1994
DOI: 10.2307/2290910
|View full text |Cite
|
Sign up to set email alerts
|

Estimation of Regression Coefficients When Some Regressors Are Not Always Observed

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

13
717
0
11

Year Published

2000
2000
2016
2016

Publication Types

Select...
8
1

Relationship

0
9

Authors

Journals

citations
Cited by 591 publications
(741 citation statements)
references
References 0 publications
13
717
0
11
Order By: Relevance
“…The main idea of our method is inverse probability weighting, that is, we weight the completely observed cases inversely proportionally to the probability of being observed. The proposed methodology and theory extend those for the inverse probability weighted conditional mean regression models (e.g., [4]). We investigate the asymptotic normality of the weighted estimator and reveal an interesting phenomenon that using estimated weights often leads to asymptotically more efficient estimator than using the true weights, which echoes the same finding for weighted estimation in linear mean regression model [4].…”
Section: Introductionmentioning
confidence: 84%
“…The main idea of our method is inverse probability weighting, that is, we weight the completely observed cases inversely proportionally to the probability of being observed. The proposed methodology and theory extend those for the inverse probability weighted conditional mean regression models (e.g., [4]). We investigate the asymptotic normality of the weighted estimator and reveal an interesting phenomenon that using estimated weights often leads to asymptotically more efficient estimator than using the true weights, which echoes the same finding for weighted estimation in linear mean regression model [4].…”
Section: Introductionmentioning
confidence: 84%
“…To address potential bias resulting from loss to followup, we repeated the linear and logistic regressions using inverse probability weighting (IPW [12,21]), a technique that places greater weight on the age 7 observations of participants who were similar to those lost to follow-up. We first estimated the probability that a subject was observed at the age 7 visit, conditioned upon values of the covariates that were empirically associated with the probability of retention at age 7, case status, and the outcomes (continuous and dichotomous).…”
Section: Discussionmentioning
confidence: 99%
“…Robins et al [19] and Robins [18] established that in a model for data O D .Y; A; X/ in which k D P .A D kjX/ is known, any regular and asymptotically linear estimator forˇcan be found as the solution to P i i aug .Y; A; XI N K / D 0 for a specific choice of N K .X /. Zhang et al [17] demonstrated the use of this theory in RCTs with univariate outcomes.…”
Section: The Simple Augmented Generalized Estimating Equationmentioning
confidence: 99%
“…N K / T aug . N K / < 1, the optimal estimator within this class for a fixed .Y; AIˇ/ is obtained by setting k opt .X i / D Ef i .Y; AIˇ/jA i D k; X i g [17][18][19]. When only two treatment arms are considered, the augmentation term…”
Section: The Simple Augmented Generalized Estimating Equationmentioning
confidence: 99%