Activity Number:
|
155
- Implementing Research-Based Recommendations in Ongoing Programs
|
Type:
|
Topic Contributed
|
Date/Time:
|
Monday, July 30, 2018 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Government Statistics Section
|
Abstract #329182
|
Presentation
|
Title:
|
Model-Assisted Regression Tree Estimator in the Occupational Employment Statistics Survey
|
Author(s):
|
Daniell Toth* and Kelly McConville
|
Companies:
|
Bureau of Labor Statistics and Swarthmore College
|
Keywords:
|
complex surveys;
recursive partitioning;
U.S. Bureau of Labor Statistics;
survey data;
post-stratification
|
Abstract:
|
Auxiliary information can increase the efficiency of survey estimators through an assisting model when the model captures some of the relationship between the auxiliary data and the study variables. Despite their superior properties, model-assisted estimators are rarely used in anything but their simplest form by statistical agencies to produce official statistics. This is in part due to the correlation of the available variables in the auxiliary data. It is quite common for variables in survey data to be co-linear, violating assumptions usually used in a model assisted approach. We propose using a regression tree estimator in a model-assisted estimate. Regression tree models are adept at handling data with collinear variables and interactions. The resulting estimator can be viewed as a post-stratification estimator where the post-strata are automatically selected by the recursive partitioning algorithm of the regression tree. We discuss the implementation of this estimator for use in the US Bureau of Labor Statistics Occupational Employment Statistics Survey and compare its performance to currently used estimator using a simulation.
|
Authors who are presenting talks have a * after their name.