Large sample theory for merged data from multiple sources

DOI10.1214/18-AOS1727MaRDI QIDQ2414098zbMATH OpenOpenAlexWikidataFDO

Authors Takumi Saegusa

Publication date 10 May 2019

Published in The Annals of Statistics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1805.07800, https://projecteuclid.org/euclid.aos/1550026850

zbMATH Keywords

calibrationcentral limit theoremdata integrationempirical processsampling without replacementsemiparametric model

Mathematics Subject Classification ID

Central limit and other weak theorems (60F05) Asymptotic distribution theory in statistics (62E20) Asymptotic properties of nonparametric inference (62G20) Classification and discrimination; cluster analysis (statistical aspects) (62H30) Censored data models (62N01)

Abstract: We develop large sample theory for merged data from multiple sources. Main statistical issues treated in this paper are (1) the same unit potentially appears in multiple datasets from overlapping data sources, (2) duplicated items are not identified, and (3) a sample from the same data source is dependent due to sampling without replacement. We propose and study a new weighted empirical process and extend empirical process theory to a dependent and biased sample with duplication. Specifically, we establish the uniform law of large numbers and uniform central limit theorem over a class of functions along with several empirical process results under conditions identical to those in the i.i.d. setting. As applications, we study infinite-dimensional M-estimation and develop its consistency, rates of convergence, and asymptotic normality. Our theoretical results are illustrated with simulation studies and a real data example.

Review text

The author develops a framework to analyse the merging of data from multiple sources, allowing for dependence between data points, both from within the same source (due to sampling without replacement) and between different sources (allowing for the possibility of the same data point being sampled by two or more different sources). Within this framework, the main results of the present paper are a law of large numbers and central limit theorems for empirical processes built from Hartley's estimator. In the setting of the central limit theorem, the asymptotic variance is given a natural interpretation in terms of sources of variation in the model. Further work in the present paper includes results on consistency and asymptotic normality of estimators within this framework (together with rates of convergence), discussion of calibration, consideration of an optimal choice of weight function for the empirical processes, and consideration of examples including parametric models and Cox models. The paper concludes with two numerical studies; the first is a simulation study to investigate properties of the proposed estimators in a Cox model, and the second is an application to a real data set.

Recommendations

• Semiparametric inference for merged data from multiple data sources
• Large sample theory of empirical distributions in biased sampling models
• Analysis of integrated data
• Quadratic inference function approach to merging longitudinal studies: validation and joint estimation
• Asymptotic theory for clustered samples

Cited in

(8)

• Nonparametric inference for distribution functions with stratified samples
• Semiparametric inference for merged data from multiple data sources
• Synergetic methods of data merging in mathematical statistics
• Complex sampling designs: uniform limit theorems and applications
• Quadratic inference function approach to merging longitudinal studies: validation and joint estimation
• Analysis of integrated data
• Improving Trial Generalizability Using Observational Studies
• Extended Glivenko—Cantelli theorem for simple random sampling without replacement from a finite population

This page was built for publication: Large sample theory for merged data from multiple sources

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2414098)