Abstract:Partial association refers to the relationship between variables <math>Y1,Y2,…,YK</math> while adjusting for a set of covariates <math>X={X1,…,Xp}</math>. To assess such an association when Yk's are recorded on ordinal scales, a classical approach is to use partial correlation between the latent continuous variables. This so-called polychoric correlation is inadequate, as it requires multivariate normality and it only reflects a linear association. We propose a new framework for studying ordinal-ordinal partial association by using Liu-Zhang's surrogate residuals. We justify that conditional on <math>X</math>, Yk, and Yl are independent if and only if their corresponding surrogate residual variables are independent. Based on this result, we develop a general measure <math>ϕ</math> to quantify association strength. As opposed to polychoric correlation, <math>ϕ</math> does not rely on normality or models with the probit link, but instead it broadly applies to models with any link functions. It can capture a nonlinear or even nonmonotonic association. Moreover, the measure <math>ϕ</math> gives rise to a general procedure for testing the hypothesis of partial independence. Our framework also permits visualization tools, such as partial regression plots and three-dimensional P-P plots, to examine the association structure, which is otherwise unfeasible for ordinal data. We stress that the whole set of tools (measures, p-values, and graphics) is developed within a single unified framework, which allows a coherent inference. The analyses of the National Election Study (K = 5) and Big Five Personality Traits (K = 50) demonstrate that our framework leads to a much fuller assessment of partial association and yields deeper insights for domain researchers. <a class="ext-link" href="https://doi.org/10.1080/01621459.2020.1796394">Supplementary materials</a> for this article are available online.

Qualitative Evaluation of Associations by the Transitivity of the Association Signs

Some association measures and their collapsibility

Assessing Partial Association Between Ordinal Variables: Quantification, Visualization, and Hypothesis Testing

Density-Based Mining of Quantitative Association Rules

Discovering Linguistic Associations from Numerical Data 1

Visual Analysis of User-Driven Association Rule Mining

Measuring Association on Topological Spaces Using Kernels and Geometric Graphs

INTEREST MEASURE--ANOTHER THRESHOLD IN ASSOCIATION RULES

Improving Associative Classification by Incorporating Novel Interestingness Measures

Quantitative Association Analysis Using Tree Hierarchies

Detecting Direct Associations in a Network by Information Theoretic Approaches

A Generalized Attributes Significance Measure Method

Comment: the Essential Tension Between Parsimony and Accuracy

Fuzzy Modeling Discovering Linguistic Associations from Numerical Data

Quantitative Criterion for Mining Both Positive and Negative Association Rules

Information-Theoretic Measures on Lattices for High-Order Interactions

Quantifying Direct Dependencies in Biological Networks by Multiscale Association Analysis

Beyond intratransaction association analysis

Capturing Tie Strength with Algebraic Topology

Mining pure linguistic associations from numerical data

A Generalized Model for Multidimensional Intransitivity