Abstract:<span>We propose a framework for inference based on an "iterative likelihood function," which provides a unified representation for a number of iterative approaches, including the EM algorithm and the generalized estimating equations (GEEs). The parameters are decoupled to facilitate construction of the inference vehicle, to simplify computation, or to ensure robustness to model misspecification and then recoupled to retain their original interpretations. For simplicity, throughout the paper, we will refer to the log-likelihood as the "likelihood." We define the global, local, and stationary estimates of an iterative likelihood and, correspondingly, the global, local, and stationary attraction points of the expected iterative likelihood. Asymptotic properties of the global, local, and stationary estimates are derived under certain assumptions. An iterative likelihood is usually constructed such that the true value of the parameter is a point of attraction of the expected log-likelihood. Often, one can only verify that the true value of the parameter is a local or stationary attraction, but not a global attraction. We show that when the true value of the parameter is a global attraction, any global estimate is consistent and asymptotically normal; when the true value is a local or stationary attraction, there exists a local or stationary estimate that is consistent and asymptotically normal, with a probability tending to 1. The behavior of the estimates under a misspecified model is also discussed. Our methodology is illustrated with three examples: (i) estimation of the treatment group difference in the level of censored HIV RNA viral load from an AIDS clinical trial; (ii) analysis of the relationship between forced expiratory volume and height in girls from a longitudinal pulmonary function study; and (iii) investigation of the impact of smoking on lung cancer in the presence of DNA adducts. Two additional examples are in the <a class="ext-link" href="https://doi.org/10.1080/10618600.2021.1904961">supplementary materials</a>, GEEs with missing covariates and an unweighted estimator for big data with subsampling. Supplementary files for this article are available online.</span>

Advancing Information Integration through Empirical Likelihood: Selective Reviews and a New Idea

A Selective Review on Calibration Information from Similar Studies Based on Parametric Likelihood or Empirical Likelihood

Empirical Likelihood Using External Summary Information

Integrating external summary information in the presence of prior probability shift: an application to assessing essential hypertension

Prior Information-Assisted Integrative Analysis of Multiple Datasets.

Individual Data Protected Integrative Regression Analysis of High-dimensional Heterogeneous Data

Privacy Preserving Integrative Regression Analysis of High-dimensional Heterogeneous Data

Integrative analysis of individual-level data and high-dimensional summary statistics

Iterative Likelihood: A Unified Inference Tool

Empirical likelihood for the analysis of experimental designs

Assumption-Lean Post-Integrated Inference with Negative Control Outcomes

Integrative High Dimensional Multiple Testing with Heterogeneity under Data Sharing Constraints

Lossless integration of multiple electronic health records for identifying pleiotropy using summary statistics

A Robust Consistent Information Criterion for Model Selection Based on Empirical Likelihood

Bi-linear Matrix-Variate Analyses, Integrative Hypothesis Tests, and Case-Control Studies

A Review on Empirical Likelihood Methods for Regression

Machine learning for integrating data in biology and medicine: Principles, practice, and opportunities

Bias and Unfairness in Information Retrieval Systems: New Challenges in the LLM Era

Efficient Integration of Aggregate Data and Individual Patient Data in One-Way Mixed Models

Machine Learning and Integrative Analysis of Biomedical Big Data

Data Integration in Causal Inference