Premium
A two‐way classification of regression estimation strategies in probability sampling
Author(s) -
Särndal CarlErik
Publication year - 1980
Publication title -
canadian journal of statistics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.804
H-Index - 51
eISSN - 1708-945X
pISSN - 0319-5724
DOI - 10.2307/3315229
Subject(s) - estimator , mathematics , correctness , statistics , population , extremum estimator , sampling (signal processing) , regression analysis , regression , m estimator , linear regression , computer science , algorithm , demography , filter (signal processing) , sociology , computer vision
This paper examines strategies for estimating the mean of a finite population in the following situation: A linear regression model is assumed to describe the population scatter. Various estimators β for the vector of regression parameters β are considered. Several ways of transforming each estimator β into a model‐based estimator for the population mean are considered. Some estimators constructed in this way become sensitive to correctness of the assumed model. The estimators favoured in this paper are the ones in which the observations are weighted to reflect the sampling design, so that asymptotic design unbiasedness is achieved. For these estimators, the randomization distribution gives protection against model breakdown.