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.

Influence of Attributes Relationship to Parameter Estimation in DINA

Atomic dynamic functional interaction patterns for characterization of ADHD.

Determining the number of attributes in the GDINA model

An Improved Parameter-Estimating Method in Bayesian Networks Applied for Cognitive Diagnosis Assessment

A Semi-supervised Learning Method for Q-Matrix Specification Under the DINA and DINO Model With Independent Structure

BNMI-DINA: A Bayesian Cognitive Diagnosis Model for Enhanced Personalized Learning

Application of Bi-factor MIRT and Higher-order CDM Models to an In-house EFL Listening Test for Diagnostic Purposes

Bayesian Analysis of Longitudinal Dyadic Data with Informative Missing Data Using a Dyadic Shared-Parameter Model

Statistical Analysis Ofq-Matrix Based Diagnostic Classification Models

The Impact of Item Model Parameter Variations on Person Parameter Estimation in Computerized Adaptive Testing With Automatically Generated Items

Sufficient and Necessary Conditions for the Identifiability of DINA Models with Polytomous Responses

Modified Cognitive Diagnostic Index and Modified Attribute-Level Discrimination Index for Test Construction

A Testlet Diagnostic Classification Model with Attribute Hierarchies

Integrative Data Analysis Where Partial Covariates Have Complex Nonlinear Effects by Using Summary Information from an External Data

Does the reliability of computational models truly improve with hierarchical modeling? Some recommendations and considerations for the assessment of model parameter reliability

DIF Statistical Inference Without Knowing Anchoring Items

Diagnostic Classification Models for Testlets: Methods and Theory

Influence Diagnostics in Meta-Regression Model

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

Cognitive Diagnostic Model Made More Practical by Genetic Algorithm.

A Semi-parametric Estimation of Personalized Dose-response Function Using Instrumental Variables.