Analyzing Incomplete Longitudinal Binary Data Using Approximate Likelihood Methods
Bibliographic record
Abstract
In longitudinal studies, an outcome variable and a set of covariates are observed repeatedly on the same subject over time.Within-subject correlation results from repeated observations on the same subject, and in order to obtain valid estimates, any method that is used to analyze the longitudinal data, should consider this correlation.Also, missingness in the outcome variable is a common problem in longitudinal data.It complicates the analysis, especially when the missingness is nonignorable, i.e., when the missing mechanism depends on the unobserved values of the outcome variable.In this case, usual approaches to missing data including imputation-based techniques fail to obtain valid inferences.We need to define a joint distribution to account for the missing data model and the longitudinal response model, at the same time.In this thesis, we investigate two approximate likelihood methods, namely, bivariate pseudo-likelihood (BPL) proposed by Sinha et al. [1], and independent pseudolikelihood (IPL) proposed by Troxel et al. [2], along with the exact likelihood method, to analyze longitudinal data with nonignorable, nonmonotone missingness in the outcome variable.We use the marginal binary models, and we present results of a simulation study and results from an application of these three methods to the RAND HRS longitudinal data.Results from our simulation study shows that with nonignorable missingness, the BPL I would like to express my deepest appreciation to Professor Sanjoy K. Sinha for supervising me as his M.Sc.student, and providing me with encouragement and patience throughout the duration of this thesis.With his ongoing support and step-by-step guidance, I was able to conduct necessary research, and complete this work.Without his invaluable contribution, knowledge and experience, I had no chance to write this thesis or conduct accompanying research.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.030 | 0.119 |
| Meta-epidemiology (narrow) | 0.001 | 0.002 |
| Meta-epidemiology (broad) | 0.003 | 0.002 |
| Bibliometrics | 0.003 | 0.005 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.003 | 0.005 |
| Open science | 0.005 | 0.003 |
| Research integrity | 0.002 | 0.005 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".