Survey Methodology
Release date: December 15, 2020
The journal Survey Methodology Volume 46, Number 2 (December 2020) contains five papers.
Regular Papers
Survey Methodology
by Cristian Oliva-Aviles, Mary C. Meyer and Jean D. Opsomer
In many large-scale surveys, estimates are produced for numerous small domains defined by cross-classifications of demographic, geographic and other variables. Even though the overall sample size of such surveys might be very large, samples sizes for domains are sometimes too small for reliable estimation. We propose an improved estimation approach that is applicable when “natural” or qualitative relationships (such as orderings or other inequality constraints) can be formulated for the domain means at the population level. We stay within a design-based inferential framework but impose constraints representing these relationships on the sample-based estimates. The resulting constrained domain estimator is shown to be design consistent and asymptotically normally distributed as long as the constraints are asymptotically satisfied at the population level. The estimator and its associated variance estimator are readily implemented in practice. The applicability of the method is illustrated on data from the 2015 U.S. National Survey of College Graduates.
Bayesian hierarchical weighting adjustment and survey inference
by Yajuan Si, Rob Trangucci, Jonah Sol Gabry and Andrew Gelman
We combine weighting and Bayesian prediction in a unified approach to survey inference. The general principles of Bayesian analysis imply that models for survey outcomes should be conditional on all variables that affect the probability of inclusion. We incorporate all the variables that are used in the weighting adjustment under the framework of multilevel regression and poststratification, as a byproduct generating model-based weights after smoothing. We improve small area estimation by dealing with different complex issues caused by real-life applications to obtain robust inference at finer levels for subdomains of interest. We investigate deep interactions and introduce structured prior distributions for smoothing and stability of estimates. The computation is done via Stan and is implemented in the open-source R package rstanarm and available for public use. We evaluate the design-based properties of the Bayesian procedure. Simulation studies illustrate how the model-based prediction and weighting inference can outperform classical weighting. We apply the method to the New York Longitudinal Study of Wellbeing. The new approach generates smoothed weights and increases efficiency for robust finite population inference, especially for subsets of the population.
Firth’s penalized likelihood for proportional hazards regressions for complex surveys
This article proposes a weight scaling method for Firth’s penalized likelihood for proportional hazards regression models. The method derives a relationship between the penalized likelihood that uses scaled weights and the penalized likelihood that uses unscaled weights, and it shows that the penalized likelihood that uses scaled weights have some desirable properties. A simulation study indicates that the penalized likelihood using scaled weights produces smaller biases in point estimates and standard errors than the biases produced by the penalized likelihood using unscaled weights. The weighted penalized likelihood is applied to estimate hazard rates for heart attacks by using a public-use data set from the National Health and Epidemiology Followup Study (NHEFS). SAS® statements to estimate hazard rates using data from complex surveys are given in the appendix.
Probability-proportional-to-size ranked-set sampling from stratified populations
by Omer Ozturk
This paper constructs a probability-proportional-to-size (PPS) ranked-set sample from a stratified population. A PPS-ranked-set sample partitions the units in a PPS sample into groups of similar observations. The construction of similar groups relies on relative positions (ranks) of units in small comparison sets. Hence, the ranks induce more structure (stratification) in the sample in addition to the data structure created by unequal selection probabilities in a PPS sample. This added data structure makes the PPS-ranked-set sample more informative then a PPS-sample. The stratified PPS-ranked-set sample is constructed by selecting a PPS-ranked-set sample from each stratum population. The paper constructs unbiased estimators for the population mean, total and their variances. The new sampling design is applied to apple production data to estimate the total apple production in Turkey.
Semi-automated classification for multi-label open-ended questions
by Hyukjun Gweon, Matthias Schonlau and Marika Wenemark
In surveys, text answers from open-ended questions are important because they allow respondents to provide more information without constraints. When classifying open-ended questions automatically using supervised learning, often the accuracy is not high enough. Alternatively, a semi-automated classification strategy can be considered: answers in the easy-to-classify group are classified automatically, answers in the hard-to-classify group are classified manually. This paper presents a semi-automated classification method for multi-label open-ended questions where text answers may be associated with multiple classes simultaneously. The proposed method effectively combines multiple probabilistic classifier chains while avoiding prohibitive computational costs. The performance evaluation on three different data sets demonstrates the effectiveness of the proposed method.
- Date modified: