the similarity between sets of correlations with other variables, researchers often use the Double-Entry Intraclass Correlation (ICCDE). Importantly, the ICCDE has been developed as an index of the similarity between profiles consisting of unbounded coefficients (e.g., profiles of personality features), whereas the correlation metric conveys a series of challenges to the applicability of the ICCDE. Using simulation studies, this research benchmarked the applicability of the ICCDE to correlation profiles under various conditions regarding characteristics of the correlation profiles subjected to analysis. More specifically, equality versus inequality of the patterns of correlation trajectories (shape), levels of variance of the compared correlation profiles (scatter), and mean levels of the profiles (elevation) were manipulated along with the level of variance within the correlation profiles. Simulation results indicate that ICCDE values can be strongly influenced by profile shape and restricted within-profile variance. These findings suggest that conclusions about similarity between correlation profiles may depend on characteristics of the compared profiles and have important implications for evaluating jingle and jangle fallacies.
Source: PsyArXiv (OSF Preprints)
Author: Christian Blötner
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7202109723
