A pairwise pseudo-likelihood approach for regression analysis of doubly truncated data

Area of Science:

• Statistics
• Epidemiology
• Astronomy
• Economics

Background:

• Double truncation, involving both left and right censoring, presents significant analytical challenges.
• Existing methods for analyzing doubly truncated data are limited and often inefficient.
• Conditional analysis is a common but potentially suboptimal approach.

Purpose of the Study:

• To develop a more efficient statistical method for analyzing doubly truncated data.
• To recover information often lost in standard conditional analysis techniques.
• To improve estimation accuracy in the presence of double truncation.

Main Methods:

• Proposed a novel pairwise pseudo-likelihood approach.
• Developed an estimator for doubly truncated data.
• Evaluated the method through an extensive simulation study.

Main Results:

• The proposed pairwise pseudo-likelihood estimator is consistent and asymptotically normal.
• Simulation studies demonstrated the practical effectiveness of the new method.
• The new method showed superior efficiency compared to the conditional analysis approach.

Conclusions:

• The pairwise pseudo-likelihood method provides a more efficient alternative for analyzing doubly truncated data.
• The methodology is applicable to real-world studies, such as the AIDS study analyzed.
• This approach enhances statistical inference in fields affected by double truncation.