Association Analyses in MerQur: From Correlation to Canonical Correlation, from Bland-Altman to VARCLUS

Abstract

Association analyses are a family of methods that describe the pattern, direction, and strength of the relationship between two or more variables. From classic Pearson correlation to Bland-Altman analysis evaluating method agreement, from effect size computations to multivariate canonical correlation and variable clustering, this family forms the foundation of both exploratory data analysis (EDA) and hypothesis testing in academic research. This study introduces in detail the 6 analyses offered in MerQur’s Association category: Correlation Analysis (Pearson/Spearman/Kendall), Bland-Altman analysis (limits of agreement), Effect Size calculator (Cohen’s d, Hedges’ g, Glass’s Δ, Cramér’s V, Cohen’s w, Cohen’s f), Canonical Correlation Analysis (CCA), Correspondence Analysis (CA), and Variable Clustering (VARCLUS). For each analysis, this review presents: (i) the type of association tested and the application context, (ii) required assumptions and data types, (iii) form fields and options in MerQur, (iv) reported statistics and visualisation outputs, and (v) an interpretation guide for a typical research question. From Pearson’s measure of linear relationship to Kendall’s tau-b’s rank-pair-based approach for ordinal data, from the agreement plots that Bland-Altman generates for clinical method comparison to CCA’s capture of the maximal linear relationship between two variable sets, from Correspondence Analysis’s geometric representation of categorical contingency tables to VARCLUS’s hierarchical clustering of highly correlated variables — the scope is a summary of the powerful repertoire MerQur offers in association analysis.